From maciej at opencsw.org Mon Aug 1 13:50:14 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 1 Aug 2011 12:50:14 +0100 Subject: [csw-maintainers] Technical Summercamp in Kiel 2011 In-Reply-To: <958B6EC9-5CE1-4B2E-BE39-2C85617D3EDD@opencsw.org> References: <958B6EC9-5CE1-4B2E-BE39-2C85617D3EDD@opencsw.org> Message-ID: 2011/7/6 Dagobert Michelsen : > Hi folks, > > time is passing and I would like to invite everybody to add their available time slots > to the doodle poll: > ?http://doodle.com/ewfzuze75afweueq > > I removed the early dates within holidays as there were anyway nobody interested and > added September. Please update soon as about 4 weeks in advance is needed for hotel > booking etc. >From the current state of the poll it looks like the optimal date is the weekend of 17-18 September 2011. So far, 7 people responded to the poll. I thought we could personally invite some maintainers to the summercamp. There might be people who just didn't think about coming. At the same time, our project is at relatively large changes (the automatic process), so the more people are there to discuss the future direction, the better. To extend the appeal to all maintainers - if you are an OpenCSW package maintainer, please consider coming over to the OpenCSW Summercamp in Kiel! Even if you maintain only a couple of packages, or you're a new maintainer, come and meet others! Knowing other maintainers will make your participation in the project easier and more fun! Just by osmosis, you will learn a lot about our build system, infrastructure, and things you didn't know existed. More importantly, you'll get to know personally other maintainers. While it might not seem that significant, it is! Email is inherently very low bandwidth; getting to know people lays the shared background necessary for effective day to day email communication. Lastly, other maintainers want to get to know you. I am coming to Kiel, and I want to meet you! I'm sure that other maintainers want the same. So, get to know others, come to Kiel! I'm looking forward to seeing you there! To get started, look at the doodle poll above and select the dates that fit you. Maciej From bwalton at opencsw.org Mon Aug 1 15:59:48 2011 From: bwalton at opencsw.org (Ben Walton) Date: Mon, 01 Aug 2011 09:59:48 -0400 Subject: [csw-maintainers] how the catalog signing works In-Reply-To: References: <1311439323-sup-2576@pinkfloyd.chass.utoronto.ca> Message-ID: <1312207128-sup-4900@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Sun Jul 24 17:14:24 -0400 2011: Hi Maciej, > If it could send email, where would you plan to send it to? Perhaps > a mail alias with a bunch of interested people? Something like a > mailing list, but with no archives (not necessary). Yes, a simple alias exploder should suffice for this. Something like gpgagent@ would work. The recipients should be those that can take action to reset the passphrase. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Tue Aug 2 01:09:20 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 2 Aug 2011 00:09:20 +0100 Subject: [csw-maintainers] gar and php question In-Reply-To: References: <4DEBB025.6010702@opencsw.org> <1307325421-sup-8953@pinkfloyd.chass.utoronto.ca> <1307365421-sup-6666@pinkfloyd.chass.utoronto.ca> <1307382737-sup-3595@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/6/6 Philip Brown : > So, please go ahead and announce , that we plan to remove it within > [insert timeframe here]. > (and to holler if this for some reason is a problem for someone, and why) > > And then after that time, we can remove it. > Since this is by definition extreme legacy software, the people may > not keep up to date with announcements, etc :-} so we may want to set > a somewhat longer timeframe than would otherwise be fine. > > July 1st? To continue the topic, the deprecation has been announced on the 8th of June: http://lists.opencsw.org/pipermail/users/2011-June/008917.html We're now at the beginning of August, so should we go ahead and drop php4 from unstable? Maciej From bwalton at opencsw.org Tue Aug 2 01:24:19 2011 From: bwalton at opencsw.org (Ben Walton) Date: Mon, 01 Aug 2011 19:24:19 -0400 Subject: [csw-maintainers] gar and php question In-Reply-To: References: <4DEBB025.6010702@opencsw.org> <1307325421-sup-8953@pinkfloyd.chass.utoronto.ca> <1307365421-sup-6666@pinkfloyd.chass.utoronto.ca> <1307382737-sup-3595@pinkfloyd.chass.utoronto.ca> Message-ID: No hesitations here. I'm ready to push the php5 update to unstable too...I just need to update the deps to fit mysql changes. Thanks -Ben "Maciej Blizi?ski" wrote: 2011/6/6 Philip Brown : > So, please go ahead and announce , that we plan to remove it within > [insert timeframe here]. > (and to holler if this for some reason is a problem for someone, and why) > > And then after that time, we can remove it. > Since this is by definition extreme legacy software, the people may > not keep up to date with announcements, etc :-} so we may want to set > a somewhat longer timeframe than would otherwise be fine. > > July 1st? To continue the topic, the deprecation has been announced on the 8th of June: http://lists.opencsw.org/pipermail/users/2011-June/008917.html We're now at the beginning of August, so should we go ahead and drop php4 from unstable? Maciej _____________________________________________ maintainers mailing list maintainers at lists.opencsw.org https://lists.opencsw.org/mailman/listinfo/maintainers .:: This mailing list's archive is public. ::. -------------- next part -------------- An HTML attachment was scrubbed... URL: From bwalton at opencsw.org Tue Aug 2 02:39:56 2011 From: bwalton at opencsw.org (Ben Walton) Date: Mon, 01 Aug 2011 20:39:56 -0400 Subject: [csw-maintainers] moving along... Message-ID: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Hi All, I'm back from a nice cottage vacation and ready to tackle another chunk of the automation process. We have two large (user visible) chunks remaining: 1. Mantis integration. 2. www/packages browsing interface. I think I'd like to work on 2 first. Can someone with access to www-mockup point me in the right direction for how we're doing our development there? I'll then proceed to check out the wordpress code and poke around our browser and the various databases to see how this should best be handled. Maciej, off the top of your head, how easy would it be to browse the same database that generates the catalogs? It's likely not carrying the svn url info presently, but would it be missing anything else? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Tue Aug 2 08:39:15 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 2 Aug 2011 07:39:15 +0100 Subject: [csw-maintainers] moving along... In-Reply-To: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/2 Ben Walton : > Maciej, off the top of your head, how easy would it be to browse the > same database that generates the catalogs? ?It's likely not carrying > the svn url info presently, but would it be missing anything else? >From the top of my head, it's also missing dependency information. A bit more information: Everything is in the pickled data structure, but not everything is also in database tables. Here's an example of the data structure: http://buildfarm.opencsw.org/pkgdb/srv4/071d260b87a06ff3339c1c2b89da1b02/ It has all the information about the package, but it's not as easily accessible. I had a git branch in the works with the data structure update. I haven't worked on it for a while now, but I can get back to it. Maciej From maciej at opencsw.org Tue Aug 2 15:57:20 2011 From: maciej at opencsw.org (Maciej (Matchek) Blizinski) Date: Tue, 2 Aug 2011 14:57:20 +0100 Subject: [csw-maintainers] Checking for ".so" in shared library file names Message-ID: Based on mantis bug #4811[1], here's an idea for a new check: all shared library must have the ".so" component in the file names. It can be in the middle (libfoo.so.1) or at the end (libfoo-1.so), but it must be present. The check would be limited to regular files, symlinks would not be checked - it's harder to implement reliably. Thoughts? Maciej [1] https://www.opencsw.org/mantis/view.php?id=4811 From maciej at opencsw.org Tue Aug 2 17:11:58 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 2 Aug 2011 16:11:58 +0100 Subject: [csw-maintainers] Buildfarm is inaccessible Message-ID: Our buildfarm seems inaccessible. Commands (e.g. uptime) started to hang on some hosts, and eventually ssh access to login.opencsw.org stopped working. The buildfarm database is inaccessible too. I've already notified buildfarm@, but got no answer so far, so I'm sending a message to a wider audience. Is anyone on the list able to investigate the issue? Maciej From bwalton at opencsw.org Tue Aug 2 17:24:09 2011 From: bwalton at opencsw.org (Ben Walton) Date: Tue, 02 Aug 2011 11:24:09 -0400 Subject: [csw-maintainers] [csw-buildfarm] Buildfarm is inaccessible In-Reply-To: References: Message-ID: <1312298236-sup-4347@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Tue Aug 02 11:11:58 -0400 2011: > Is anyone on the list able to investigate the issue? I can confirm your findings but can't do anything about it. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bwalton at opencsw.org Wed Aug 3 04:59:24 2011 From: bwalton at opencsw.org (Ben Walton) Date: Tue, 02 Aug 2011 22:59:24 -0400 Subject: [csw-maintainers] Checking for ".so" in shared library file names In-Reply-To: References: Message-ID: <1312340272-sup-1773@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej (Matchek) Blizinski's message of Tue Aug 02 09:57:20 -0400 2011: > Based on mantis bug #4811[1], here's an idea for a new check: all > shared library must have the ".so" component in the file names. It can > be in the middle (libfoo.so.1) or at the end (libfoo-1.so), but it > must be present. The check would be limited to regular files, > symlinks would not be checked - it's harder to implement reliably. It's a reasonable test I think. For symlinks, you'd need to do a full resolution and if the target is supposed to end in .so, so should the link. Likely a pile of (mostly unnecessary) work for that though. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bwalton at opencsw.org Wed Aug 3 05:51:42 2011 From: bwalton at opencsw.org (Ben Walton) Date: Tue, 02 Aug 2011 23:51:42 -0400 Subject: [csw-maintainers] libxml2 broken In-Reply-To: <1307407557-sup-3296@pinkfloyd.chass.utoronto.ca> References: <4DEBD2F3.8070808@opencsw.org> <4DEC251C.5060106@opencsw.org> <4DEC2D37.2010809@opencsw.org> <65D204EA-6681-4CAD-87C7-D383CD6E1875@opencsw.org> <1307407557-sup-3296@pinkfloyd.chass.utoronto.ca> Message-ID: <1312343279-sup-6076@pinkfloyd.chass.utoronto.ca> Excerpts from Ben Walton's message of Mon Jun 06 20:46:20 -0400 2011: > > Always good to keep the eyes open. I just looked at the libiconv recipe and > > "NOTES" from libiconv-1.13.1 says > > > > Q: Support EBCDIC ? > > A: No! > > > > However, the packaged and released libxml2 2.7.8 does also not pass this test. > > > > Ben: Do you want to report this upstream? > > Sure. Let me get a good set of info together to shoot up there and > I'll do so. Picking up an old thread that got lost in my mailbox... This is a known bug that also affects (at least) OSX[1]. I'm determining how much tampering is required to produce the requisite fix. The other option is to patch out the troublesome test file. Time availability may decide the outcome. I've got a test build running right now that will hopefully strip out the ebcdic handling support. If that works, I'll build the autoconf support next. Thanks -Ben [1] https://bugzilla.gnome.org/show_bug.cgi?id=603432 -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From jesse at opencsw.org Wed Aug 3 06:13:24 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 13:43:24 +0930 Subject: [csw-maintainers] [csw-buildfarm] Buildfarm is inaccessible In-Reply-To: <1312298236-sup-4347@pinkfloyd.chass.utoronto.ca> References: <1312298236-sup-4347@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/3 Ben Walton > Excerpts from Maciej Blizi?ski's message of Tue Aug 02 11:11:58 -0400 2011 > : > > > Is anyone on the list able to investigate the issue? > > I can confirm your findings but can't do anything about it. > > Seems OK now. -------------- next part -------------- An HTML attachment was scrubbed... URL: From jesse at opencsw.org Wed Aug 3 06:59:37 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 14:29:37 +0930 Subject: [csw-maintainers] Help for newbie creating a new package Message-ID: Hello Please forgive my dumbass newbie questions... I understand that I should use 'mgar newpkg' to create a new package definition, but it looks like I first need to have created a Makefile, or some other type of definitions file, for it says; unstable10x:~/opencsw/rbgems $ mgar newpkg mysql ERROR: No package build description found in the current directory (I am trying to create a package for the mysql ruby gem.) Where should I get a template 'package build description' from, and where should I place it, before running 'mgar newpkg mysql'? And then what happens? Note that so far all I've done is: - copied ben?s .garrc and modified paths to reference my home dir rather than his - run ?mgar init? - initialise the opencsw build dir (~/opencsw) - run ?mgar up --all? - fetch the package build descriptions - takes a good while (many hours!) - run ?mgar index? to create (update) the index of the package build tree Thank you Jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From jesse at opencsw.org Wed Aug 3 07:43:02 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 15:13:02 +0930 Subject: [csw-maintainers] Where to report bugs for mgar ? Message-ID: Hello Do bugs in the mgar command get reported in the gar_devel project in Mantis? (There's also a gar_dev). In this case it's a minor thing - the usage output doesn't include the 'newpkg' command. The man page says: REPORTING BUGS Please report bugs at http://www.opencsw.org/mantis/. But not which project. ... Thank you Jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From jesse at opencsw.org Wed Aug 3 09:06:45 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 16:36:45 +0930 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: On 3 August 2011 14:29, Jesse Reynolds wrote: > Hello > > Please forgive my dumbass newbie questions... > > I understand that I should use 'mgar newpkg' to create a new package > definition, but it looks like I first need to have created a Makefile, or > some other type of definitions file, for it says; > > unstable10x:~/opencsw/rbgems $ mgar newpkg mysql > ERROR: No package build description found in the current directory > > (I am trying to create a package for the mysql ruby gem.) > > Where should I get a template 'package build description' from, and where > should I place it, before running 'mgar newpkg mysql'? > > And then what happens? > In my wanderings I've found the following wiki page: http://wiki.opencsw.org/project-rubygems Which says the sqlite gem Makefile is a good staring point for other gem package Makefiles: http://sourceforge.net/apps/trac/gar/browser/csw/mgar/pkg/rbgems/sqlite3-ruby/trunk/Makefile And these are part of the rbgems category which has it's Makefile here: http://sourceforge.net/apps/trac/gar/browser/csw/mgar/gar/v2/categories/rbgems/category.mk But yes, I don't quite get what to do. I tried creating a Makefile and running 'mgar newpkg gperf' as per one of the examples in the wiki but it complained ... unstable9s:~/test/gperf $ mgar fetch svn: '.' is not a working copy ERROR: No build system found at /home/jesse/opencsw/.buildsys/ Clearly I'm missing something basic about how, or where, to run mgar. tia. > > Note that so far all I've done is: > > - copied ben?s .garrc and modified paths to reference my home dir rather > than his > - run ?mgar init? - initialise the opencsw build dir (~/opencsw) > - run ?mgar up --all? - fetch the package build descriptions - takes a good > while (many hours!) > - run ?mgar index? to create (update) the index of the package build tree > > Thank you > Jesse > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From jesse at opencsw.org Wed Aug 3 09:10:36 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 16:40:36 +0930 Subject: [csw-maintainers] [csw-buildfarm] Buildfarm is inaccessible In-Reply-To: References: <1312298236-sup-4347@pinkfloyd.chass.utoronto.ca> Message-ID: On 3 August 2011 13:43, Jesse Reynolds wrote: > > > 2011/8/3 Ben Walton > >> Excerpts from Maciej Blizi?ski's message of Tue Aug 02 11:11:58 -0400 >> 2011: >> >> > Is anyone on the list able to investigate the issue? >> >> I can confirm your findings but can't do anything about it. >> >> > Seems OK now. > Er, the VSphere machines seem to have gone away again. I was logged into unstable9x and unstable10x earlier but now they've frozen up. I can still log in to unstable9s and the other sparc build machines. But I can't ping or ssh to any of the x86 build machines that are VMs on vsphere right now. Jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciej at opencsw.org Wed Aug 3 09:24:59 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 08:24:59 +0100 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: References: Message-ID: 2011/8/3 Jesse Reynolds : > Do bugs in the mgar command get reported in the gar_devel project in Mantis? > (There's also a gar_dev). > In this case it's a minor thing - the usage output doesn't include the > 'newpkg' command. > The man page says: > REPORTING BUGS > ? ? ?Please report bugs at http://www.opencsw.org/mantis/. > But not which project. ... It looks like mgar is not registered in mantis. Mantis integration is one of the missing bits which we haven't solved since the former release manager's departure. We're working on it. [1] For now, you can file a bug against gar_dev, and Ben will route the bug to Sebastian, or you can send Sebastian (skayser@) an email. Another thing you can do is checking out mgar sources[2], editing them and sending a patch to devel at . Maciej [1] http://lists.opencsw.org/pipermail/maintainers/2011-August/015087.html [2] https://opencsw.svn.sourceforge.net/svnroot/opencsw/gar-wrapper/ From maciej at opencsw.org Wed Aug 3 09:38:31 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 08:38:31 +0100 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: 2011/8/3 Jesse Reynolds : > I tried creating a Makefile and running 'mgar newpkg gperf' as per one of > the examples in the wiki but it complained ... > unstable9s:~/test/gperf $ mgar fetch > svn: '.' is not a working copy > ERROR: No build system found at /home/jesse/opencsw/.buildsys/ > Clearly I'm missing something basic about how, or where, to run mgar. The error you're seeing is the one that shows when the GARTYPE setting is missing from the Makefile you're attempting to build. This setting tells mgar which branch of mgar (v1, v2, perhaps another one) to use. You can also debug mgar. It is written in shell, so you can run the following command: /opt/csw/bin/bash -x /opt/csw/bin/mgar (...) If the debug output doesn't help you, please post it here (or on IRC via a paste service), it should shed more light on what's going on. Maciej From maciej at opencsw.org Wed Aug 3 09:44:35 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 08:44:35 +0100 Subject: [csw-maintainers] Checking for ".so" in shared library file names In-Reply-To: <1312340272-sup-1773@pinkfloyd.chass.utoronto.ca> References: <1312340272-sup-1773@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/3 Ben Walton : > It's a reasonable test I think. ?For symlinks, you'd need to do a full > resolution and if the target is supposed to end in .so, so should the > link. To be more precise, if the target is a shared library as indicated by libmagic, there needs to be an ".so" somewhere in the file name. At times, you have 2 or 3 levels of symlinks before you reach the regular file. >?Likely a pile of (mostly unnecessary) work for that though. One of the problems might be that the symlink target is not available, e.g. it's not in the package set that checkpkg is examining. In such case, checkpkg could reach and fetch that information from the database, but mime metadata are currently not cached in MySQL tables. It's something I intend to change, but it'll require time, and we have more urgent problems to solve. The regular file / shared library check, is easy to implement. Maciej From jesse at opencsw.org Wed Aug 3 10:02:31 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 17:32:31 +0930 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: 2011/8/3 Maciej Blizi?ski > 2011/8/3 Jesse Reynolds : > > I tried creating a Makefile and running 'mgar newpkg gperf' as per one of > > the examples in the wiki but it complained ... > > unstable9s:~/test/gperf $ mgar fetch > > svn: '.' is not a working copy > > ERROR: No build system found at /home/jesse/opencsw/.buildsys/ > > Clearly I'm missing something basic about how, or where, to run mgar. > > The error you're seeing is the one that shows when the GARTYPE setting > is missing from the Makefile you're attempting to build. This setting > tells mgar which branch of mgar (v1, v2, perhaps another one) to use. > You can also debug mgar. It is written in shell, so you can run the > following command: > > /opt/csw/bin/bash -x /opt/csw/bin/mgar (...) > > If the debug output doesn't help you, please post it here (or on IRC > via a paste service), it should shed more light on what's going on. > > Thanks Maciej. I've added the following to the Makefile: GARTYPE = v2 And how it seems happy. I've now done the following for the gperf example: mgar fetch mgar makesum mgar checksum mgar extract mgar configure mgar build mgar test Yay! However: mgar package runs into some errors. Here's the tail end of the output: CSWgperf: If any of the reported errors were false positives, you can override them pasting the lines below to the GAR recipe. CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/opt/SUNWspro/lib/v8|opt/csw/bin/gperf CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/lib|opt/csw/bin/gperf CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/opt/SUNWspro/lib|opt/csw/bin/gperf CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/opt/SUNWspro/lib/rw7|opt/csw/bin/gperf Please note that checkpkg isn't suggesting you should simply add these overrides do the Makefile. It only informs what the overrides could look like. You need to understand what are the reported issues about and use your best judgement to decide whether to fix the underlying problems or override them. For more information, scroll up and read the detailed messages. gmake: *** [pkgcheck] Error 2 Getting much closer though, thank you. Jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciej at opencsw.org Wed Aug 3 10:10:25 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 09:10:25 +0100 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: 2011/8/3 Jesse Reynolds : > CHECKPKG_OVERRIDES_CSWgperf += > bad-rpath-entry|/opt/SUNWspro/lib/v8|opt/csw/bin/gperf > CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/lib|opt/csw/bin/gperf > CHECKPKG_OVERRIDES_CSWgperf += > bad-rpath-entry|/opt/SUNWspro/lib|opt/csw/bin/gperf > CHECKPKG_OVERRIDES_CSWgperf += > bad-rpath-entry|/opt/SUNWspro/lib/rw7|opt/csw/bin/gperf This check deals with the RPATH setting in the resulting binaries. Sun Studio by default adds extra RPATH entries, e.g. "/opt/SUNWspro/lib/v8". To fix this, you need to tell Sun Studio to stop doing that: EXTRA_CFLAGS = -xnorunpath The EXTRA_CFLAGS variable value will be put by GAR into the right places for autotools to pick up. Maciej From jesse at opencsw.org Wed Aug 3 10:33:04 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 18:03:04 +0930 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: 2011/8/3 Maciej Blizi?ski > 2011/8/3 Jesse Reynolds : > > CHECKPKG_OVERRIDES_CSWgperf += > > bad-rpath-entry|/opt/SUNWspro/lib/v8|opt/csw/bin/gperf > > CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/lib|opt/csw/bin/gperf > > CHECKPKG_OVERRIDES_CSWgperf += > > bad-rpath-entry|/opt/SUNWspro/lib|opt/csw/bin/gperf > > CHECKPKG_OVERRIDES_CSWgperf += > > bad-rpath-entry|/opt/SUNWspro/lib/rw7|opt/csw/bin/gperf > > This check deals with the RPATH setting in the resulting binaries. > Sun Studio by default adds extra RPATH entries, e.g. > "/opt/SUNWspro/lib/v8". To fix this, you need to tell Sun Studio to > stop doing that: > > EXTRA_CFLAGS = -xnorunpath > > The EXTRA_CFLAGS variable value will be put by GAR into the right > places for autotools to pick up. > Thanks. So I put that line into gperf/Makefile and ran 'mgar clean', 'mgar package' and it ends up with the same errors again (though in a different order): CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/opt/SUNWspro/lib|opt/csw/bin/gperf CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/opt/SUNWspro/lib/rw7|opt/csw/bin/gperf CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/opt/SUNWspro/lib/v8|opt/csw/bin/gperf CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/lib|opt/csw/bin/gperf This is on unstable9s, I think I ran it last time on one of the other sparc machines in the build farm. -------------- next part -------------- An HTML attachment was scrubbed... URL: From jesse at opencsw.org Wed Aug 3 10:41:16 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 18:11:16 +0930 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: On 3 August 2011 18:03, Jesse Reynolds wrote: > > > 2011/8/3 Maciej Blizi?ski > >> 2011/8/3 Jesse Reynolds : >> > CHECKPKG_OVERRIDES_CSWgperf += >> > bad-rpath-entry|/opt/SUNWspro/lib/v8|opt/csw/bin/gperf >> > CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/lib|opt/csw/bin/gperf >> > CHECKPKG_OVERRIDES_CSWgperf += >> > bad-rpath-entry|/opt/SUNWspro/lib|opt/csw/bin/gperf >> > CHECKPKG_OVERRIDES_CSWgperf += >> > bad-rpath-entry|/opt/SUNWspro/lib/rw7|opt/csw/bin/gperf >> >> This check deals with the RPATH setting in the resulting binaries. >> Sun Studio by default adds extra RPATH entries, e.g. >> "/opt/SUNWspro/lib/v8". To fix this, you need to tell Sun Studio to >> stop doing that: >> >> EXTRA_CFLAGS = -xnorunpath >> >> The EXTRA_CFLAGS variable value will be put by GAR into the right >> places for autotools to pick up. >> > > Thanks. So I put that line into gperf/Makefile and ran 'mgar clean', 'mgar > package' and it ends up with the same errors again (though in a different > order): > > CHECKPKG_OVERRIDES_CSWgperf += > bad-rpath-entry|/opt/SUNWspro/lib|opt/csw/bin/gperf > CHECKPKG_OVERRIDES_CSWgperf += > bad-rpath-entry|/opt/SUNWspro/lib/rw7|opt/csw/bin/gperf > CHECKPKG_OVERRIDES_CSWgperf += > bad-rpath-entry|/opt/SUNWspro/lib/v8|opt/csw/bin/gperf > CHECKPKG_OVERRIDES_CSWgperf += bad-rpath-entry|/lib|opt/csw/bin/gperf > > This is on unstable9s, I think I ran it last time on one of the other sparc > machines in the build farm. > My Makefile is currently looking like this: unstable9s:~/test/gperf $ cat Makefile NAME = gperf VERSION = 3.0.3 CATEGORIES = devel DESCRIPTION = A perfect hash function generator define BLURB GNU gperf is a perfect hash function generator. For a given list of strings, it produces a hash function and hash table, in form of C or C++ code, for looking up a value depending on the input string. The hash function is perfect, which means that the hash table has no collisions, and the hash table lookup needs a single string comparison only. endef MASTER_SITES = $(GNU_MIRROR) DISTFILES = $(NAME)-$(VERSION).tar.gz GARTYPE = v2 TEST_TARGET = check EXTRA_CFLAGS = -xnorunpath include gar/category.mk -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciej at opencsw.org Wed Aug 3 10:53:43 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 09:53:43 +0100 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: 2011/8/3 Jesse Reynolds : > EXTRA_CFLAGS = -xnorunpath This was under the assumption that it's a C (not C++) code. To enable the same thing for C++ binaries, set: EXTRA_CXXFLAGS = -norunpath (I'm not sure if this is what the problem is, it's a guess that it's a C++ binary.) Maciej From maciej at opencsw.org Wed Aug 3 11:15:44 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 10:15:44 +0100 Subject: [csw-maintainers] Checking for ".so" in shared library file names In-Reply-To: References: <1312340272-sup-1773@pinkfloyd.chass.utoronto.ca> Message-ID: Committed in r15249. Also, run against libdnet which now shows the error tags. http://sourceforge.net/apps/trac/gar/changeset/15249 http://buildfarm.opencsw.org/pkgdb/srv4/c3becaf3633bfdbc0f315b73bc1d798c/ http://buildfarm.opencsw.org/pkgdb/srv4/cbf909119bdfed1a7c07a147741ff511/ Maciej From jesse at opencsw.org Wed Aug 3 11:17:07 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 18:47:07 +0930 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: 2011/8/3 Maciej Blizi?ski > 2011/8/3 Jesse Reynolds : > > EXTRA_CFLAGS = -xnorunpath > > This was under the assumption that it's a C (not C++) code. To enable > the same thing for C++ binaries, set: > > EXTRA_CXXFLAGS = -norunpath > > (I'm not sure if this is what the problem is, it's a guess that it's a > C++ binary.) > Thanks. Alas now I get a syntax error on ld - doesn't like -xn... /opt/SUNWspro/bin/CC -xO3 -m32 -xarch=v8 -xnorunpath -m32 -xarch=v8 -L/opt/csw/lib -o gperf version.o positions.o options.o keyword.o keyword-list.o input.o bool-array.o hash-table.o search.o output.o main.o ../lib/libgp.a -lm CC: Warning: Option -xnorunpath passed to ld, if ld is invoked, ignored otherwise /usr/ccs/bin/ld: illegal option -- x /usr/ccs/bin/ld: illegal option -- n ld: warning: option -o appears more than once, first setting taken usage: ld [-6:abc:d:e:f:h:il:mo:p:rstu:z:B:CD:F:GI:L:M:N:P:Q:R:S:VY:?] file(s) -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciej at opencsw.org Wed Aug 3 11:34:53 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 10:34:53 +0100 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: 2011/8/3 Jesse Reynolds : > Thanks. Alas now I get a syntax error on ld - doesn't like -xn... This happens when CFLAGS get passed to the C++ compiler. If you set the flags correctly and it happens, then might be a bug in gperf. Maciej From jesse at opencsw.org Wed Aug 3 15:10:40 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 22:40:40 +0930 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: 2011/8/3 Maciej Blizi?ski > 2011/8/3 Jesse Reynolds : > > Thanks. Alas now I get a syntax error on ld - doesn't like -xn... > > This happens when CFLAGS get passed to the C++ compiler. If you set > the flags correctly and it happens, then might be a bug in gperf. > Right. But this is from the GAR example on the wiki, version 3.0.3 of gperf which presumably used to work at some point. http://sourceforge.net/apps/trac/gar/wiki/ExplainedGperf -------------- next part -------------- An HTML attachment was scrubbed... URL: From bwalton at opencsw.org Wed Aug 3 15:22:11 2011 From: bwalton at opencsw.org (Ben Walton) Date: Wed, 03 Aug 2011 09:22:11 -0400 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: References: Message-ID: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Wed Aug 03 03:24:59 -0400 2011: > For now, you can file a bug against gar_dev, and Ben will route the > bug to Sebastian, or you can send Sebastian (skayser@) an email. > Another thing you can do is checking out mgar sources[2], editing them > and sending a patch to devel at . Yes, I'll handle routing that to Sebastian, but sending a patch to devel@ is definitely the best option as Sebastian can simply integrate and release an update. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Wed Aug 3 15:27:13 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 14:27:13 +0100 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: 2011/8/3 Jesse Reynolds : > > > 2011/8/3 Maciej Blizi?ski >> >> 2011/8/3 Jesse Reynolds : >> > Thanks. Alas now I get a syntax error on ld - doesn't like -xn... >> >> This happens when CFLAGS get passed to the C++ compiler. ?If you set >> the flags correctly and it happens, then might be a bug in gperf. > > Right. But this is from the GAR example on the wiki, version 3.0.3 of gperf > which presumably used to work at some > point.?http://sourceforge.net/apps/trac/gar/wiki/ExplainedGperf The check set in checkpkg is continuously updated, and build recipes that looked good so far, might start triggering errors. This particular error is not serious, you could override it. However, if you set a variable and it doesn't propagate well, it means that either there's something wrong with the gperf build system and it doesn't process the environment properly, or you don't have the full understanding of how the build works. Just to double-check: did you notice the difference between -norunpath and -xnorunpath? Maciej Maciej From jesse at opencsw.org Wed Aug 3 15:51:38 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 23:21:38 +0930 Subject: [csw-maintainers] Help for newbie creating a new package In-Reply-To: References: Message-ID: 2011/8/3 Maciej Blizi?ski > 2011/8/3 Jesse Reynolds : > > > > > > 2011/8/3 Maciej Blizi?ski > >> > >> 2011/8/3 Jesse Reynolds : > >> > Thanks. Alas now I get a syntax error on ld - doesn't like -xn... > >> > >> This happens when CFLAGS get passed to the C++ compiler. If you set > >> the flags correctly and it happens, then might be a bug in gperf. > > > > Right. But this is from the GAR example on the wiki, version 3.0.3 of > gperf > > which presumably used to work at some > > point. http://sourceforge.net/apps/trac/gar/wiki/ExplainedGperf > > The check set in checkpkg is continuously updated, and build recipes > that looked good so far, might start triggering errors. > Right, OK. > > This particular error is not serious, you could override it. However, > if you set a variable and it doesn't propagate well, it means that > either there's something wrong with the gperf build system and it > doesn't process the environment properly, or you don't have the full > understanding of how the build works. > OK right. I'm just trying to get a feel for the system by following an example. > > Just to double-check: did you notice the difference between -norunpath > and -xnorunpath? > Ack! No! Apologies. Trying that now ... Perfect. The following packages have been built: CSWgperf /home/jesse/staging/gperf-3.0.3,REV=2011.08.03-SunOS5.9-sparc-NOTVERSIONED.pkg.gz [package] complete for gperf. -------------- next part -------------- An HTML attachment was scrubbed... URL: From jesse at opencsw.org Wed Aug 3 15:58:16 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Wed, 3 Aug 2011 23:28:16 +0930 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> References: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> Message-ID: On 3 August 2011 22:52, Ben Walton wrote: > Excerpts from Maciej Blizi?ski's message of Wed Aug 03 03:24:59 -0400 2011: > > > For now, you can file a bug against gar_dev, and Ben will route the > > bug to Sebastian, or you can send Sebastian (skayser@) an email. > > Another thing you can do is checking out mgar sources[2], editing them > > and sending a patch to devel at . > > Yes, I'll handle routing that to Sebastian, but sending a patch to > devel@ is definitely the best option as Sebastian can simply integrate > and release an update. > > OK, so what's the best one line description of what 'mgar newpkg' does? Perhaps: newpkg - initialises a package build directory, requires a Makefile with package build description As I'm still getting my head around this stuff .. -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciej at opencsw.org Wed Aug 3 16:19:37 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 15:19:37 +0100 Subject: [csw-maintainers] Issues with libdnet Message-ID: [cc:spikey] Hello maintainers, There is a problem with our libdnet package: shared libraries are missing the .so suffix. I've already implemented a check to prevent this from going unnoticed in the future, but we need to fix the existing libdnet package. It is necessary for Spikey (cc'd) to port ArpON to Solaris. I have replicated the issue on our buildfarm, and it looks like it's a bug in the libdnet build scripts. I haven't got the bandwidth to further investigate and fix this. Could any of you jump in and figure out what's wrong? Maciej [1] https://www.opencsw.org/mantis/view.php?id=4811 From bonivart at opencsw.org Wed Aug 3 16:46:49 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Wed, 3 Aug 2011 16:46:49 +0200 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: References: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> Message-ID: On Wed, Aug 3, 2011 at 3:58 PM, Jesse Reynolds wrote: > OK, so what's the best one line description of what 'mgar newpkg' does? Are you sure you're running the latest version? bonivart at unstable9s[~]$ which mgar /home/bonivart/bin/mgar bonivart at unstable9s[~]$ mgar version 380 bonivart at unstable9s[~]$ mgar Wrapper around the GAR build system. Usage: mgar [options] Main actions: init [dir] Initialize package build tree, defaults to ~/opencsw/ index Build/update the package build tree index locate Search a package within the package build tree index up [--all] Update current (or all) package build descriptions newpkg [ver] Create new package build directory Package build actions (to be called from a package build directory): fetch Download the package source makesum Generate a checksum for the package source extract Extract the packages source configure Configure the packages source build Build the package source package Assemble the package platforms Assemble packages for all platforms clean Reset current directory working files (current platform) spotless Reset the package build directory (all platforms) commit -m Commit build recipe (prefixes msg with package path) show-srcdir Show the location of the extracted package source show-stagedir Show the location of the package staging directory find-file Find a file within the extracted package source edit-file Edit a file within the extracted package source Other actions: modenv Display employed compiler options scm Pass a command to the underlying SCM (currently: svn) show-buildsys Display the location and version of the build system show-pkgtree Show the location of the package build tree version Display mgar version information Note that it does indeed mention newpkg. /peter From bwalton at opencsw.org Wed Aug 3 19:53:55 2011 From: bwalton at opencsw.org (Ben Walton) Date: Wed, 03 Aug 2011 13:53:55 -0400 Subject: [csw-maintainers] [csw-buildfarm] vsphere vms for opencsw build farm inaccessable Message-ID: <1312394016-sup-2972@pinkfloyd.chass.utoronto.ca> Excerpts from Jan Holzhueter's message of Wed Aug 03 13:44:17 -0400 2011: [+maintainers] > The x86 build servers should be up again I hope. Thanks for the update Jan. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bonivart at opencsw.org Wed Aug 3 19:56:22 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Wed, 3 Aug 2011 19:56:22 +0200 Subject: [csw-maintainers] [csw-buildfarm] vsphere vms for opencsw build farm inaccessable In-Reply-To: <4E3988F1.3040408@baltic-online.de> References: <4E394195.3050001@baltic-online.de> <4E3988F1.3040408@baltic-online.de> Message-ID: On Wed, Aug 3, 2011 at 7:44 PM, Jan Holzhueter wrote: > Am 03.08.11 14:39, schrieb Jan Holzhueter: > >> Hi, >> we did have some network Problems. >> Should be fixed now though. >> But it looks like it killed the X build Servers. >> I can't restart them atm. Will try to reach someone who can. > > The x86 build servers should be up again I hope. Yes, they are. Thanks for helping out with this on your vacation! :) /peter From jesse at opencsw.org Thu Aug 4 00:22:22 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Thu, 4 Aug 2011 07:52:22 +0930 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: References: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> Message-ID: On 4 August 2011 00:16, Peter Bonivart wrote: > On Wed, Aug 3, 2011 at 3:58 PM, Jesse Reynolds wrote: > > OK, so what's the best one line description of what 'mgar newpkg' does? > > Are you sure you're running the latest version? > Ah, No, I'm running the version installed on unstable9s: unstable9s:~ $ which mgar /opt/csw/bin/mgar unstable9s:~ $ mgar version 360 So it's best practice to checkout mgar into ~/bin/ ? What other utilities / libraries fall into this category? > bonivart at unstable9s[~]$ which mgar > /home/bonivart/bin/mgar > bonivart at unstable9s[~]$ mgar version > 380 > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciej at opencsw.org Thu Aug 4 00:29:47 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 23:29:47 +0100 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: References: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/3 Jesse Reynolds : > > > On 4 August 2011 00:16, Peter Bonivart wrote: >> >> On Wed, Aug 3, 2011 at 3:58 PM, Jesse Reynolds wrote: >> > OK, so what's the best one line description of what 'mgar newpkg' does? >> >> Are you sure you're running the latest version? > > Ah, No, I'm running the version installed on unstable9s: > unstable9s:~ $ which mgar > /opt/csw/bin/mgar > unstable9s:~ $ mgar version > > 360 > > So it's best practice to checkout mgar into ~/bin/ ? What other utilities / > libraries fall into this category? We don't have good procedures to deploy these tools on the buildfarm. For now, we need to email buildfarm@ with the request to upgrade mgar. You can do that if you want to. Maciej From bwalton at opencsw.org Thu Aug 4 00:41:49 2011 From: bwalton at opencsw.org (Ben Walton) Date: Wed, 03 Aug 2011 18:41:49 -0400 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: References: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> Message-ID: <1312411253-sup-6572@pinkfloyd.chass.utoronto.ca> Excerpts from Jesse Reynolds's message of Wed Aug 03 18:22:22 -0400 2011: > > Are you sure you're running the latest version? > > > > Ah, No, I'm running the version installed on unstable9s: > > unstable9s:~ $ which mgar > /opt/csw/bin/mgar > unstable9s:~ $ mgar version > > 360 I'll just updated the boxes with the latest release... :) > So it's best practice to checkout mgar into ~/bin/ ? What other > utilities / libraries fall into this category? Outside of GAR itself (which includes checkpkg), not much. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Thu Aug 4 00:44:19 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 3 Aug 2011 23:44:19 +0100 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: <1312411253-sup-6572@pinkfloyd.chass.utoronto.ca> References: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> <1312411253-sup-6572@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/3 Ben Walton : >> So it's best practice to checkout mgar into ~/bin/ ? What other >> utilities / libraries fall into this category? > > Outside of GAR itself (which includes checkpkg), not much. In GAR sources, there are also pkgdb and csw-upload-pkg. The latter is also packaged and installed on login, but the packaged version might at times lag behind the one in the source tree. Maciej From jesse at opencsw.org Thu Aug 4 01:26:34 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Thu, 4 Aug 2011 08:56:34 +0930 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: <1312411253-sup-6572@pinkfloyd.chass.utoronto.ca> References: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> <1312411253-sup-6572@pinkfloyd.chass.utoronto.ca> Message-ID: On 4 August 2011 08:11, Ben Walton wrote: > Excerpts from Jesse Reynolds's message of Wed Aug 03 18:22:22 -0400 2011: > > > > Are you sure you're running the latest version? > > > > > > > Ah, No, I'm running the version installed on unstable9s: > > > > unstable9s:~ $ which mgar > > /opt/csw/bin/mgar > > unstable9s:~ $ mgar version > > > > 360 > > I'll just updated the boxes with the latest release... :) > Thanks Ben, I can see it's updated yep. > > > So it's best practice to checkout mgar into ~/bin/ ? What other > > utilities / libraries fall into this category? > > Outside of GAR itself (which includes checkpkg), not much. > OK thanks, I'll get GAR as well, to be sure, to be sure. Now, mgar 380 is quite different to mgar 360 ... with 360 I worked out I could: mkdir -p ~/test/gperf cd ~/test/gperf cat > Makefile ... paste in the build description ... mgar newpkg mgar package And have it do everything, now it seems the newpkg command does quite different stuff and so I'm back to not being able to create a package again. newpkg now wants to create the directory for the package, and seemingly doesn't care about a build description - presumably now you have to run 'mgar newpkg gperf' under some place in the actual checked out build descriptions heirarchy? And then add your custom Makefile after it's created the directory structure? Or ...? Thank you Jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciej at opencsw.org Thu Aug 4 01:51:04 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Thu, 4 Aug 2011 00:51:04 +0100 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: References: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> <1312411253-sup-6572@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/4 Jesse Reynolds : > presumably now you have to run 'mgar newpkg gperf' under some place in the > actual checked out build descriptions heirarchy? Yes, I think that the idea is that you run this from the pkg/ directory. From bonivart at opencsw.org Thu Aug 4 10:43:30 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Thu, 4 Aug 2011 10:43:30 +0200 Subject: [csw-maintainers] Where to report bugs for mgar ? In-Reply-To: References: <1312377656-sup-2607@pinkfloyd.chass.utoronto.ca> Message-ID: On Thu, Aug 4, 2011 at 12:22 AM, Jesse Reynolds wrote: > So it's best practice to checkout mgar into ~/bin/ ? I would say no. I did this before mgar was packaged at all. Even though the latest version wasn't packaged until recently and then not installed I think that mgar will probably stabilize soon and a gap like this will not matter much. It's much easier for everyone if it's in /opt/csw/bin on the build hosts. /peter From maciej at opencsw.org Thu Aug 4 16:09:31 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Thu, 4 Aug 2011 15:09:31 +0100 Subject: [csw-maintainers] Issues with libdnet In-Reply-To: References: Message-ID: 2011/8/3 Maciej Blizi?ski : > I have replicated the issue on our buildfarm, and it looks like it's a > bug in the libdnet build scripts. ?I haven't got the bandwidth to > further investigate and fix this. ?Could any of you jump in and figure > out what's wrong? I found a bit of debian documentation[1] indicating that it could be a problem with old libtool. I tried the following: http://sourceforge.net/apps/trac/gar/changeset/15258 Unfortunately, it did not solve the issue. (By the way, I assume that autoreconf calls libtoolize.) Does anyone have an idea where to look? Maciej [1] http://www.v7w.com/debian/libtool-missing_so.html From maciej at opencsw.org Thu Aug 4 17:59:25 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Thu, 4 Aug 2011 16:59:25 +0100 Subject: [csw-maintainers] moving along... In-Reply-To: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/2 Ben Walton : > 1. Mantis integration. I will take that part. The rough plan is to write an integration job which will pull data from the buildfarm db via a RESTful interface (which we control) and stuff data into the bugtracker using XML RPC... ...but there is a problem with XML RPC: we're running an old version of Mantis with broken XML RPC interface. We need to upgrade... ...but there is a problem with the upgrade: Mantis is not written for the number of projects that we have. It is meant to handle a two digit number of projects, and we have a four digit one. If we just upgrade to the newest Mantis, our bugtracker will stop working, there will be page timeouts. Our currently deployed version of mantis has been patched to optimize and make it work with the number of projects that we have. Unfortunately, these patches have not been pushed upstream, and they don't apply any more to the newer Mantis, so we need to start over again with profiling and writing patches. So this is where I'll start. Here are the notes about the mantis speedup project. Sebastian has already done a fair bit of work there. http://wiki.opencsw.org/project-mantis-speedup Maciej From bonivart at opencsw.org Thu Aug 4 18:06:09 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Thu, 4 Aug 2011 18:06:09 +0200 Subject: [csw-maintainers] moving along... In-Reply-To: References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/4 Maciej Blizi?ski : > So this is where I'll start. Here are the notes about the mantis > speedup project. Sebastian has already done a fair bit of work there. > > http://wiki.opencsw.org/project-mantis-speedup Isn't this a lot of effort if we're anyway switch to Jira or whatever? Couldn't we instead: 1. Register new projects in current Mantis manually when a new package is released (not _that_ frequent) 2. Put the same effort into getting Jira/? working on the side. Installing, testing, migrating. Thoughts? /peter From maciej at opencsw.org Thu Aug 4 19:56:12 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Thu, 4 Aug 2011 18:56:12 +0100 Subject: [csw-maintainers] moving along... In-Reply-To: References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/4 Peter Bonivart : > 2011/8/4 Maciej Blizi?ski : >> So this is where I'll start. Here are the notes about the mantis >> speedup project. Sebastian has already done a fair bit of work there. >> >> http://wiki.opencsw.org/project-mantis-speedup > > Isn't this a lot of effort if we're anyway switch to Jira or whatever? > > Couldn't we instead: > > 1. Register new projects in current Mantis manually when a new package > is released (not _that_ frequent) > 2. Put the same effort into getting Jira/? working on the side. > Installing, testing, migrating. > > Thoughts? It's hard to tell what will require more work, switching to jira or taming mantis. I'm personally not inclined to spearhead the jira migration right now (although I would not mind another person doing that). In my experience it's a good idea to make small, effective steps. For example, we could start with a script that doesn't write to mantis, but queries both systems and tells you the difference; what to do in mantis to bring it to the state that matches the catalog. This shouldn't be that hard, and is already a step in the right direction. Maciej From bonivart at opencsw.org Thu Aug 4 20:05:46 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Thu, 4 Aug 2011 20:05:46 +0200 Subject: [csw-maintainers] moving along... In-Reply-To: References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/4 Maciej Blizi?ski : > It's hard to tell what will require more work, switching to jira or > taming mantis. Assuming we will anyway move from Mantis it's obviously more work to both fix it and migrate from it. > In my experience it's a good idea to make small, effective > steps. I don't think it's efficient when we already see a need to switch to something else. Also, small things tend to grow which will just delay the switch. /peter From maciej at opencsw.org Thu Aug 4 20:53:10 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Thu, 4 Aug 2011 19:53:10 +0100 Subject: [csw-maintainers] moving along... In-Reply-To: References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/4 Peter Bonivart : >> In my experience it's a good idea to make small, effective >> steps. > > I don't think it's efficient when we already see a need to switch to > something else. Also, small things tend to grow which will just delay > the switch. Yes, I agree that one does works against the other. I'm thinking that the migration to jira will require a large amount of work in one go, which I cannot pull off at the moment. Going down that path is a risk: we might start the work on the migration, and not finish it. This way, we might end up with having spent the time and effort, and getting no bug tracker integration. It's about the risk, and I'm personally inclined going the safer path. I would not mind it if someone else spearheaded the jira migration. The things to do would be: 1. set up an instance of jira, 2. migrate the data (is there a tool for such migration?), 3. tackle the problem of existing links to mantis bugs, they shouldn't return 404 errors, 4. integrate with the buildfarm database. It's probably not a one evening kind of effort. How much time would the migration take, do you think? Maciej From maciej at opencsw.org Fri Aug 5 09:15:06 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Fri, 5 Aug 2011 08:15:06 +0100 Subject: [csw-maintainers] Using an architecture-specific includes in GAR Message-ID: Hello maintainers, Suppose I have a package which has different headers for different architectures (e.g. libgmp). What's the best place for the headers? Is the following sane? /opt/csw/include/amd64 d none 0755 root bin CSWlibgmp-dev /opt/csw/include/amd64/gmp.h f none 0644 root bin 86215 39079 1312523810 CSWlibgmp-dev /opt/csw/include/gmp.h f none 0644 root bin 86204 38065 1312523796 CSWlibgmp-dev /opt/csw/include/pentium d none 0755 root bin CSWlibgmp-dev /opt/csw/include/pentium/gmp.h f none 0644 root bin 86208 38680 1312523801 CSWlibgmp-dev Second question: Is there an idiom in GAR that automatically pick the right include file? For example, when I'm building for amd64, I want the /opt/csw/include/amd64 directory to be first in the include search path. I know I can fiddle with EXTRA_INC in my build description, but perhaps there's already something in the GAR code to handle this case. Maciej From maciej at opencsw.org Fri Aug 5 10:41:23 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Fri, 5 Aug 2011 09:41:23 +0100 Subject: [csw-maintainers] how to set up a local buildhost? In-Reply-To: References: <4DEA660B.4080102@opencsw.org> <4DEA8DA4.8000304@opencsw.org> Message-ID: 2011/6/27 Maciej Blizi?ski : > 2011/6/4 John Ellson : >> What about gar scripts? ? Just "pkg-get -i gar" ? > > There is also the mgar package, which I highly recommend. > > It was recently impossible to run checkpkg in an off-buildfarm > setting. ?Last week was a bugfix week for checkpkg, thanks to Igor who > attacked and didn't let go until checkpkg was working. ?Currently, > checkpkg will work off-buildfarm if CATALOG_RELEASE is set to current > or unstable in your ~/.garrc. ?I will send a separate, more detailed > update to the list later in the week. I've pushed an update that displays a more helpful error message: It says what catalog release was requested, and what catalog releases are available in the database. http://sourceforge.net/apps/trac/gar/changeset/15261 Maciej From maciej at opencsw.org Fri Aug 5 12:13:55 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Fri, 5 Aug 2011 11:13:55 +0100 Subject: [csw-maintainers] how the catalog signing works In-Reply-To: <1311439323-sup-2576@pinkfloyd.chass.utoronto.ca> References: <1311439323-sup-2576@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/7/23 Ben Walton : > On the designated zone (named cswsign) there is an account named > catalogsign. ?Anyone needing access to this account should have their > ssh key seeded there. Who would be the people to contact if signing stops working? From maciej at opencsw.org Fri Aug 5 12:25:59 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Fri, 5 Aug 2011 11:25:59 +0100 Subject: [csw-maintainers] how the catalog signing works In-Reply-To: <1311439323-sup-2576@pinkfloyd.chass.utoronto.ca> References: <1311439323-sup-2576@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/7/23 Ben Walton : > The monitor cannot currently emit email so the sign of breakage at the > 3 day interval will be that the generate-unstable script will die when > the signature fails. ?Dago (at least) should get an email in this > situation. What should happen when signing fails - should catalog generation stop and the result not be pushed to the mirror, or should it continue and publish an unsigned catalog? I was thinking that for unstable, it could publish unsigned catalogs, but for named releases, it should stop. By the way, the signing script has currently stopped working, and I've made a change that makes it continue and publish an unsigned catalog. I don't mean that it should be the target state. I did it to be able to continue working on libgmp and libmpfr. Maciej From bwalton at opencsw.org Fri Aug 5 14:21:55 2011 From: bwalton at opencsw.org (Ben Walton) Date: Fri, 05 Aug 2011 08:21:55 -0400 Subject: [csw-maintainers] how the catalog signing works In-Reply-To: References: <1311439323-sup-2576@pinkfloyd.chass.utoronto.ca> Message-ID: <1312546842-sup-6421@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Fri Aug 05 06:25:59 -0400 2011: > What should happen when signing fails - should catalog generation > stop and the result not be pushed to the mirror, or should it > continue and publish an unsigned catalog? I was thinking that for > unstable, it could publish unsigned catalogs, but for named > releases, it should stop. Well, yes, I thought that it should trigger the 'set -e' and modified the catalog generation step accordingly. If we want different policies for unstable vs named release, that's fine with me. > By the way, the signing script has currently stopped working, and > I've made a change that makes it continue and publish an unsigned > catalog. I don't mean that it should be the target state. I did it > to be able to continue working on libgmp and libmpfr. I've re-activated the passphrase in gpg-agent so they should run signed again. When Dago is back, we'll get mail notifications going. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Fri Aug 5 14:26:43 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Fri, 5 Aug 2011 13:26:43 +0100 Subject: [csw-maintainers] Issues with libdnet In-Reply-To: References: Message-ID: 2011/8/5 Andrea Di Pasquale : > Hi guys, > > if you rename /usr/lib/libdnet in /usr/lib/libdnet.so the cmake works well. > Can you rename this file in your package repository please? It's more than just the libdnet.so file. The soname of the library is wrong (libdnet.1) and once that gets into other binaries, the situation gets harder to clean up. If you just want to link it for now, create a libdnet.so symlink in a directory you have the write permission to, and pass it through the -L flag to the linker. I'd like OpenCSW to solve the problem in a clean way. Maciej From bwalton at opencsw.org Fri Aug 5 15:02:27 2011 From: bwalton at opencsw.org (Ben Walton) Date: Fri, 05 Aug 2011 09:02:27 -0400 Subject: [csw-maintainers] how the catalog signing works In-Reply-To: References: <1311439323-sup-2576@pinkfloyd.chass.utoronto.ca> Message-ID: <1312549286-sup-2768@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Fri Aug 05 06:13:55 -0400 2011: > 2011/7/23 Ben Walton : > > On the designated zone (named cswsign) there is an account named > > catalogsign. ?Anyone needing access to this account should have their > > ssh key seeded there. > > Who would be the people to contact if signing stops working? Well, anyone authorized to hold the passphrase, I think. Currently that would be yourself and Ihsan. We can (and should) have at least one more person authorized to re-instate the gpg-agent. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Fri Aug 5 17:43:28 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Fri, 5 Aug 2011 16:43:28 +0100 Subject: [csw-maintainers] Issues with libdnet In-Reply-To: <8C3F6485-EC9D-4706-85BC-8D02CAD5E109@gmail.com> References: <8C3F6485-EC9D-4706-85BC-8D02CAD5E109@gmail.com> Message-ID: 2011/8/5 Andrea Di Pasquale : > sorry but soft link is a bad solution. :\ It's exactly the same solution as renaming /opt/csw/lib/libdnet to /opt/csw/lib/libdnet.so; if you look at it, /opt/csw/libdnet is a symlink. Here is the pkgmap entry: 1 s none /opt/csw/lib/sparcv9/libdnet=libdnet.1.0.1 The "s" stands for symlink. I thought that if you accept the renaming solution, you'd also accept creating a similar symlink in another place. That would cause the wrong soname (libdnet.1) to be used in the compiled binaries, but I guess it's a tangential problem for porting. You could work on your port without waiting for the libdnet fix. > I can't to do the port of ArpON for Solaris for libdnet problem, I'm sorry. The progress so far looks like this: I've created an issue in the libdnet bug tracker[2]. I also investigated the scenario described in Debian docs[4], but regenerating all the autotools files did not help with the issue. I've committed my patches anyway[3] and also submitted them upstream for inclusion[5]. I still don't know what's the core cause of this issue, I think that it's likely a problem with libtool, but I haven't been able to pinpoint it. I've also contacted one of the committers to see if the project is alive; the last commits are from October 2010. I heard that the primary developer is currently heavily occupied with other projects, so we can't expect a quick response from the upstream and sadly, we're on our own. Maciej [1] http://buildfarm.opencsw.org/pkgdb/srv4/c3becaf3633bfdbc0f315b73bc1d798c/ [2] http://code.google.com/p/libdnet/issues/detail?id=18 [3] http://sourceforge.net/apps/trac/gar/changeset/15258 [4] http://www.v7w.com/debian/libtool-missing_so.html [5] http://code.google.com/p/libdnet/issues/detail?id=22 From dam at opencsw.org Fri Aug 5 22:35:06 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Fri, 5 Aug 2011 22:35:06 +0200 Subject: [csw-maintainers] Using an architecture-specific includes in GAR In-Reply-To: References: Message-ID: <16C1D249-39DD-4109-AD06-24D23FB2AA07@opencsw.org> Hi folks, Am 05.08.2011 um 09:15 schrieb Maciej Blizi?ski: > Suppose I have a package which has different headers for different > architectures (e.g. libgmp). What's the best place for the headers? > Is the following sane? > > /opt/csw/include/amd64 d none 0755 root bin CSWlibgmp-dev > /opt/csw/include/amd64/gmp.h f none 0644 root bin 86215 39079 > 1312523810 CSWlibgmp-dev > /opt/csw/include/gmp.h f none 0644 root bin 86204 38065 1312523796 CSWlibgmp-dev > /opt/csw/include/pentium d none 0755 root bin CSWlibgmp-dev > /opt/csw/include/pentium/gmp.h f none 0644 root bin 86208 38680 > 1312523801 CSWlibgmp-dev > > Second question: Is there an idiom in GAR that automatically pick the > right include file? For example, when I'm building for amd64, I want > the /opt/csw/include/amd64 directory to be first in the include search > path. I know I can fiddle with EXTRA_INC in my build description, but > perhaps there's already something in the GAR code to handle this case. (I'll get back on monday, just a quick heads up for now, haven't checked any other emails) The includes can usually be branched by compile-specific flags for the ISA to compile. One notable example for this is curlbuild.h which contains isa-specific definitions of different wide types when compiled on 32 and 64 bits. Traditionally, this has been done by a rename of isa-specific headers and a meta-header with the branch: http://sourceforge.net/apps/trac/gar/browser/csw/mgar/pkg/curl/tags/curl-7.20.0%2CREV%3D2010.02.15/Makefile#L72 http://sourceforge.net/apps/trac/gar/browser/csw/mgar/pkg/curl/tags/curl-7.20.0%2CREV%3D2010.02.15/files/curlbuild.h After that I found out that there is a specific diff-mode which can generate define-shielded diffs in C header format to make just one unified diff: http://sourceforge.net/apps/trac/gar/browser/csw/mgar/pkg/libcurl4/trunk/Makefile#L157 (this was "lend" from some recipe from the OpenSolaris contrib tools tree). I plan to introduce that automatically in GAR to always diff like this between isa-default-32 and isa-default-64 headers which will usually yield in just the 32 bit default headers being the same without any branching. Best regards -- Dago -- "You don't become great by trying to be great, you become great by wanting to do something, and then doing it so hard that you become great in the process." - xkcd #896 From ellson at opencsw.org Sat Aug 6 06:10:45 2011 From: ellson at opencsw.org (John Ellson) Date: Sat, 06 Aug 2011 00:10:45 -0400 Subject: [csw-maintainers] Using an architecture-specific includes in GAR In-Reply-To: References: Message-ID: <4E3CBEC5.5050300@opencsw.org> Maciej, On Fedora, /usr/include/gmp.h is just a wrapper, ... it starts with this comment which explains: /* * This gmp.h is a wrapper include file for the original gmp.h, which has been * renamed to gmp-.h. There are conflicts for the original gmp.h on * multilib systems, which result from arch-specific configuration options. * Please do not use the arch-specific file directly. * * Copyright (C) 2006 Red Hat, Inc. * Thomas Woerner */ #ifdef gmp_wrapper_h #error "gmp_wrapper_h should not be defined!" #endif #define gmp_wrapper_h #if defined(__arm__) #include "gmp-arm.h" #elif defined(__i386__) #include "gmp-i386.h" #elif defined(__ia64__) #include "gmp-ia64.h" #elif defined(__powerpc64__) #include "gmp-ppc64.h" #elif defined(__powerpc__) #include "gmp-ppc.h" #elif defined(__s390x__) #include "gmp-s390x.h" #elif defined(__s390__) #include "gmp-s390.h" #elif defined(__x86_64__) #include "gmp-x86_64.h" #elif defined(__alpha__) #include "gmp-alpha.h" #elif defined(__sh__) #include "gmp-sh.h" #elif defined(__sparc__) && defined (__arch64__) #include "gmp-sparc64.h" #elif defined(__sparc__) #include "gmp-sparc.h" #else #error "The gmp-devel package is not usable with the architecture." #endif #undef gmp_wrapper_h ------------------------------------------------------------------------ John On 08/05/2011 03:15 AM, Maciej Blizi?ski wrote: > Hello maintainers, > > Suppose I have a package which has different headers for different > architectures (e.g. libgmp). What's the best place for the headers? > Is the following sane? > > /opt/csw/include/amd64 d none 0755 root bin CSWlibgmp-dev > /opt/csw/include/amd64/gmp.h f none 0644 root bin 86215 39079 > 1312523810 CSWlibgmp-dev > /opt/csw/include/gmp.h f none 0644 root bin 86204 38065 1312523796 CSWlibgmp-dev > /opt/csw/include/pentium d none 0755 root bin CSWlibgmp-dev > /opt/csw/include/pentium/gmp.h f none 0644 root bin 86208 38680 > 1312523801 CSWlibgmp-dev > > Second question: Is there an idiom in GAR that automatically pick the > right include file? For example, when I'm building for amd64, I want > the /opt/csw/include/amd64 directory to be first in the include search > path. I know I can fiddle with EXTRA_INC in my build description, but > perhaps there's already something in the GAR code to handle this case. > > Maciej > _______________________________________________ > maintainers mailing list > maintainers at lists.opencsw.org > https://lists.opencsw.org/mailman/listinfo/maintainers > .:: This mailing list's archive is public. ::. > From maciej at opencsw.org Sat Aug 6 09:16:23 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sat, 6 Aug 2011 08:16:23 +0100 Subject: [csw-maintainers] Issues with libdnet In-Reply-To: References: <8C3F6485-EC9D-4706-85BC-8D02CAD5E109@gmail.com> Message-ID: 2011/8/5 Maciej Blizi?ski : > I also investigated the scenario described in > Debian docs[4], but regenerating all the autotools files did not help > with the issue. (...) I banged my head against the keyboard[1], which turned out to be a viable strategy to fix the libdnet problem. It seems to me that removing the old aclocal.m4 file from the sources was necessary. After removing it can calling a random sequence of autotools-related commands, libdnet libraries suddenly got their .so extensions back. Updated libdnet packages are available from unstable[2]. Enjoy! Maciej [1] http://sourceforge.net/apps/trac/gar/changeset/15273 [2] http://mirror.opencsw.org/opencsw-future/unstable/ From pfelecan at opencsw.org Sat Aug 6 18:36:44 2011 From: pfelecan at opencsw.org (Peter FELECAN) Date: Sat, 06 Aug 2011 18:36:44 +0200 Subject: [csw-maintainers] moving along... In-Reply-To: ("Maciej =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Thu, 4 Aug 2011 16:59:25 +0100") References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: Maciej Blizi?ski writes: > Mantis is not written for > the number of projects that we have. It is meant to handle a two digit > number of projects, and we have a four digit one Debian is using a BTS for a bigger magnitude, isn't it? What about using their system? Just a thought... -- Peter From bwalton at opencsw.org Sun Aug 7 00:17:35 2011 From: bwalton at opencsw.org (Ben Walton) Date: Sat, 06 Aug 2011 18:17:35 -0400 Subject: [csw-maintainers] web visibility of packages Message-ID: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> Hi All, I'm now looking at the web code to display package info. Unless we're going to modify the way things work so that we can pull primary info from the catalog database backing csw-upload-pkg, we need to complete the mantis integration at the same time as the web package database. The wordpress code uses both databases. Maciej: Would you like to pool resources on this? This _could_ be done as another restful interface that is called by the backend of csw-upload-pkg. For each package in the set of things uploaded, the backend would make a request of the form: /register/pkgname/?maintainer=...&... /deregister/pkgname A re-registration (eg: version bump) would force in fresh information. This would accommodate takeovers as well. Separating this logic to a separate unit would have the advantage that when we're ready to switch BTS we have a nice modular place to do it. Additionally, this interface should manage both the mantis and the other database as a single unit. This would make things a one-stop-shop for the csw-upload-pkg backend and make it easier to fully merge these databases in the future. Am I missing useful commands to the restful interface? It might make sense to have some sort of generalized query too, but I see that as less important than stuffing info into the database. Thoughts? Thanks -Ben [1] I'm in the lets get things running before replacing everything camp for now as I don't have the time to invest in the larger project yet. -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Sun Aug 7 10:54:46 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sun, 7 Aug 2011 09:54:46 +0100 Subject: [csw-maintainers] moving along... In-Reply-To: References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/6 Peter FELECAN : > Maciej Blizi?ski writes: > >> Mantis is not written for >> the number of projects that we have. It is meant to handle a two digit >> number of projects, and we have a four digit one > > Debian is using a BTS for a bigger magnitude, isn't it? What about using > their system? Just a thought... Debian BTS / Debbugs has indeed work with a large number of packages, and the sources are available[1]. However, if you the summary on Wikipedia, it it's an email based system, and although you can search and view bugs from a web interface, the main input and output is email. What I like about it, is that there is the reportbug utility which drives the bug reporting process. Overall, it feels dated. Another system that is in use, is Bugzilla. It's used by e. g. Red Hat. It is a web based system, and if it has an email input, it's only an addon. In the case of a packaging project, the main consideration is this: the user who reports the issue, has a rough idea against which package the issue needs to be filed. It still can be fuzzy in the case of split packages, e.g. cups vs cupsd vs cupsclient, but let's assume that the user knows which package the bugs belongs to. What mantis does for us, is that it shows the list of packages to the user, and knows which maintainer is assigned to which package. This is a great feature. I talked once to a Gentoo developer and their bugzilla installation. I learned that their bugzilla does not have that mapping, and a human is necessary for the bug to be routed to the right person. In this light, Mantis does a pretty good job for us, and if we look for a new system, we need to make sure that it has an easy user interface to map from packages to maintainers. That also requires having some form of abstraction that we can use to represent packages. In Mantis it's projects, but in a different bug tracker it could be a different abstraction, e.g. a component or simply a tag. I'm with Ben on that I'd rather look for a way to restore the catalog?mantis synchronization, the cheaper the better, and go from there. I don't want discourage anyone from looking for a better bugtracker - if we had a better bugtracker, I'm all for it! There already is a wiki page about the topic[3], so if anyone makes progress, I'd encourage them to share it. Maciej [1] http://bugs.debian.org/debbugs-source/ [2] http://en.wikipedia.org/wiki/Debian_bug_tracking_system [3] http://wiki.opencsw.org/project-bugtracker From maciej at opencsw.org Sun Aug 7 11:38:27 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sun, 7 Aug 2011 10:38:27 +0100 Subject: [csw-maintainers] web visibility of packages In-Reply-To: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/6 Ben Walton : > I'm now looking at the web code to display package info. ?Unless we're > going to modify the way things work so that we can pull primary info > from the catalog database backing csw-upload-pkg, we need to complete > the mantis integration at the same time as the web package database. > The wordpress code uses both databases. > > Maciej: Would you like to pool resources on this? ?This _could_ be > done as another restful interface that is called by the backend of > csw-upload-pkg. I'd suggest one modification: Instead of making it part of pkgdb (the backend of csw-upload-pkg), it could be a separate program which would reach out to both pkgdb and mantis for information, and that could be run asynchronously to package uploads. This way, each component would be itself simpler, which is a point for maintainability. > For each package in the set of things uploaded, the > backend would make a request of the form: > > /register/pkgname/?maintainer=...&... > /deregister/pkgname > > A re-registration (eg: version bump) would force in fresh > information. ?This would accommodate takeovers as well. > > Separating this logic to a separate unit would have the advantage that > when we're ready to switch BTS we have a nice modular place to do it. > > Additionally, this interface should manage both the mantis and the > other database as a single unit. ?This would make things a > one-stop-shop for the csw-upload-pkg backend and make it easier to > fully merge these databases in the future. > > Am I missing useful commands to the restful interface? ?It might make > sense to have some sort of generalized query too, but I see that as > less important than stuffing info into the database. There is currently no separate documentation of the RESTful interface. However, the code itself is (I think) not that hard to read. http://sourceforge.net/apps/trac/gar/browser/csw/mgar/gar/v2/lib/web If you look at pkgdb_web,py, you'll see the list of URLs at line 16, which already gives you an idea what kind of queries can you currently make. Each URL maps to a class, and if you look up the class, you can see what HTTP methods are implemented. Usually it's only GET, because the public pkgdb interface is read-only. All the code that allows writing is in releases_web.py, which is available only on the internal network (see example-apache.conf). Looking at pkgdb_web.py, you can see this line: r'/rest/srv4/([0-9a-f]{32})/', 'RestSrv4Detail', This means that you can call http://buildfarm.opencsw.org/pkgdb/rest/srv4// and retrieve package details. For example: maciej at login [login]:~ > gmd5sum .pkg.gz 688d3a5b038d172a36a4cba30272441f .pkg.gz How that we can identify the package by its md5 sum, we can query pkgdb. The last echo is to inject a newline character. maciej at login [login]:~ > curl -s http://buildfarm.opencsw.org/pkgdb/rest/srv4/688d3a5b038d172a36a4cba30272441f/ ; echo {"maintainer_full_name": null, "version_string": "1.8.7p334,REV=2011.03.24", "basename": "ruby18-1.8.7p334,REV=2011.03.24-SunOS5.9-sparc-CSW.pkg.gz", "maintainer_email": "bwalton at opencsw.org", "mtime": "2011-04-01 17:42:37", "file_basename": "ruby18-1.8.7p334,REV=2011.03.24-SunOS5.9-sparc-CSW.pkg.gz", "arch": "sparc", "osrel": "SunOS5.9", "size": 1365872, "md5_sum": "688d3a5b038d172a36a4cba30272441f", "pkgname": "CSWruby18", "rev": "2011.03.24", "filename_arch": "sparc", "version": "1.8.7p334,REV=2011.03.24", "catalogname": "ruby18"} A similar example which parses JSON and pretty-prints it: maciej at login [login]:~ > python -c "import json, urllib2, pprint; pprint.pprint(json.load(urllib2.urlopen('http://buildfarm.opencsw.org/pkgdb/rest/srv4/688d3a5b038d172a36a4cba30272441f/')))" {u'arch': u'sparc', u'basename': u'ruby18-1.8.7p334,REV=2011.03.24-SunOS5.9-sparc-CSW.pkg.gz', u'catalogname': u'ruby18', u'file_basename': u'ruby18-1.8.7p334,REV=2011.03.24-SunOS5.9-sparc-CSW.pkg.gz', u'filename_arch': u'sparc', u'maintainer_email': u'bwalton at opencsw.org', u'maintainer_full_name': None, u'md5_sum': u'688d3a5b038d172a36a4cba30272441f', u'mtime': u'2011-04-01 17:42:37', u'osrel': u'SunOS5.9', u'pkgname': u'CSWruby18', u'rev': u'2011.03.24', u'size': 1365872, u'version': u'1.8.7p334,REV=2011.03.24', u'version_string': u'1.8.7p334,REV=2011.03.24'} We can add more fields to the JSON structure, or add more URLs. This RESTful interface has been written by my pretty much solo, and has not been reviewed. I only implemented as much as I needed to get csw-upload-pkg to work. I'll be happy to receive comments on how to develop it further. Maciej From dam at opencsw.org Sun Aug 7 11:40:29 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Sun, 7 Aug 2011 11:40:29 +0200 Subject: [csw-maintainers] moving along... In-Reply-To: References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: <5403D1FE-6970-46D4-B73E-6A05D3A6150B@opencsw.org> Hi Maciej, Am 04.08.2011 um 19:56 schrieb Maciej Blizi?ski: > 2011/8/4 Peter Bonivart : >> 2011/8/4 Maciej Blizi?ski : >>> So this is where I'll start. Here are the notes about the mantis >>> speedup project. Sebastian has already done a fair bit of work there. >>> >>> http://wiki.opencsw.org/project-mantis-speedup >> >> Isn't this a lot of effort if we're anyway switch to Jira or whatever? >> >> Couldn't we instead: >> >> 1. Register new projects in current Mantis manually when a new package >> is released (not _that_ frequent) >> 2. Put the same effort into getting Jira/? working on the side. >> Installing, testing, migrating. >> >> Thoughts? > > It's hard to tell what will require more work, switching to jira or > taming mantis. I'm personally not inclined to spearhead the jira > migration right now (although I would not mind another person doing > that). In my experience it's a good idea to make small, effective > steps. For example, we could start with a script that doesn't write > to mantis, but queries both systems and tells you the difference; what > to do in mantis to bring it to the state that matches the catalog. > This shouldn't be that hard, and is already a step in the right > direction. As I understand it most of the work is updating Mantis and reapplying the patches. How about making a custom REST interface to Mantis ourselves which just harcode the structures and modify the database directly? From the experience of my (read-only) bug viewer interface that shouldn't be too hard, make a clean interface for now and won't waste resources on things we probably won't need later on. Best regards -- Dago -- "You don't become great by trying to be great, you become great by wanting to do something, and then doing it so hard that you become great in the process." - xkcd #896 From maciej at opencsw.org Sun Aug 7 11:41:59 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sun, 7 Aug 2011 10:41:59 +0100 Subject: [csw-maintainers] Using an architecture-specific includes in GAR In-Reply-To: <16C1D249-39DD-4109-AD06-24D23FB2AA07@opencsw.org> References: <16C1D249-39DD-4109-AD06-24D23FB2AA07@opencsw.org> Message-ID: 2011/8/5 Dagobert Michelsen : > The includes can usually be branched by compile-specific flags for the > ISA to compile. One notable example for this is curlbuild.h which contains > isa-specific definitions of different wide types when compiled on 32 and > 64 bits. Traditionally, this has been done by a rename of isa-specific > headers and a meta-header with the branch: > ?http://sourceforge.net/apps/trac/gar/browser/csw/mgar/pkg/curl/tags/curl-7.20.0%2CREV%3D2010.02.15/Makefile#L72 > ?http://sourceforge.net/apps/trac/gar/browser/csw/mgar/pkg/curl/tags/curl-7.20.0%2CREV%3D2010.02.15/files/curlbuild.h Thanks Dago and John for the suggestions. I've improved the libgmp_dev to include a switch header file, like the one in curl. Maciej From maciej at opencsw.org Sun Aug 7 15:01:46 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sun, 7 Aug 2011 14:01:46 +0100 Subject: [csw-maintainers] moving along... In-Reply-To: <5403D1FE-6970-46D4-B73E-6A05D3A6150B@opencsw.org> References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> <5403D1FE-6970-46D4-B73E-6A05D3A6150B@opencsw.org> Message-ID: 2011/8/7 Dagobert Michelsen : > How about making a custom REST interface to Mantis ourselves > which just harcode the structures and modify the database directly? > From the experience of my (read-only) bug viewer interface that shouldn't > be too hard, make a clean interface for now and won't waste resources > on things we probably won't need later on. Certainly doable. Perhaps we could only implement a minimal set of functions that are not available through the SOAP interface. For instance, the Mantis SOAP API[1] in theory provides the mc_projects_get_user_accessible function, but in practice, it only returns a "(500, Null)" tuple, whatever that means. If the other functions (e.g. add a new project) work, we could use SOAP to do these. The integration job that we need, would be something like: - get the list of projects in mantis - get the list of packages in $catalog_release - generate the diff - take renames into account (not yet sure how) - add missing packages to mantis - print the list of projects in mantis with no corresponding packages (I wouldn't recommend deleting them automatically) - make sure that the package - maintainer assignments are up to date / correct Perhaps more. As far as reading, I thought about cheating and parsing HTML... Maciej [1] http://www.mantisbt.org/bugs/api/soap/mantisconnect.php From dam at opencsw.org Sun Aug 7 16:36:31 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Sun, 7 Aug 2011 16:36:31 +0200 Subject: [csw-maintainers] moving along... In-Reply-To: References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> <5403D1FE-6970-46D4-B73E-6A05D3A6150B@opencsw.org> Message-ID: <9C0E4619-5F04-4934-8CC6-D7E61C2FC989@opencsw.org> Hi Maciej, Am 07.08.2011 um 15:01 schrieb Maciej Blizi?ski: > 2011/8/7 Dagobert Michelsen : >> How about making a custom REST interface to Mantis ourselves >> which just harcode the structures and modify the database directly? >> From the experience of my (read-only) bug viewer interface that shouldn't >> be too hard, make a clean interface for now and won't waste resources >> on things we probably won't need later on. > > Certainly doable. Perhaps we could only implement a minimal set of > functions that are not available through the SOAP interface. For > instance, the Mantis SOAP API[1] in theory provides the > mc_projects_get_user_accessible function, but in practice, it only > returns a "(500, Null)" tuple, whatever that means. If the other > functions (e.g. add a new project) work, we could use SOAP to do > these. If was thinking of completely ignoring the broken, incomplete API and do something separetely modifying the database directly. > The integration job that we need, would be something like: > > - get the list of projects in mantis Easy, I have SQL for that. > - get the list of packages in $catalog_release > - generate the diff > - take renames into account (not yet sure how) > - add missing packages to mantis This needs to be looked at. Should be doable in an hour or so. Do we still have mysqladmin or something on the database somewhere? > - print the list of projects in mantis with no corresponding packages > (I wouldn't recommend deleting them automatically) > - make sure that the package - maintainer assignments are up to date / correct > > Perhaps more. As far as reading, I thought about cheating and parsing HTML... of packages/* ? Should not be necessary as reading the DB directly is pretty straight forward. Best regards -- Dago -- "You don't become great by trying to be great, you become great by wanting to do something, and then doing it so hard that you become great in the process." - xkcd #896 From bwalton at opencsw.org Mon Aug 8 03:02:48 2011 From: bwalton at opencsw.org (Ben Walton) Date: Sun, 07 Aug 2011 21:02:48 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> Message-ID: <1312765060-sup-1798@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Sun Aug 07 05:38:27 -0400 2011: > I'd suggest one modification: Instead of making it part of pkgdb > (the backend of csw-upload-pkg), it could be a separate program > which would reach out to both pkgdb and mantis for information, and > that could be run asynchronously to package uploads. This way, each > component would be itself simpler, which is a point for > maintainability. This is workable, but is asynchronous action desirable? If I upload a new package to unstable, I'd like it to be visible right away in mantis and on the web. It is a nice point that it wouldn't require modification of pkgdb and thus increases the modularity and (decreases the coupling) of things. As long as it ran reasonably freqently (maybe tied to the catalog updating?) the delay between push and publish wouldn't be all that bad. I think I like Dago's leanings in the mantis area...screw the api. Hit the db directly. That's what the old scripts were doing and Dago indicates that it isn't too bad to decipher the table structure. > Looking at pkgdb_web.py, you can see this line: > > r'/rest/srv4/([0-9a-f]{32})/', 'RestSrv4Detail', This is quite easy to parse and understand. > We can add more fields to the JSON structure, or add more URLs. > This RESTful interface has been written by my pretty much solo, and > has not been reviewed. I only implemented as much as I needed to > get csw-upload-pkg to work. I'll be happy to receive comments on > how to develop it further. It looks reasonable for data extraction, so that's enough to start with. If it requires modification for some reason, we can address that. A further advantage of this design is that there would be no daemon requirement. A simple cron job would suffice. So, that would leave us with a job that: * Pulls each catalog from pkgdb every period. * If the entry exists in mantis, ensure the owner is correct. * If the entry doesn't exist, add it. * Do the same for the web package database. The owner check will see each package queried (based on the md5sum in the catalog) so the number of queries will grow linearly with the catalog size. Maybe we should only do owner checks daily or weekly instead of at every run? The package rename problem stems from the fact that we use catalog name as the primary key in mantis but package name most everywhere else. This leads me to think about package removals...a rename in pkgdb is a remove and then an add. If this maintenance were tied synchronously to pkgdb interaction, renames are no longer a problem for mantis maintenance. Sticking with the asynchronous model though, removals aren't bad. We'd need to extract the full list of packages from mantis and compare to the catalog, but that's pretty easy. I think that if we start at the mirror and use the generated catalogs from bldcat, we get the perfect starting point to ensure mantis and the web db are up to date... Before I break ground and start writing code, does this sound ok to everyone? Any obvious gotchas here? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Mon Aug 8 11:13:43 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 8 Aug 2011 10:13:43 +0100 Subject: [csw-maintainers] web visibility of packages In-Reply-To: <1312765060-sup-1798@pinkfloyd.chass.utoronto.ca> References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1312765060-sup-1798@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/8 Ben Walton : > Excerpts from Maciej Blizi?ski's message of Sun Aug 07 05:38:27 -0400 2011: > > >> I'd suggest one modification: Instead of making it part of pkgdb >> (the backend of csw-upload-pkg), it could be a separate program >> which would reach out to both pkgdb and mantis for information, and >> that could be run asynchronously to package uploads. ?This way, each >> component would be itself simpler, which is a point for >> maintainability. > > This is workable, but is asynchronous action desirable? ?If I upload a > new package to unstable, I'd like it to be visible right away in > mantis and on the web. It could be a matter of setting the expectations right. Currently, you run the utility, and you wait up to 3h for the confirmation email. It's reasonable to have a similarly working mantis integration too. > It is a nice point that it wouldn't require modification of pkgdb and > thus increases the modularity and (decreases the coupling) of things. > As long as it ran reasonably freqently (maybe tied to the catalog > updating?) the delay between push and publish wouldn't be all that > bad. Yes, it makes sense to run it right after the existing db?disk integration. > I think I like Dago's leanings in the mantis area...screw the api. > Hit the db directly. ?That's what the old scripts were doing and Dago > indicates that it isn't too bad to decipher the table structure. Fair enough. We'll do that. > So, that would leave us with a job that: > > * Pulls each catalog from pkgdb every period. The assumption behind the way mantis integration was handled in the past, is that there is only one catalog ('current'). What if we had e.g. package foo, released by maintainer A to testing and by maintainer B to unstable? Perhaps we should for now assume that there is only one catalog ('unstable'), against which bugs are filed. Even if maintainer A is the person who released the package to testing, it's the maintainer B who is the person actively working on the package. > * If the entry exists in mantis, ensure the owner is correct. > * If the entry doesn't exist, add it. > * Do the same for the web package database. > > The owner check will see each package queried (based on the md5sum in > the catalog) so the number of queries will grow linearly with the > catalog size. ?Maybe we should only do owner checks daily or weekly > instead of at every run? The integration job could keep a cache of the md5sum?maintainer mapping and only query the db for new packages. > The package rename problem stems from the fact that we use catalog > name as the primary key in mantis but package name most everywhere > else. ?This leads me to think about package removals...a rename in > pkgdb is a remove and then an add. ?If this maintenance were tied > synchronously to pkgdb interaction, renames are no longer a problem > for mantis maintenance. csw-upload-pkg doesn't understand the notion of a rename, so this information couldn't be used in the mantis integration. > Sticking with the asynchronous model though, removals aren't bad. > We'd need to extract the full list of packages from mantis and compare > to the catalog, but that's pretty easy. We could start with just displaying the information what would needed to be done for the two to match. > I think that if we start at the mirror and use the generated catalogs > from bldcat, we get the perfect starting point to ensure mantis and > the web db are up to date... +1 > Before I break ground and start writing code, does this sound ok to > everyone? ?Any obvious gotchas here? You are thinking about writing the buildfarm?package_db integration, right? Maciej From bonivart at opencsw.org Mon Aug 8 11:30:26 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Mon, 8 Aug 2011 11:30:26 +0200 Subject: [csw-maintainers] web visibility of packages In-Reply-To: References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1312765060-sup-1798@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/8 Maciej Blizi?ski : > The assumption behind the way mantis integration was handled in the > past, is that there is only one catalog ('current'). ?What if we had > e.g. package foo, released by maintainer A to testing and by > maintainer B to unstable? ?Perhaps we should for now assume that there > is only one catalog ('unstable'), against which bugs are filed. ?Even > if maintainer A is the person who released the package to testing, > it's the maintainer B who is the person actively working on the > package. There's probably not much to gain from having the mapping for, e.g., the stable maintainer, if that maintainer is currently active he will most likely just refer the bug to the unstable maintainer anyway. I think all mappings should be to the unstable catalog since that's where stuff is happening currently. The package browser should be able to display all variants of arch and catalog, I've always thought that was a weakness with the current one. James Lee's browser can do that but only by choosing an entry point, I think it would be nicer if we just used a table to display the variants on the same page. But the most important thing for me is when can we expect catalog updates again for current? Last I heard it was e-mail lacking from the signing host, Dago is now back. Or is that for Ihsan to take care of? /peter From maciej at opencsw.org Mon Aug 8 11:49:07 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 8 Aug 2011 10:49:07 +0100 Subject: [csw-maintainers] GCC 4.6 progress Message-ID: Let's continue the effort to build the new GCC. I've updated GMP and MPFR last week, the GCC build seems to be happy with them. There's one more library to build, MPC. Unfortunately, this week is quite choppy for me, I won't be able to make progress, could anyone build (or at least start building) this library this week? http://www.multiprecision.org/ I would like to ask Dago, to update, if possible, the gar_override.mk file, which seems to be a pre-platforms collection of settings for GAR. Could you update it so that it's compatible with platforms in GAR? http://sourceforge.net/apps/trac/gar/browser/csw/mgar/pkg/gcc4/trunk/files/gar_override.mk Maciej From bonivart at opencsw.org Mon Aug 8 16:58:40 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Mon, 8 Aug 2011 16:58:40 +0200 Subject: [csw-maintainers] Technical Summercamp in Kiel 2011 In-Reply-To: References: <958B6EC9-5CE1-4B2E-BE39-2C85617D3EDD@opencsw.org> Message-ID: 2011/8/1 Maciej Blizi?ski : > From the current state of the poll it looks like the optimal date is > the weekend of 17-18 September 2011. So far, 7 people responded to the > poll. I thought we could personally invite some maintainers to the > summercamp. ?There might be people who just didn't think about coming. > ?At the same time, our project is at relatively large changes (the > automatic process), so the more people are there to discuss the future > direction, the better. Could we maybe reduce the choice to the two most popular weekends to simplify things? September 17-18 and 24-25 are the ones where most people can attend currently. Honestly, the earlier weekends are now too close and we need to set a date to book transport and hotels. I would like to have it September 17-18 myself but two others have it marked with a maybe, that's Ihsan and Rafael. Could they be persuaded to make it a yes? :) The weekend of September 24-25 looks good for everyone but myself unfortunately. :( If there are others who want to join one of these weekends, please sign up in the Doodle poll now! http://doodle.com/ewfzuze75afweueq /peter From maciej at opencsw.org Mon Aug 8 20:37:14 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 8 Aug 2011 19:37:14 +0100 Subject: [csw-maintainers] Technical Summercamp in Kiel 2011 In-Reply-To: References: <958B6EC9-5CE1-4B2E-BE39-2C85617D3EDD@opencsw.org> Message-ID: 2011/8/8 Peter Bonivart : > I would like to have it September 17-18 myself but two others have it > marked with a maybe, that's Ihsan and Rafael. Could they be persuaded > to make it a yes? :) I think that these options mean "yes if necessary", meaning that they'd prefer other dates but they are willing to come if there's no better choice. So do we settle for the weekend of 17-18 September? Maciej From bwalton at opencsw.org Tue Aug 9 02:59:05 2011 From: bwalton at opencsw.org (Ben Walton) Date: Mon, 08 Aug 2011 20:59:05 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1312765060-sup-1798@pinkfloyd.chass.utoronto.ca> Message-ID: <1312851355-sup-6124@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Mon Aug 08 05:13:43 -0400 2011: > > Before I break ground and start writing code, does this sound ok to > > everyone? ?Any obvious gotchas here? > > You are thinking about writing the buildfarm?package_db integration, > right? The tool I'm going to write will ensure that www.opencsw.org/packages and mantis both have current information at their disposal. It will be asynchronous and use the catalog in the mirror tree as the authoritative source of packages. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bwalton at opencsw.org Tue Aug 9 03:02:34 2011 From: bwalton at opencsw.org (Ben Walton) Date: Mon, 08 Aug 2011 21:02:34 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1312765060-sup-1798@pinkfloyd.chass.utoronto.ca> Message-ID: <1312851550-sup-165@pinkfloyd.chass.utoronto.ca> Excerpts from Peter Bonivart's message of Mon Aug 08 05:30:26 -0400 2011: > There's probably not much to gain from having the mapping for, e.g., > the stable maintainer, if that maintainer is currently active he > will most likely just refer the bug to the unstable maintainer > anyway. I think all mappings should be to the unstable catalog since > that's where stuff is happening currently. I think this is pretty accurate. No point in making it more complex than is required...not to say we can't improve it in the future, but presently we're just getting the functionality we used to have fully automated. > The package browser should be able to display all variants of arch > and catalog, I've always thought that was a weakness with the > current one. James Lee's browser can do that but only by choosing > an entry point, I think it would be nicer if we just used a table to > display the variants on the same page. Given what I've seen so far, I think this will require database changes as not enough info is stored. It certainly would be nice, especially as we transition to named releases. Initially though, lets just get 'out of the hole' and moving. Improvements can be made later. > But the most important thing for me is when can we expect catalog > updates again for current? Last I heard it was e-mail lacking from > the signing host, Dago is now back. Or is that for Ihsan to take > care of? We need to be able to relay mail out of the cswsign host and that's not presently possible. Maciej: Is there anything standing in the way of doing the unstable->current integration using the merge tool? Can we simply install the new gpg key in cswsign when mail is ready and push changes with the idea that the web and mantis will 'catch up' afterward? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Tue Aug 9 10:10:34 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 9 Aug 2011 09:10:34 +0100 Subject: [csw-maintainers] web visibility of packages In-Reply-To: <1312851550-sup-165@pinkfloyd.chass.utoronto.ca> References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1312765060-sup-1798@pinkfloyd.chass.utoronto.ca> <1312851550-sup-165@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/9 Ben Walton : > Maciej: Is there anything standing in the way of doing the > unstable->current integration using the merge tool? ?Can we simply > install the new gpg key in cswsign when mail is ready and push changes > with the idea that the web and mantis will 'catch up' afterward? There's nothing standing in the way, we can use the tool and to the integration. What do you think about the layout currently presented at opencsw-future? http://mirror.opencsw.org/opencsw-future/ One of the major points: 'current' is now a symlink to dublin, and we would integrate from unstable to dublin. Maciej From maciej at opencsw.org Tue Aug 9 12:53:44 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 9 Aug 2011 11:53:44 +0100 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: References: Message-ID: 2011/8/8 Maciej Blizi?ski : > Unfortunately, this week is quite > choppy for me, I won't be able to make progress, could anyone build > (or at least start building) this library this week? > > http://www.multiprecision.org/ I took a quick stab at it and it turned out to be an easy build, no patching required. Once that's installed on the buildfarm, I'll see what will the gcc configure script say next. I'm also waiting for Dago to kindly update gar_override.mk. Maciej From maciej at opencsw.org Tue Aug 9 20:38:02 2011 From: maciej at opencsw.org (Maciej (Matchek) Blizinski) Date: Tue, 9 Aug 2011 19:38:02 +0100 Subject: [csw-maintainers] sparcv8plus with GCC Message-ID: Does anyone know if there's a fundamental problem with building sparcv8plus with GCC? My attempt ended with: /opt/csw/gcc4/bin/gcc -std=gnu99 -c -DHAVE_CONFIG_H -I. -I.. -D__GMP_WITHIN_GMP -I.. -DOPERATION_submul_1 -I/opt/csw/include -mcpu=v8 -mv8plus tmp-submul_1.s -fPIC -DPIC -o .libs/submul_1.o tmp-mul_1.s: Assembler messages: tmp-mul_1.s:84: Error: Architecture mismatch on "stx". tmp-mul_1.s:84: (Requires v9|v9a|v9b; requested architecture is v8.) Or is it a problem with the flags passed to GCC? Maciej From bwalton at opencsw.org Wed Aug 10 03:01:52 2011 From: bwalton at opencsw.org (Ben Walton) Date: Tue, 09 Aug 2011 21:01:52 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1312765060-sup-1798@pinkfloyd.chass.utoronto.ca> <1312851550-sup-165@pinkfloyd.chass.utoronto.ca> Message-ID: <1312937844-sup-4192@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Tue Aug 09 04:10:34 -0400 2011: > There's nothing standing in the way, we can use the tool and to the > integration. Is it easy (or possible) to ask the tool for only 2+ week old packages or does it simply do all differences. (Personally I'm ok with a full push in this manual case as we need to open the flow again. This goes against the idea of unstable though, which is to have a gating period.) We do need signing in place though. Dago, any ideas about getting mail from cswsign to the world? I'd happily do an exim config on the web host in the buildfarm that can accept mail from internal systems and relay to mail.opencsw.org if you'd like...maybe Baltic Online has something else in place for this though? > What do you think about the layout currently presented at opencsw-future? > > http://mirror.opencsw.org/opencsw-future/ We need to get pkgutil-ALL.pkg in place there, but otherwise it looks good to me. > One of the major points: 'current' is now a symlink to dublin, and > we would integrate from unstable to dublin. Yes, I used the wrong names, but the idea is the same. I think we should push anything from unstable to dublin that has been there for 2+ weeks. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bwalton at opencsw.org Wed Aug 10 03:07:08 2011 From: bwalton at opencsw.org (Ben Walton) Date: Tue, 09 Aug 2011 21:07:08 -0400 Subject: [csw-maintainers] gpg key for signing Message-ID: <1312938121-sup-613@pinkfloyd.chass.utoronto.ca> Hi All, It seems that the gpg key has been so well protected that not even we can access it now. One half of the key is presently inaccessible and although there is some chance of recovery yet, I'd propose that we move on with a brand new signing key. I think we should do this even if the old key is recovered. The new key would still be held by the secretary and the treasurer and possibly by a third person able to unlock it for use on cswsign if we decide this is beneficial and select a designated person. Any objections? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Wed Aug 10 10:35:50 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 10 Aug 2011 09:35:50 +0100 Subject: [csw-maintainers] __sync_fetch_and_add_4 woes Message-ID: __sync_fetch_and_add_4 has been biting my badly in recent days, when was working towards the new GCC. It's not a new problem, you can find a number of past postings to the mailing list about this issue. Our porting FAQ suggests a workaround: disable optimization. The problem occurs with i386 and sparcv8, and I need to set -O0 to make these ISAs build. I suppose that for a less used program it could be fine, but I'm building the compiler and its dependent libraries. Are we OK with GMP, PPL and their client, GCC, built with no optimization? Maciej [1] http://lists.opencsw.org/pipermail/maintainers/2009-December/010834.html [2] http://lists.opencsw.org/pipermail/maintainers/2010-October/012869.html [3] http://opencsw.wikidot.com/porting-faq#toc6 From maciej at opencsw.org Wed Aug 10 12:39:00 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 10 Aug 2011 11:39:00 +0100 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: References: Message-ID: The next library was PPL. PPL needed the C++ bindings in GMP, so I needed to rebuild it again, with C++ enabled. I also realized that I have to rebuild it with GCC, because of different name mangling between GCC and Solaris Studio. I worked around the __sync_fetch_and_add_4 problem by compiling with -O0, which I think sucks, but it gets me the built library, so I went with it for now. PPL accepted the updated GMP and built, but there are 2 failing tests. I've inspected one of them, which turns out to be a coredump: maciej at unstable9s :~/src/opencsw/pkg/ppl/trunk > work/solaris9-sparc/build-isa-sparcv8/ppl-0.11.2/tests/Partially_Reduced_Product/bounds1 Bus Error (core dumped) maciej at unstable9s :~/src/opencsw/pkg/ppl/trunk > pstack ./core core './core' of 5719: /home/maciej/src/opencsw/pkg/ppl/trunk/work/solaris9-sparc/build-isa-s fed5e4d4 emutls_destroy (bb218, fed5e494, 0, bb618, 1, ffbff80c) + 40 febd6aa8 keys_destruct (0, 9f384, 5068c, feb8bb08, 5, 0) + 68 feb9cce8 _exithandle (511c4, 2, fb, 10034, ffffffff, fec3c000) + 70 fec207c0 exit (0, ffbff914, ffbff91c, b8fdc, 7d8, ffbffa34) + 24 000511c4 _start (0, ffbff914, 1, ff3dc608, ff3ee834, ff3ee000) + 64 Not sure what the problem is, maybe it's a similar problem to the one with GMP headers. If anyone wants to jump and help debugging, help would be appreciated. Maciej From dam at opencsw.org Wed Aug 10 13:47:01 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Wed, 10 Aug 2011 13:47:01 +0200 Subject: [csw-maintainers] Technical Summercamp in Kiel 2011 In-Reply-To: References: <958B6EC9-5CE1-4B2E-BE39-2C85617D3EDD@opencsw.org> Message-ID: Hi folks, Am 08.08.2011 um 16:58 schrieb Peter Bonivart: > 2011/8/1 Maciej Blizi?ski : >> From the current state of the poll it looks like the optimal date is >> the weekend of 17-18 September 2011. So far, 7 people responded to the >> poll. I thought we could personally invite some maintainers to the >> summercamp. There might be people who just didn't think about coming. >> At the same time, our project is at relatively large changes (the >> automatic process), so the more people are there to discuss the future >> direction, the better. > > Could we maybe reduce the choice to the two most popular weekends to > simplify things? September 17-18 and 24-25 are the ones where most > people can attend currently. Honestly, the earlier weekends are now > too close and we need to set a date to book transport and hotels. > > I would like to have it September 17-18 myself but two others have it > marked with a maybe, that's Ihsan and Rafael. Could they be persuaded > to make it a yes? :) Ok then, I have closed the poll for 17-18. I have 8 rooms reserved for "OpenCSW" for booking until the 20th of August (ten days from now) in the hotel Astor as referenced from the webpage: http://wiki.opencsw.org/summercamp-2011 *** Please make sure to book until that date if possible *** Previous camps have shown that it is useful to have some more time, so you may want to arrive on friday or even thursday and stay until monday for enhanced hacking. Best regards -- Dago -- "You don't become great by trying to be great, you become great by wanting to do something, and then doing it so hard that you become great in the process." - xkcd #896 From maciej at opencsw.org Wed Aug 10 19:01:04 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 10 Aug 2011 18:01:04 +0100 Subject: [csw-maintainers] Technical Summercamp in Kiel 2011 In-Reply-To: References: <958B6EC9-5CE1-4B2E-BE39-2C85617D3EDD@opencsw.org> Message-ID: 2011/8/10 Dagobert Michelsen : > Ok then, I have closed the poll for 17-18. There's a problem with the date for me: there is a work event I need to attend, which showed up after I filled in the doodle poll. If we do the camp on the weekend of the 17th, I can't come! Maciej From pfelecan at opencsw.org Wed Aug 10 20:56:59 2011 From: pfelecan at opencsw.org (Peter FELECAN) Date: Wed, 10 Aug 2011 20:56:59 +0200 Subject: [csw-maintainers] __sync_fetch_and_add_4 woes In-Reply-To: ("Maciej =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Wed, 10 Aug 2011 09:35:50 +0100") References: Message-ID: Maciej Blizi?ski writes: > __sync_fetch_and_add_4 has been biting my badly in recent days, when > was working towards the new GCC. It's not a new problem, you can find > a number of past postings to the mailing list about this issue. Our > porting FAQ suggests a workaround: disable optimization. The problem > occurs with i386 and sparcv8, and I need to set -O0 to make these ISAs > build. I suppose that for a less used program it could be fine, but > I'm building the compiler and its dependent libraries. Are we OK with > GMP, PPL and their client, GCC, built with no optimization? For those 2 clunky platforms: yes; for the others: no. Don't ask... -- Peter From bwalton at opencsw.org Thu Aug 11 03:59:02 2011 From: bwalton at opencsw.org (Ben Walton) Date: Wed, 10 Aug 2011 21:59:02 -0400 Subject: [csw-maintainers] interesting solaris resource Message-ID: <1313027599-sup-8123@pinkfloyd.chass.utoronto.ca> Hi All, I met Thomas (Cc'd) on freenode today and he is the man behind http://wesunsolve.net, which aims to fill a gap left in the Solaris ecosystem after the Oracle takeover. He asked if we'd mind linking to his site from ours and I think that's a fine idea as many of our users would be happy to find this tool. (There is an update in the works too, which can be seen at http://dev.wesunsolve.net.) Dago suggested creating a "Friends" section in the site where we link to things like this. Any objections? Any suggestions on how to place these links nicely? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Thu Aug 11 10:16:05 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Thu, 11 Aug 2011 09:16:05 +0100 Subject: [csw-maintainers] Technical Summercamp in Kiel 2011 In-Reply-To: References: <958B6EC9-5CE1-4B2E-BE39-2C85617D3EDD@opencsw.org> Message-ID: 2011/8/10 Maciej Blizi?ski : > 2011/8/10 Dagobert Michelsen : >> Ok then, I have closed the poll for 17-18. > > There's a problem with the date for me: there is a work event I need > to attend, which showed up after I filled in the doodle poll. ?If we > do the camp on the weekend of the 17th, I can't come! Peter, Dago and I met on IRC this morning to discuss possible solutions. Almost everyone declared the next weekend (24-25 of September) as available. Peter figured out a way to make that weekend available for him, so all 7 people are available now. In conclusion, we're changing the date of Summercamp 2011 to 24-25 of September. Sorry for the inconvenience! Maciej From maciej at opencsw.org Thu Aug 11 22:01:35 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Thu, 11 Aug 2011 21:01:35 +0100 Subject: [csw-maintainers] web visibility of packages In-Reply-To: <1312937844-sup-4192@pinkfloyd.chass.utoronto.ca> References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1312765060-sup-1798@pinkfloyd.chass.utoronto.ca> <1312851550-sup-165@pinkfloyd.chass.utoronto.ca> <1312937844-sup-4192@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/10 Ben Walton : > Excerpts from Maciej Blizi?ski's message of Tue Aug 09 04:10:34 -0400 2011: > >> There's nothing standing in the way, we can use the tool and to the >> integration. > > Is it easy (or possible) to ask the tool for only 2+ week old packages > or does it simply do all differences. ?(Personally I'm ok with a full > push in this manual case as we need to open the flow again. ?This goes > against the idea of unstable though, which is to have a gating > period.) It only does the full diff right now. It could be augmented to do that, the information is there already (the rev field). >> One of the major points: 'current' is now a symlink to dublin, and >> we would integrate from unstable to dublin. > > Yes, I used the wrong names, but the idea is the same. ?I think we > should push anything from unstable to dublin that has been there for > 2+ weeks. +1 Maciej From ihsan at opencsw.org Sat Aug 13 20:14:46 2011 From: ihsan at opencsw.org (=?UTF-8?B?xLBoc2FuwqBEb8SfYW4=?=) Date: Sat, 13 Aug 2011 20:14:46 +0200 Subject: [csw-maintainers] Technical Summercamp in Kiel 2011 In-Reply-To: References: <958B6EC9-5CE1-4B2E-BE39-2C85617D3EDD@opencsw.org> Message-ID: <4E46BF16.50602@opencsw.org> Am 11.08.2011 10:16, schrieb Maciej Blizi?ski: >> There's a problem with the date for me: there is a work event I need >> to attend, which showed up after I filled in the doodle poll. If we >> do the camp on the weekend of the 17th, I can't come! > > Peter, Dago and I met on IRC this morning to discuss possible > solutions. Almost everyone declared the next weekend (24-25 of > September) as available. Peter figured out a way to make that weekend > available for him, so all 7 people are available now. > > In conclusion, we're changing the date of Summercamp 2011 to 24-25 of September. Just booked the hotel. I will arrive on Friday afternoon and leave on Monday noon. See you guys soon in Kiel. :-) Ihsan -- ihsan at dogan.ch http://blog.dogan.ch/ From maciej at opencsw.org Mon Aug 15 00:23:47 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sun, 14 Aug 2011 23:23:47 +0100 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: References: Message-ID: I'm currently stuck at the following error: maciej at unstable9s :~/src/opencsw/pkg/gcc4/branches/bootstrap-4.6 > (cd /home/maciej/src/opencsw/pkg/gcc4/branches/bootstrap-4.6/work/solaris9-sparc/build-isa-sparcv8/gcc-4.6.1/host-sparc-sun-solaris2.9/lto-plugin; /bin/bash ./libtool --tag=CC --tag=disable-static --mode=link /opt/SUNWspro/bin/cc -g -module -bindir /opt/csw/gcc4/libexec/gcc/sparc-sun-solaris2.9/4.6.1 -m32 -xarch=v8 -L/opt/csw/lib -o liblto_plugin.la -rpath /opt/csw/gcc4/libexec/gcc/sparc-sun-solaris2.9/4.6.1 lto-plugin.lo -Wc,../libiberty/pic/libiberty.a) libtool: link: /opt/SUNWspro/bin/cc -shared .libs/lto-plugin.o -L/opt/csw/lib -lc -m32 -xarch=v8 ../libiberty/pic/libiberty.a -Wl,-soname -Wl,liblto_plugin.so.0 -o .libs/liblto_plugin.so.0.0.0 ld: warning: option -o appears more than once, first setting taken ld: fatal: file liblto_plugin.so.0: open failed: No such file or directory ld: fatal: File processing errors. No output written to .libs/liblto_plugin.so.0.0.0 Does anyone have an idea where to look for diagnostic information? Maciej From bwalton at opencsw.org Mon Aug 15 02:00:03 2011 From: bwalton at opencsw.org (Ben Walton) Date: Sun, 14 Aug 2011 20:00:03 -0400 Subject: [csw-maintainers] gpg key for signing In-Reply-To: <1312938121-sup-613@pinkfloyd.chass.utoronto.ca> References: <1312938121-sup-613@pinkfloyd.chass.utoronto.ca> Message-ID: <1313366233-sup-2282@pinkfloyd.chass.utoronto.ca> Excerpts from Ben Walton's message of Tue Aug 09 21:07:08 -0400 2011: > Any objections? Maciej and Ihsan, would you please generate a new key and install it into the catalogsign account on cswsign in the buildfarm? I'll ensure that you're notified when the passphrase times out. We should maybe add a third person with the ability to reset the passphrase, but that can be a separate discussion. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bwalton at opencsw.org Mon Aug 15 02:07:04 2011 From: bwalton at opencsw.org (Ben Walton) Date: Sun, 14 Aug 2011 20:07:04 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> Message-ID: <1313366514-sup-1442@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Sun Aug 07 05:38:27 -0400 2011: Hi Maciej, > {"maintainer_full_name": null, "version_string": > "1.8.7p334,REV=2011.03.24", "basename": > "ruby18-1.8.7p334,REV=2011.03.24-SunOS5.9-sparc-CSW.pkg.gz", > "maintainer_email": "bwalton at opencsw.org", "mtime": "2011-04-01 > 17:42:37", "file_basename": > "ruby18-1.8.7p334,REV=2011.03.24-SunOS5.9-sparc-CSW.pkg.gz", "arch": > "sparc", "osrel": "SunOS5.9", "size": 1365872, "md5_sum": > "688d3a5b038d172a36a4cba30272441f", "pkgname": "CSWruby18", "rev": > "2011.03.24", "filename_arch": "sparc", "version": > "1.8.7p334,REV=2011.03.24", "catalogname": "ruby18"} I'd like to get more information than what is currently available. I suspect that this will require a schema change and thus a full re-index of the package tree. The current database table presented at www/packages stores: $software $pkgname $desc $version $vendorurl $email $maintlogin $repourl The things that I'm missing are $vendorurl and $repourl. Is it possible to extend pkgdb such that this is collected? It would save having to fetch each updated package, unroll it and extract the info from it directly. As the pkgdb rest calls will already be expensive, I'd like to avoid this additional overhead if possible. What do you think? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bwalton at opencsw.org Mon Aug 15 02:08:07 2011 From: bwalton at opencsw.org (Ben Walton) Date: Sun, 14 Aug 2011 20:08:07 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1312765060-sup-1798@pinkfloyd.chass.utoronto.ca> <1312851550-sup-165@pinkfloyd.chass.utoronto.ca> <1312937844-sup-4192@pinkfloyd.chass.utoronto.ca> Message-ID: <1313366857-sup-3367@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Thu Aug 11 16:01:35 -0400 2011: > It only does the full diff right now. It could be augmented to do > that, the information is there already (the rev field). > > >> One of the major points: 'current' is now a symlink to dublin, and > >> we would integrate from unstable to dublin. > > > > Yes, I used the wrong names, but the idea is the same. ?I think we > > should push anything from unstable to dublin that has been there for > > 2+ weeks. > > +1 Could either you or Dago (or someone else equally as familiar) make this a reality? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Mon Aug 15 09:23:02 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 15 Aug 2011 08:23:02 +0100 Subject: [csw-maintainers] web visibility of packages In-Reply-To: <1313366514-sup-1442@pinkfloyd.chass.utoronto.ca> References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1313366514-sup-1442@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/15 Ben Walton : > Excerpts from Maciej Blizi?ski's message of Sun Aug 07 05:38:27 -0400 2011: > > Hi Maciej, > >> {"maintainer_full_name": null, "version_string": >> "1.8.7p334,REV=2011.03.24", "basename": >> "ruby18-1.8.7p334,REV=2011.03.24-SunOS5.9-sparc-CSW.pkg.gz", >> "maintainer_email": "bwalton at opencsw.org", "mtime": "2011-04-01 >> 17:42:37", "file_basename": >> "ruby18-1.8.7p334,REV=2011.03.24-SunOS5.9-sparc-CSW.pkg.gz", "arch": >> "sparc", "osrel": "SunOS5.9", "size": 1365872, "md5_sum": >> "688d3a5b038d172a36a4cba30272441f", "pkgname": "CSWruby18", "rev": >> "2011.03.24", "filename_arch": "sparc", "version": >> "1.8.7p334,REV=2011.03.24", "catalogname": "ruby18"} > > I'd like to get more information than what is currently available. > I suspect that this will require a schema change and thus a full > re-index of the package tree. > > The current database table presented at www/packages stores: > > $software > $pkgname > $desc > $version > $vendorurl > $email > $maintlogin > $repourl > > The things that I'm missing are $vendorurl and $repourl. ?Is it > possible to extend pkgdb such that this is collected? ?It would save > having to fetch each updated package, unroll it and extract the info > from it directly. ?As the pkgdb rest calls will already be expensive, > I'd like to avoid this additional overhead if possible. > > What do you think? Done, vendor url and repository url are now in the JSON data structure. See the commit for details. http://sourceforge.net/apps/trac/gar/changeset/15341 Maciej From maciej at opencsw.org Mon Aug 15 10:02:34 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 15 Aug 2011 09:02:34 +0100 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: References: Message-ID: 2011/8/14 Maciej Blizi?ski : > ld: fatal: file liblto_plugin.so.0: open failed: No such file or directory > ld: fatal: File processing errors. No output written to > .libs/liblto_plugin.so.0.0.0 > > Does anyone have an idea where to look for diagnostic information? If anyone wants reproduce the problem, all code is committed and available in pkg/gcc4/branches/bootstrap-4.6. From pfelecan at opencsw.org Mon Aug 15 17:48:46 2011 From: pfelecan at opencsw.org (Peter FELECAN) Date: Mon, 15 Aug 2011 17:48:46 +0200 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: ("Maciej =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Mon, 15 Aug 2011 09:02:34 +0100") References: Message-ID: Maciej Blizi?ski writes: > 2011/8/14 Maciej Blizi?ski : >> ld: fatal: file liblto_plugin.so.0: open failed: No such file or directory >> ld: fatal: File processing errors. No output written to >> .libs/liblto_plugin.so.0.0.0 >> >> Does anyone have an idea where to look for diagnostic information? > > If anyone wants reproduce the problem, all code is committed and > available in pkg/gcc4/branches/bootstrap-4.6. Are you trying to build with Sun's C compiler? (I never succeeded to build gcc with that compiler) In the affirmative you'll have a heavy time generating the Ada compiler. -- Peter From maciej at opencsw.org Mon Aug 15 17:59:35 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 15 Aug 2011 16:59:35 +0100 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: References: Message-ID: 2011/8/15 Peter FELECAN : > Maciej Blizi?ski writes: > >> 2011/8/14 Maciej Blizi?ski : >>> ld: fatal: file liblto_plugin.so.0: open failed: No such file or directory >>> ld: fatal: File processing errors. No output written to >>> .libs/liblto_plugin.so.0.0.0 >>> >>> Does anyone have an idea where to look for diagnostic information? >> >> If anyone wants reproduce the problem, all code is committed and >> available in pkg/gcc4/branches/bootstrap-4.6. > > Are you trying to build with Sun's C compiler? (I never succeeded to build > gcc with that compiler) In the affirmative you'll have a heavy time > generating the Ada compiler. Yes, I am trying to get the stage 1 gcc built with Sun Studio. I've found a proof of concept how to work around the libtool problem above, but it's not the proper fix yet. Dago got the 4.3.6 version to build. Built packages generate a ton of checkpkg errors, such as sparcv8+ binaries and file conflicts. It will take a while to clean all that up. In general, I find the existing gcc recipe very complex and I'll see if I can simplify it. I have a question: What is the rationale behind having the gcc package split off to so many subpackages? I understand splitting off packages with the shared libraries, this is required to make the catalog maintainable. But all the specific compilers? It can't be the disk space argument, because if you have a development machine, you need to have a lot of disk space anyway. What is it then? If there is no really good reason, I would be inclined to switching gcc to a single package + shared libraries -- this way its maintenance will be much easier. Maciej From bwalton at opencsw.org Mon Aug 15 18:01:21 2011 From: bwalton at opencsw.org (Ben Walton) Date: Mon, 15 Aug 2011 12:01:21 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1313366514-sup-1442@pinkfloyd.chass.utoronto.ca> Message-ID: <1313423990-sup-1384@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Mon Aug 15 03:23:02 -0400 2011: > Done, vendor url and repository url are now in the JSON data > structure. See the commit for details. > > http://sourceforge.net/apps/trac/gar/changeset/15341 Great! So this is already stored, it just needed to be spit out in the json structure then? That's how I read the changeset. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Mon Aug 15 18:07:32 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 15 Aug 2011 17:07:32 +0100 Subject: [csw-maintainers] web visibility of packages In-Reply-To: <1313423990-sup-1384@pinkfloyd.chass.utoronto.ca> References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1313366514-sup-1442@pinkfloyd.chass.utoronto.ca> <1313423990-sup-1384@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/15 Ben Walton : > Excerpts from Maciej Blizi?ski's message of Mon Aug 15 03:23:02 -0400 2011: > >> Done, vendor url and repository url are now in the JSON data >> structure. See the commit for details. >> >> http://sourceforge.net/apps/trac/gar/changeset/15341 > > Great! ?So this is already stored, it just needed to be spit out in > the json structure then? ?That's how I read the changeset. Yes. There are 2 places where the metadata are stored. One is the pickled Python data structure. This is everything that has been collected about the package. It is what's used during package checks. The second place is the mysql schema and tables. A subset of metadata is put into mysql tables during something I call package "registration". It essentially means "please mark this package as one that can be added to catalogs". The gain from having these information in the tables is that you can use SQL power to query it. To access the pickled data, you have to unpickle, and unpickling the whole catalog takes about 8 minutes, while making a table join takes a fraction of a second. In this case, we're only requesting information about one package, so it's possible to unpickle it and reach all the data we want. Maciej From pfelecan at opencsw.org Mon Aug 15 20:14:39 2011 From: pfelecan at opencsw.org (Peter FELECAN) Date: Mon, 15 Aug 2011 20:14:39 +0200 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: ("Maciej =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Mon, 15 Aug 2011 16:59:35 +0100") References: Message-ID: Maciej Blizi?ski writes: > 2011/8/15 Peter FELECAN : >> Maciej Blizi?ski writes: >> >>> 2011/8/14 Maciej Blizi?ski : >>>> ld: fatal: file liblto_plugin.so.0: open failed: No such file or directory >>>> ld: fatal: File processing errors. No output written to >>>> .libs/liblto_plugin.so.0.0.0 >>>> >>>> Does anyone have an idea where to look for diagnostic information? >>> >>> If anyone wants reproduce the problem, all code is committed and >>> available in pkg/gcc4/branches/bootstrap-4.6. >> >> Are you trying to build with Sun's C compiler? (I never succeeded to build >> gcc with that compiler) In the affirmative you'll have a heavy time >> generating the Ada compiler. > > Yes, I am trying to get the stage 1 gcc built with Sun Studio. I've > found a proof of concept how to work around the libtool problem above, > but it's not the proper fix yet. > > Dago got the 4.3.6 version to build. Built packages generate a ton of > checkpkg errors, such as sparcv8+ binaries and file conflicts. It > will take a while to clean all that up. And how did you generate the Ada compiler? > In general, I find the existing gcc recipe very complex and I'll see > if I can simplify it. Complex package, complex recipe. > I have a question: What is the rationale behind having the gcc package > split off to so many subpackages? I understand splitting off packages > with the shared libraries, this is required to make the catalog > maintainable. But all the specific compilers? It can't be the disk > space argument, because if you have a development machine, you need to > have a lot of disk space anyway. What is it then? If there is no > really good reason, I would be inclined to switching gcc to a single > package + shared libraries -- this way its maintenance will be much > easier. non exhaustive reasons for which the packaging is done that way: 1. no need to install all the compilers when one needs only 1 language 2. traditional packaging on all the platforms that I know is done this way 3. there are common components which are not libraries The disk space is not an argument in our case but it is in other cases. Unfortunately you'll end up with more packages than the current ones and in my opinion that it's not a bad thing. Good granularity is a quality factor for our project. -- Peter From jesse at opencsw.org Tue Aug 16 00:31:09 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Tue, 16 Aug 2011 08:01:09 +0930 Subject: [csw-maintainers] nginx + passenger Message-ID: Hi Andy (and maintainers) I'm a newbee maintainer so apologies in advance for the dumb questions. I see you maintain nginx - yay. I use nginx with passenger for semi-high availability ruby on rails hosting (doing both front end load balancing / failover and the nginx-passenger-ruby instances at the app tier). What are your thoughts about adding the passenger module to the nginx package? My first thought is that this creates too many dependencies, ie ruby, but perhaps it can be only an optional runtime dependency? I'm not sure what nginx would do if it's compiled with passenger support and you remove ruby (and the passenger ruby gem). So, failing that, would it make sense to have a separate package named nginx+passenger or something? Or is there no room for such things with the opencsw naming convention? Or do you think this falls in to the basked of 'sorry opencsw can't help you'? Cheers Jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From bwalton at opencsw.org Tue Aug 16 00:35:14 2011 From: bwalton at opencsw.org (Ben Walton) Date: Mon, 15 Aug 2011 18:35:14 -0400 Subject: [csw-maintainers] nginx + passenger In-Reply-To: References: Message-ID: <89940428-62da-4c26-adb6-9ac952f100ae@email.android.com> Hi Jesse, A short reply as I'm in flight...multiple nginx packages with different options could be facilitated via alternatives. This would require some coordination but is possible. Thanks -Ben Jesse Reynolds wrote: Hi Andy (and maintainers) I'm a newbee maintainer so apologies in advance for the dumb questions. I see you maintain nginx - yay. I use nginx with passenger for semi-high availability ruby on rails hosting (doing both front end load balancing / failover and the nginx-passenger-ruby instances at the app tier). What are your thoughts about adding the passenger module to the nginx package? My first thought is that this creates too many dependencies, ie ruby, but perhaps it can be only an optional runtime dependency? I'm not sure what nginx would do if it's compiled with passenger support and you remove ruby (and the passenger ruby gem). So, failing that, would it make sense to have a separate package named nginx+passenger or something? Or is there no room for such things with the opencsw naming convention? Or do you think this falls in to the basked of 'sorry opencsw can't help you'? Cheers Jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciej at opencsw.org Tue Aug 16 00:45:12 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 15 Aug 2011 23:45:12 +0100 Subject: [csw-maintainers] nginx + passenger In-Reply-To: References: Message-ID: 2011/8/15 Jesse Reynolds : > What are your thoughts about adding the passenger module to the nginx > package? My first thought is that this creates too many dependencies, ie > ruby, but perhaps it can be only an optional runtime dependency? What does adding this module mean in the terms of files on disk, and package building? Is creating a new package sufficient, or is there a need to modify the nginx package? If so, what kind of modification is required? Maciej From jesse at opencsw.org Tue Aug 16 01:34:43 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Tue, 16 Aug 2011 09:04:43 +0930 Subject: [csw-maintainers] nginx + passenger In-Reply-To: References: Message-ID: 2011/8/16 Maciej Blizi?ski > 2011/8/15 Jesse Reynolds : > > What are your thoughts about adding the passenger module to the nginx > > package? My first thought is that this creates too many dependencies, ie > > ruby, but perhaps it can be only an optional runtime dependency? > > What does adding this module mean in the terms of files on disk, and > package building? Is creating a new package sufficient, or is there a > need to modify the nginx package? If so, what kind of modification is > required? > > I'd need to test this to be 100%, but I believe the nginx executable will be different with passenger support. It has no plugin architecture - you have to recompile nginx to add in it's modules. Andy already includes a bunch of modules with configure lines like --with-http_dav_module and so on. The only binary executable in the (sparc) package is /opt/csw/sbin/sparcv8plus/nginx. -------------- next part -------------- An HTML attachment was scrubbed... URL: From bwalton at opencsw.org Tue Aug 16 03:18:29 2011 From: bwalton at opencsw.org (Ben Walton) Date: Mon, 15 Aug 2011 21:18:29 -0400 Subject: [csw-maintainers] [offtopic] git presentation Message-ID: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> Hi All, I'm going to be putting together a git presentation for a mostly technical audience. I expect that most of these folks will not have used git although a few will have. I'm curious what those of you who use git would present. What features do you: find most useful? find hardest to understand? wish you'd learned earlier? find most dangerous? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From jesse at opencsw.org Tue Aug 16 04:40:46 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Tue, 16 Aug 2011 12:10:46 +0930 Subject: [csw-maintainers] moving along... In-Reply-To: References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/7 Maciej Blizi?ski > 2011/8/6 Peter FELECAN : > > Maciej Blizi?ski writes: > > > >> Mantis is not written for > >> the number of projects that we have. It is meant to handle a two digit > >> number of projects, and we have a four digit one > > > > Debian is using a BTS for a bigger magnitude, isn't it? What about using > > their system? Just a thought... > > Debian BTS / Debbugs has indeed work with a large number of packages, > and the sources are available[1]. However, if you the summary on > Wikipedia, it it's an email based system, and although you can search > and view bugs from a web interface, the main input and output is > email. What I like about it, is that there is the reportbug utility > which drives the bug reporting process. Overall, it feels dated. > > Another system that is in use, is Bugzilla. It's used by e. g. Red > Hat. It is a web based system, and if it has an email input, it's > only an addon. > > In the case of a packaging project, the main consideration is this: > the user who reports the issue, has a rough idea against which package > the issue needs to be filed. It still can be fuzzy in the case of > split packages, e.g. cups vs cupsd vs cupsclient, but let's assume > that the user knows which package the bugs belongs to. What mantis > does for us, is that it shows the list of packages to the user, and > knows which maintainer is assigned to which package. This is a great > feature. I talked once to a Gentoo developer and their bugzilla > installation. I learned that their bugzilla does not have that > mapping, and a human is necessary for the bug to be routed to the > right person. In this light, Mantis does a pretty good job for us, > and if we look for a new system, we need to make sure that it has an > easy user interface to map from packages to maintainers. That also > requires having some form of abstraction that we can use to represent > packages. In Mantis it's projects, but in a different bug tracker it > could be a different abstraction, e.g. a component or simply a tag. > > I'm with Ben on that I'd rather look for a way to restore the > catalog?mantis synchronization, the cheaper the better, and go from > there. I don't want discourage anyone from looking for a better > bugtracker - if we had a better bugtracker, I'm all for it! There > already is a wiki page about the topic[3], so if anyone makes > progress, I'd encourage them to share it. > > I see that progress has been made on moving to Jira. Sounds good to me. Jira has the idea of a 'project leader' that can be the default assignee per project (or you can configure someone else as the default assignee, or an abstract role user). The wiki page was last updated in September 2010 though, is this still 'the plan'? Given the plan looks to be to switch to Jira, with Crowd as a single sign-on solution, might as well grab Fisheye as well for browsing code and linking to Jira issues. Atlassian do their community licence for Fisheye as well. -------------- next part -------------- An HTML attachment was scrubbed... URL: From rupert at opencsw.org Tue Aug 16 06:36:10 2011 From: rupert at opencsw.org (rupert THURNER) Date: Tue, 16 Aug 2011 06:36:10 +0200 Subject: [csw-maintainers] moving along... In-Reply-To: References: <1312245431-sup-5389@pinkfloyd.chass.utoronto.ca> Message-ID: On Tue, Aug 16, 2011 at 04:40, Jesse Reynolds wrote: > > > 2011/8/7 Maciej Blizi?ski > >> 2011/8/6 Peter FELECAN : >> > Maciej Blizi?ski writes: >> > >> >> Mantis is not written for >> >> the number of projects that we have. It is meant to handle a two digit >> >> number of projects, and we have a four digit one >> > >> > Debian is using a BTS for a bigger magnitude, isn't it? What about using >> > their system? Just a thought... >> >> Debian BTS / Debbugs has indeed work with a large number of packages, >> and the sources are available[1]. However, if you the summary on >> Wikipedia, it it's an email based system, and although you can search >> and view bugs from a web interface, the main input and output is >> email. What I like about it, is that there is the reportbug utility >> which drives the bug reporting process. Overall, it feels dated. >> >> Another system that is in use, is Bugzilla. It's used by e. g. Red >> Hat. It is a web based system, and if it has an email input, it's >> only an addon. >> >> In the case of a packaging project, the main consideration is this: >> the user who reports the issue, has a rough idea against which package >> the issue needs to be filed. It still can be fuzzy in the case of >> split packages, e.g. cups vs cupsd vs cupsclient, but let's assume >> that the user knows which package the bugs belongs to. What mantis >> does for us, is that it shows the list of packages to the user, and >> knows which maintainer is assigned to which package. This is a great >> feature. I talked once to a Gentoo developer and their bugzilla >> installation. I learned that their bugzilla does not have that >> mapping, and a human is necessary for the bug to be routed to the >> right person. In this light, Mantis does a pretty good job for us, >> and if we look for a new system, we need to make sure that it has an >> easy user interface to map from packages to maintainers. That also >> requires having some form of abstraction that we can use to represent >> packages. In Mantis it's projects, but in a different bug tracker it >> could be a different abstraction, e.g. a component or simply a tag. >> >> I'm with Ben on that I'd rather look for a way to restore the >> catalog?mantis synchronization, the cheaper the better, and go from >> there. I don't want discourage anyone from looking for a better >> bugtracker - if we had a better bugtracker, I'm all for it! There >> already is a wiki page about the topic[3], so if anyone makes >> progress, I'd encourage them to share it. >> >> > I see that progress has been made on moving to Jira. Sounds good to me. > Jira has the idea of a 'project leader' that can be the default assignee per > project (or you can configure someone else as the default assignee, or an > abstract role user). The wiki page was last updated in September 2010 > though, is this still 'the plan'? > > Given the plan looks to be to switch to Jira, with Crowd as a single > sign-on solution, might as well grab Fisheye as well for browsing code and > linking to Jira issues. Atlassian do their community licence for Fisheye as > well. > if we really want to go commercial as with jira, then we might have a look at youtrack: * http://www.jetbrains.com/youtrack/ * http://youtrack.jetbrains.com/dashboard (300'000+ issues) * http://confluence.jetbrains.net/display/YTD3/Import+from+Mantis * http://www.jetbrains.com/youtrack/documentation/linux_installation.html both, atlassian, and jetbrains offer similar licenses, i.e. free usage for opensource projects. while i like jira lot and consider it an excellent product, youtrack was late enough to get a slim and simple user interface paired with the same basic feature set than jira. rupert. -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciej at opencsw.org Tue Aug 16 09:54:03 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 16 Aug 2011 08:54:03 +0100 Subject: [csw-maintainers] nginx + passenger In-Reply-To: References: Message-ID: 2011/8/16 Jesse Reynolds : > It has no plugin architecture - you have to recompile nginx to add in it's > modules. Andy already includes a bunch of modules with configure lines like > ?--with-http_dav_module and so on. If that's the case, you could (as you said) see if you can recompile nginx with passenger support, and separate out the files responsible for handling that support, if that's possible. You would just need to make sure that nging would still work with these files missing (i.e. the nginx-passenger package not installed). Maciej From jesse at opencsw.org Tue Aug 16 10:38:20 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Tue, 16 Aug 2011 18:08:20 +0930 Subject: [csw-maintainers] nginx + passenger In-Reply-To: References: Message-ID: 2011/8/16 Maciej Blizi?ski > 2011/8/16 Jesse Reynolds : > > It has no plugin architecture - you have to recompile nginx to add in > it's > > modules. Andy already includes a bunch of modules with configure lines > like > > --with-http_dav_module and so on. > > If that's the case, you could (as you said) see if you can recompile > nginx with passenger support, and separate out the files responsible > for handling that support, if that's possible. You would just need to > make sure that nging would still work with these files missing (i.e. > the nginx-passenger package not installed). > > Yeah, as we discussed on irc just now, that won't work because when you compile modules into nginx they are literally compiled into the one monolithic binary executable. It doesn't link to a shared object or anything like that. So I think it's a case of either: a) creating an nginx_passenger package, which incompatables the existing nginx package or b) adding passenger support to the existing nginx package, with a soft dependency on ruby and the passenger gem (ie you can use passenger configs with passenger if you've also installed ruby and the passenger gem) In any case, I'll try to create a package for the passenger gem as this will be required for either option. What's best? -------------- next part -------------- An HTML attachment was scrubbed... URL: From dam at opencsw.org Tue Aug 16 10:49:27 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Tue, 16 Aug 2011 10:49:27 +0200 Subject: [csw-maintainers] nginx + passenger In-Reply-To: References: Message-ID: <3FEBC3D7-7BCF-4EC2-BD00-FC1599BB8ECB@opencsw.org> Hi, Am 16.08.2011 um 10:38 schrieb Jesse Reynolds: > 2011/8/16 Maciej Blizi?ski > 2011/8/16 Jesse Reynolds : > > It has no plugin architecture - you have to recompile nginx to add in it's > > modules. Andy already includes a bunch of modules with configure lines like > > --with-http_dav_module and so on. > > If that's the case, you could (as you said) see if you can recompile > nginx with passenger support, and separate out the files responsible > for handling that support, if that's possible. You would just need to > make sure that nging would still work with these files missing (i.e. > the nginx-passenger package not installed). > > > Yeah, as we discussed on irc just now, that won't work because when you compile modules into nginx they are literally compiled into the one monolithic binary executable. It doesn't link to a shared object or anything like that. So I think it's a case of either: > a) creating an nginx_passenger package, which incompatables the existing nginx package or > b) adding passenger support to the existing nginx package, with a soft dependency on ruby and the passenger gem (ie you can use passenger configs with passenger if you've also installed ruby and the passenger gem) > > In any case, I'll try to create a package for the passenger gem as this will be required for either option. I suggest adding alternatives with nginx as base package and nginx_passenger as add-on which takes precedence on the binary and depends on ruby etc. Best regards -- Dago -- "You don't become great by trying to be great, you become great by wanting to do something, and then doing it so hard that you become great in the process." - xkcd #896 From skayser at opencsw.org Tue Aug 16 11:18:08 2011 From: skayser at opencsw.org (Sebastian Kayser) Date: Tue, 16 Aug 2011 11:18:08 +0200 Subject: [csw-maintainers] [offtopic] git presentation In-Reply-To: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> References: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> Message-ID: <20110816091808.GA21790@sebastiankayser.de> * Ben Walton wrote: > I'm going to be putting together a git presentation for a mostly > technical audience. I expect that most of these folks will not have > used git although a few will have. I'm curious what those of you who > use git would present. What features do you: Disclaimer: entry-level git user > find most useful? * easy to maintain, local repository (no server) * local commits (even when working with a server) * git stash * git commit --amend I'd probably also add github's super-easy-to-use fork + pull request workflow. > find hardest to understand? * when getting started: git's notion of an index/stage * detached HEAD Sebastian From dam at opencsw.org Tue Aug 16 11:41:39 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Tue, 16 Aug 2011 11:41:39 +0200 Subject: [csw-maintainers] [offtopic] git presentation In-Reply-To: <20110816091808.GA21790@sebastiankayser.de> References: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> <20110816091808.GA21790@sebastiankayser.de> Message-ID: <124C1EB7-14E3-4E82-B1ED-CEAA4F274FE8@opencsw.org> Hi, Am 16.08.2011 um 11:18 schrieb Sebastian Kayser: > * Ben Walton wrote: >> I'm going to be putting together a git presentation for a mostly >> technical audience. I expect that most of these folks will not have >> used git although a few will have. I'm curious what those of you who >> use git would present. What features do you: > > Disclaimer: entry-level git user > >> find most useful? > > * easy to maintain, local repository (no server) > * local commits (even when working with a server) > * git stash > * git commit --amend > > I'd probably also add github's super-easy-to-use fork + pull request > workflow. * branching/mering in 1 second * keeping the original author of a patch (as opposed to the committer) >> find hardest to understand? > > * when getting started: git's notion of an index/stage If that is the commit-cache: +1 > * detached HEAD * rebasing >> wish you'd learned earlier? * proper working with remotes >> find most dangerous? * rebasing Best regards -- Dago -- "You don't become great by trying to be great, you become great by wanting to do something, and then doing it so hard that you become great in the process." - xkcd #896 From maciej.blizinski at gmail.com Thu Aug 11 13:09:57 2011 From: maciej.blizinski at gmail.com (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Thu, 11 Aug 2011 12:09:57 +0100 Subject: [csw-maintainers] interesting solaris resource In-Reply-To: <1313027599-sup-8123@pinkfloyd.chass.utoronto.ca> References: <1313027599-sup-8123@pinkfloyd.chass.utoronto.ca> Message-ID: Em 11/08/2011 02:59, "Ben Walton" escreveu: > > > Hi All, > > I met Thomas (Cc'd) on freenode today and he is the man behind > http://wesunsolve.net, which aims to fill a gap left in the Solaris > ecosystem after the Oracle takeover. He asked if we'd mind linking to > his site from ours and I think that's a fine idea as many of our users > would be happy to find this tool. (There is an update in the works > too, which can be seen at http://dev.wesunsolve.net.) > > Dago suggested creating a "Friends" section in the site where we link > to things like this. Any objections? No objections? Inconceivable! > Any suggestions on how to place > these links nicely? A link in a sidebar would be good. Maciej -------------- next part -------------- An HTML attachment was scrubbed... URL: From andrew.nicols at luns.net.uk Tue Aug 16 11:23:06 2011 From: andrew.nicols at luns.net.uk (Andrew Robert Nicols) Date: Tue, 16 Aug 2011 10:23:06 +0100 Subject: [csw-maintainers] [offtopic] git presentation In-Reply-To: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> References: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> Message-ID: Hi Ben, As a developer, I use the following git commands each and every day. These are in no particular order: git grep git checkout -b foo / git checkout / git checkout git add git add -pi git rebase git rebase -i HEAD~5 git merge git cherry-pick git log git log -p git log -p -1 git log -p -1 git reset HEAD~1 (to unstage commits when editing commits with git rebase -i HEAD~x) git clean -df (use with care) git commit --amend git commit git blame git fetch git push git status git tag -s (I always sign my tags) I use each and everyone one of these pretty much *every* day. I also have the following aliases which make life easy. These are listed in order of frequency of use: git st = git status git co = git checkout git br = git branch git wdiff = git diff --color-words git stdel = !git st | grep delete | awk '{print $3}' | xargs git rm And the following commands are also really useful from time to time: git bisect (to find when a bug began it's existence - very useful) git gc (improves the speed of busy repositories by compressing the repo, and removing dangling links) git map (Perl version: https://github.com/clarkema/git-map; bash version: https://github.com/icefox/git-map. Allows you to check the status of multiple git repos easily) git remote add git mv gitk --all (I never use it without --all) git show : git vimdiff (alias to !GIT_EXTERNAL_DIFF='gitvimdiff' git --no-pager diff "$@" where gitvimdiff is a bash script in my $PATH running exec vimdiff "$2" "$5") tig --all (I never use it without --all) git reflog I personally don't like git pull There are also tools like gerrit which are brilliant - you can see how we use it in the mahara project at https://reviews.mahara.org. We also have Jenkins (continuous integration) hooked in to run various tests which feed into the process and prevent submission of commits to the repo if there are errors. I use git in the following situations daily: * as a contributor on the moodle project (http://www.moodle.org) - an open source virtual learning environment written in php; * as a contributor on the mahara project (http://www.mahara.org) - an open source ePortfolio application written in php; * as a systems developer to manage our configuration management system (git/puppet/perl/template toolkit/other custom tools); * for storage of debian package sources for packages we write/build/maintain; and * for internal projects. And also frequently: * as a debian package maintainer for mahara; * for texinfo documents; * for latex documents; * ... I also have a custom .bashrc snippet which displays my current branch if I'm in a repository, and colours this according to status (red = dirty; yellow = on master and clean; green = on another branch and clean) which is really handy. You'll find it at ~nicols/.bashrc on login.opencsw.org if you're interested. Hope that this helps, Andrew On 16 August 2011 02:18, Ben Walton wrote: > > Hi All, > > I'm going to be putting together a git presentation for a mostly > technical audience. I expect that most of these folks will not have > used git although a few will have. I'm curious what those of you who > use git would present. What features do you: > > find most useful? > find hardest to understand? > wish you'd learned earlier? > find most dangerous? > > Thanks > -Ben > -- > Ben Walton > Systems Programmer - CHASS > University of Toronto > C:416.407.5610 | W:416.978.4302 > > _______________________________________________ > maintainers mailing list > maintainers at lists.opencsw.org > https://lists.opencsw.org/mailman/listinfo/maintainers > .:: This mailing list's archive is public. ::. > -- Systems Developer e: andrew.nicols at luns.net.uk im: a.nicols at jabber.lancs.ac.uk t: +44 (0)1524 5 10147 Lancaster University Network Services is a limited company registered in England and Wales. Registered number: 04311892. Registered office: University House, Lancaster University, Lancaster, LA1 4YW -------------- next part -------------- An HTML attachment was scrubbed... URL: From andrew.nicols at luns.net.uk Tue Aug 16 11:29:28 2011 From: andrew.nicols at luns.net.uk (Andrew Robert Nicols) Date: Tue, 16 Aug 2011 10:29:28 +0100 Subject: [csw-maintainers] [offtopic] git presentation In-Reply-To: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> References: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> Message-ID: On 16 August 2011 02:18, Ben Walton wrote: > I'm going to be putting together a git presentation for a mostly > technical audience. I expect that most of these folks will not have > used git although a few will have. I'm curious what those of you who > use git would present. What features do you: > And I forgot to actually reply to the rest of it ;). > find hardest to understand? > git rebase is probably the hardest to explain - unless people get the concept of modifying your history and how it differs from a merge. > wish you'd learned earlier? > git rebase -i HEAD~x; then edit a commit; then git reset HEAD~1; then git add -pi . the relevant bits to break the one commit into sensible logical comits. find most dangerous? > it's pretty difficult to actually lose commits with git - if all else fails, you have git reflog. That said: git reset --hard HEAD~1 is pretty dangerous if you don't mean to use it. Also: *Always* start by branching before doing anything like a git rebase, or a git merge, or a feature branch, or a bug fix. When you start using tools like gerrit, it's almost impossible to use without feature branches - and rightly so! Andrew -- Systems Developer e: andrew.nicols at luns.net.uk im: a.nicols at jabber.lancs.ac.uk t: +44 (0)1524 5 10147 Lancaster University Network Services is a limited company registered in England and Wales. Registered number: 04311892. Registered office: University House, Lancaster University, Lancaster, LA1 4YW -------------- next part -------------- An HTML attachment was scrubbed... URL: From rupert.thurner at gmail.com Tue Aug 16 08:27:05 2011 From: rupert.thurner at gmail.com (rupert THURNER) Date: Tue, 16 Aug 2011 08:27:05 +0200 Subject: [csw-maintainers] [offtopic] git presentation Message-ID: just some ideas beside the necessary workflow, and the storage with rep, stage, working copy: # clone, local clones for feature development # commit before integrate others work # branch, merge # change history, e.g. amend to commit, rebase # if people from large svn installations are there, submodules # use a client like gitg On Aug 16, 2011 3:18 AM, "Ben Walton" wrote: -------------- next part -------------- An HTML attachment was scrubbed... URL: From jesse at opencsw.org Tue Aug 16 13:57:54 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Tue, 16 Aug 2011 21:27:54 +0930 Subject: [csw-maintainers] nginx + passenger In-Reply-To: <3FEBC3D7-7BCF-4EC2-BD00-FC1599BB8ECB@opencsw.org> References: <3FEBC3D7-7BCF-4EC2-BD00-FC1599BB8ECB@opencsw.org> Message-ID: On 16 August 2011 18:19, Dagobert Michelsen wrote: > > I suggest adding alternatives with nginx as base package and > nginx_passenger as > add-on which takes precedence on the binary and depends on ruby etc. > > OK, so that would mean changing the nginx base package to replace the /opt/csw/sbin/sparcv8plus/nginx (and x86) binaries with symlinks etc, yes? -------------- next part -------------- An HTML attachment was scrubbed... URL: From dam at opencsw.org Tue Aug 16 14:00:20 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Tue, 16 Aug 2011 14:00:20 +0200 Subject: [csw-maintainers] nginx + passenger In-Reply-To: References: <3FEBC3D7-7BCF-4EC2-BD00-FC1599BB8ECB@opencsw.org> Message-ID: Hi Jesse, Am 16.08.2011 um 13:57 schrieb Jesse Reynolds: > On 16 August 2011 18:19, Dagobert Michelsen wrote: >> I suggest adding alternatives with nginx as base package and nginx_passenger as >> add-on which takes precedence on the binary and depends on ruby etc. > > OK, so that would mean changing the nginx base package to replace the /opt/csw/sbin/sparcv8plus/nginx (and x86) binaries with symlinks etc, yes? Yes, as described here for GAR: http://sourceforge.net/apps/trac/gar/wiki/Alternatives Best regards -- Dago -- "You don't become great by trying to be great, you become great by wanting to do something, and then doing it so hard that you become great in the process." - xkcd #896 From bwalton at opencsw.org Tue Aug 16 14:02:05 2011 From: bwalton at opencsw.org (Ben Walton) Date: Tue, 16 Aug 2011 08:02:05 -0400 Subject: [csw-maintainers] nginx + passenger In-Reply-To: References: <3FEBC3D7-7BCF-4EC2-BD00-FC1599BB8ECB@opencsw.org> Message-ID: <6fc15654-5326-4d0d-adef-f2b22b12010e@email.android.com> Yes, Andy would need to move his binary to nginx.std or something and add alternatives support to the package. Your package would then deliver nginx.passenger and have alternatives support but a higher priority. You've need to ensure that you only skip non-conflicting files though, so your package might contain only the binary? Jesse Reynolds wrote: On 16 August 2011 18:19, Dagobert Michelsen wrote: I suggest adding alternatives with nginx as base package and nginx_passenger as add-on which takes precedence on the binary and depends on ruby etc. OK, so that would mean changing the nginx base package to replace the /opt/csw/sbin/sparcv8plus/nginx (and x86) binaries with symlinks etc, yes? -------------- next part -------------- An HTML attachment was scrubbed... URL: From jcraig at opencsw.org Tue Aug 16 14:05:33 2011 From: jcraig at opencsw.org (Jonathan Craig) Date: Tue, 16 Aug 2011 08:05:33 -0400 Subject: [csw-maintainers] nginx + passenger In-Reply-To: <3FEBC3D7-7BCF-4EC2-BD00-FC1599BB8ECB@opencsw.org> References: <3FEBC3D7-7BCF-4EC2-BD00-FC1599BB8ECB@opencsw.org> Message-ID: On Tue, Aug 16, 2011 at 4:49 AM, Dagobert Michelsen wrote: > > I suggest adding alternatives with nginx as base package and nginx_passenger as > add-on which takes precedence on the binary and depends on ruby etc. > So we would have a package that provides function set 'A' as a base package and then create a package that provides function set 'A' + 'B' and uses alternatives to take precedence. The benefit being that the base package avoids a dependency on ruby? Would it be better to roll the 'a' + 'b' functionality into the base and create a shell package that appends the ruby dependency but doesn't actually make a binary change. The benefit of this would be just one actual binary to build and troubleshoot. From dam at opencsw.org Tue Aug 16 14:08:18 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Tue, 16 Aug 2011 14:08:18 +0200 Subject: [csw-maintainers] nginx + passenger In-Reply-To: References: <3FEBC3D7-7BCF-4EC2-BD00-FC1599BB8ECB@opencsw.org> Message-ID: <75AEAA44-EAB1-4FEC-A0FA-24BBBD1D901E@opencsw.org> Hi Jon, Am 16.08.2011 um 14:05 schrieb Jonathan Craig: > On Tue, Aug 16, 2011 at 4:49 AM, Dagobert Michelsen wrote: >> >> I suggest adding alternatives with nginx as base package and nginx_passenger as >> add-on which takes precedence on the binary and depends on ruby etc. > > So we would have a package that provides function set 'A' as a base > package and then create a package that provides function set 'A' + 'B' > and uses alternatives to take precedence. The benefit being that the > base package avoids a dependency on ruby? Yes. > Would it be better to roll > the 'a' + 'b' functionality into the base and create a shell package > that appends the ruby dependency but doesn't actually make a binary > change. That will probably not work as the binaries in the base package link to dependent libraries not installed. It would also complicate checking if dependencies are entered correctly. > The benefit of this would be just one actual binary to build > and troubleshoot. True, but for the above reason it usually doesn't work this way. Best regards -- Dago -- "You don't become great by trying to be great, you become great by wanting to do something, and then doing it so hard that you become great in the process." - xkcd #896 From jesse at opencsw.org Tue Aug 16 14:27:13 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Tue, 16 Aug 2011 21:57:13 +0930 Subject: [csw-maintainers] nginx + passenger In-Reply-To: <6fc15654-5326-4d0d-adef-f2b22b12010e@email.android.com> References: <3FEBC3D7-7BCF-4EC2-BD00-FC1599BB8ECB@opencsw.org> <6fc15654-5326-4d0d-adef-f2b22b12010e@email.android.com> Message-ID: On 16 August 2011 21:32, Ben Walton wrote: > ** Yes, Andy would need to move his binary to nginx.std or something and > add alternatives support to the package. Your package would then deliver > nginx.passenger and have alternatives support but a higher priority. You've > need to ensure that you only skip non-conflicting files though, so your > package might contain only the binary? > Yes I was thinking it'd only contain the binary. Have to work out how to do that :-) (and many other things!) -------------- next part -------------- An HTML attachment was scrubbed... URL: From jesse at opencsw.org Tue Aug 16 14:34:49 2011 From: jesse at opencsw.org (Jesse Reynolds) Date: Tue, 16 Aug 2011 22:04:49 +0930 Subject: [csw-maintainers] nginx + passenger In-Reply-To: <75AEAA44-EAB1-4FEC-A0FA-24BBBD1D901E@opencsw.org> References: <3FEBC3D7-7BCF-4EC2-BD00-FC1599BB8ECB@opencsw.org> <75AEAA44-EAB1-4FEC-A0FA-24BBBD1D901E@opencsw.org> Message-ID: On 16 August 2011 21:38, Dagobert Michelsen wrote: > Hi Jon, > > Am 16.08.2011 um 14:05 schrieb Jonathan Craig: > > On Tue, Aug 16, 2011 at 4:49 AM, Dagobert Michelsen > wrote: > >> > >> I suggest adding alternatives with nginx as base package and > nginx_passenger as > >> add-on which takes precedence on the binary and depends on ruby etc. > > > > So we would have a package that provides function set 'A' as a base > > package and then create a package that provides function set 'A' + 'B' > > and uses alternatives to take precedence. The benefit being that the > > base package avoids a dependency on ruby? > > Yes. > > > Would it be better to roll > > the 'a' + 'b' functionality into the base and create a shell package > > that appends the ruby dependency but doesn't actually make a binary > > change. > > That will probably not work as the binaries in the base package link > to dependent libraries not installed. It would also complicate > checking if dependencies are entered correctly. > I'd need to test this, but I think if ruby were not installed, and you ran up this nginx+passenger binary, it would run fine unless you put passenger config items into it and expected it to work. That is, I think all the passenger module provides to nginx is a means to spawn ruby VMs running the ruby side of the passenger stuff, and to subsequently hand requests to it's child ruby processes. nginx with passenger support does not link to any new libraries. > > > The benefit of this would be just one actual binary to build > > and troubleshoot. > > True, but for the above reason it usually doesn't work this way. > If it were to work this way in this case, is it a good idea to go down this path? -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciej at opencsw.org Tue Aug 16 18:16:53 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 16 Aug 2011 17:16:53 +0100 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: References: Message-ID: 2011/8/15 Peter FELECAN : > Maciej Blizi?ski writes: > >> 2011/8/15 Peter FELECAN : >>> Maciej Blizi?ski writes: >>> >>>> 2011/8/14 Maciej Blizi?ski : >>>>> ld: fatal: file liblto_plugin.so.0: open failed: No such file or directory >>>>> ld: fatal: File processing errors. No output written to >>>>> .libs/liblto_plugin.so.0.0.0 >>>>> >>>>> Does anyone have an idea where to look for diagnostic information? >>>> >>>> If anyone wants reproduce the problem, all code is committed and >>>> available in pkg/gcc4/branches/bootstrap-4.6. >>> >>> Are you trying to build with Sun's C compiler? (I never succeeded to build >>> gcc with that compiler) In the affirmative you'll have a heavy time >>> generating the Ada compiler. >> >> Yes, I am trying to get the stage 1 gcc built with Sun Studio. I've >> found a proof of concept how to work around the libtool problem above, >> but it's not the proper fix yet. >> >> Dago got the 4.3.6 version to build. ?Built packages generate a ton of >> checkpkg errors, such as sparcv8+ binaries and file conflicts. ?It >> will take a while to clean all that up. > > And how did you generate the Ada compiler? For 4.3.6, it built fine. The 4.6.1 version requires the PPL library, which I got to compile for sparc, but not for i386, so I did not attempt to build Ada for 4.6.1. When you were working on it, did you also got stuck on PPL? Meanwhile, I talked with Dago on IRC; Dago took a look at PPL and stamped out compilation errors in PPL. >> In general, I find the existing gcc recipe very complex and I'll see >> if I can simplify it. > > Complex package, complex recipe. > >> I have a question: What is the rationale behind having the gcc package >> split off to so many subpackages? I understand splitting off packages >> with the shared libraries, this is required to make the catalog >> maintainable. But all the specific compilers? It can't be the disk >> space argument, because if you have a development machine, you need to >> have a lot of disk space anyway. What is it then? If there is no >> really good reason, I would be inclined to switching gcc to a single >> package + shared libraries -- this way its maintenance will be much >> easier. > > non exhaustive reasons for which the packaging is done that way: > > 1. no need to install all the compilers when one needs only 1 language > > 2. traditional packaging on all the platforms that I know is done this > ? way > > 3. there are common components which are not libraries > > The disk space is not an argument in our case but it is in other cases. > > Unfortunately you'll end up with more packages than the current ones and > in my opinion that it's not a bad thing. Good granularity is a quality > factor for our project. Fair enough, I'll slice the package separating out language-dependent files into subpackages. The main difference in packaging between the previous and the new gcc packages will be that I'm putting the shared libraries into /opt/csw/lib and creating library-specific packages, conforming to our packaging policy. There is one issue, in which gcc3corert and gcc3g++rt try to deliver the same files, so we need to repackage them in a way that makes room for gcc4 shared libs. Maciej From maciej at opencsw.org Tue Aug 16 18:40:01 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 16 Aug 2011 17:40:01 +0100 Subject: [csw-maintainers] progress report In-Reply-To: <1311179300-sup-3323@pinkfloyd.chass.utoronto.ca> References: <1311126906-sup-2591@pinkfloyd.chass.utoronto.ca> <1311179300-sup-3323@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/7/20 Ben Walton : > These will be handled like any other orphaned packages. ?If somebody > wants an uprev or bug fix, they're open to takeover. Are there any packages that we especially care about? Looking at the list, there's a few that caught my eye: alternatives, binutils, common, netcat, tcpwrappers. If any of the junior project members are looking for some practice, feel free to look at these and write GAR recipes for them. You can always post your work in progress to devel@ for reviews and help. Maciej From maciej at opencsw.org Tue Aug 16 20:06:16 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 16 Aug 2011 19:06:16 +0100 Subject: [csw-maintainers] gar download from launchpad? In-Reply-To: References: Message-ID: 2011/7/18 rupert THURNER : > ==> Running setup.py ?in work/solaris9-i386/build-isa-i386/subvertpy-0.8.2 > Traceback (most recent call last): > ? File "./setup.py", line 254, in > ? ? (svn_includedirs, svn_libdirs, svn_link_flags, extra_libs) = > svn_build_data() > ? File "./setup.py", line 125, in svn_build_data > ? ? raise Exception("Subversion development files not found. " > Exception: Subversion development files not found. Please set SVN_PREFIX or > (SVN_LIBRARY_PATH and SVN_HEADER_PATH) environment variable. Have you tried what the error message suggests? From maciej at opencsw.org Tue Aug 16 20:26:59 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 16 Aug 2011 19:26:59 +0100 Subject: [csw-maintainers] [csw-pkgsubmissions] newpkgs puppet, puppetmaster In-Reply-To: <7B579C1E-8AB7-40B4-847A-8BFD252F12F7@opencsw.org> References: <201105171549.p4HFn6NP024022@login.bo.opencsw.org> <1305649802-sup-9503@pinkfloyd.chass.utoronto.ca> <1305650813-sup-2188@pinkfloyd.chass.utoronto.ca> <2634A9E0-740B-4C32-B6BE-6DF37E168392@opencsw.org> <7B579C1E-8AB7-40B4-847A-8BFD252F12F7@opencsw.org> Message-ID: 2011/5/17 Mark Phillips : > Crikey, you go out for the evening and it all goes mad. > > Sorry, just catching up with this. Did I do something wrong? :) You broked the catalog generation! ;-) In other words, you helped me found a bug in the catalog generation, but instead of fixing it, I rolled back your puppet packages. The problem was that you're not supposed to rename pkg files, and there was no check for that. I will implement a check that makes csw-upload-pkg spit packages out if they have been renamed. Maciej From maciej at opencsw.org Tue Aug 16 20:36:00 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 16 Aug 2011 19:36:00 +0100 Subject: [csw-maintainers] [board] [Mantis] Account registration In-Reply-To: References: <1309884703-sup-1267@pinkfloyd.chass.utoronto.ca> <1309913219-sup-4680@pinkfloyd.chass.utoronto.ca> <1309977657-sup-28@pinkfloyd.chass.utoronto.ca> <1309979186-sup-9810@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/7/8 Asif Iqbal : > I noticed opencsw.org has the hobbit pkg 4.2 which is from 2006. > Since I heavily depend on csw (started with blastwave) I was looking > for a new hobbit pkg. > > So I volunteered to try to make xymon pkg and maintain. I have built > hobbit pkg for our company but that is not easy to maintain. I used the sun > doc to build the pkg and it works fine. > > However I rather join a mature community like this one and maintain > it in csw repo using the method designed to follow. It will definitely > benefit us. Great to see you here, Asif! How is the work on xymon going? I see a stub of the xymon recipe in the repository, is there anything we can help with? Maciej From rupert at opencsw.org Wed Aug 17 01:42:13 2011 From: rupert at opencsw.org (rupert THURNER) Date: Wed, 17 Aug 2011 01:42:13 +0200 Subject: [csw-maintainers] [offtopic] git presentation In-Reply-To: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> References: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> Message-ID: On Tue, Aug 16, 2011 at 03:18, Ben Walton wrote: > > Hi All, > > I'm going to be putting together a git presentation for a mostly > technical audience. I expect that most of these folks will not have > used git although a few will have. I'm curious what those of you who > use git would present. What features do you: > > find most useful? > find hardest to understand? > wish you'd learned earlier? > find most dangerous? > besides the things already mentioned by others, and the obvious "first commit then get in others commits" workflow: # how to change history, ammend to commit, rebase # how to undo a / any command. # how to develop a feature (branch or clone or whatever) # if people from large svn installations are there, submodules # use a client like gitg rupert -------------- next part -------------- An HTML attachment was scrubbed... URL: From bwalton at opencsw.org Wed Aug 17 01:57:18 2011 From: bwalton at opencsw.org (Ben Walton) Date: Tue, 16 Aug 2011 19:57:18 -0400 Subject: [csw-maintainers] [offtopic] git presentation In-Reply-To: References: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> Message-ID: <1313538940-sup-7914@pinkfloyd.chass.utoronto.ca> Excerpts from rupert THURNER's message of Tue Aug 16 19:42:13 -0400 2011: Hi All, > besides the things already mentioned by others, and the obvious "first > commit then get in others commits" workflow: > # how to change history, ammend to commit, rebase > # how to undo a / any command. > # how to develop a feature (branch or clone or whatever) > # if people from large svn installations are there, submodules > # use a client like gitg Thanks for a great set of responses. It's nice to get a sense of how others perceive this. This will help me target the presentation such that I can hopefully give people a good introduction to git and maybe save a few toes from being stubbed. Much appreaciated! Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bwalton at opencsw.org Wed Aug 17 02:04:18 2011 From: bwalton at opencsw.org (Ben Walton) Date: Tue, 16 Aug 2011 20:04:18 -0400 Subject: [csw-maintainers] progress report In-Reply-To: References: <1311126906-sup-2591@pinkfloyd.chass.utoronto.ca> <1311179300-sup-3323@pinkfloyd.chass.utoronto.ca> Message-ID: <1313539411-sup-1117@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Tue Aug 16 12:40:01 -0400 2011: > Are there any packages that we especially care about? Looking at > the list, there's a few that caught my eye: alternatives, binutils, > common, netcat, tcpwrappers. If any of the junior project members > are looking for some practice, feel free to look at these and write > GAR recipes for them. You can always post your work in progress to > devel@ for reviews and help. The CSWcommon should be an especially easy build as it's just a bunch of mkdir commands. Alternatives likely isn't bad either. -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Wed Aug 17 09:39:46 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 17 Aug 2011 08:39:46 +0100 Subject: [csw-maintainers] [offtopic] git presentation In-Reply-To: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> References: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/16 Ben Walton : > What features do you: > > find most useful? 1. git stage -p 2. Rebasing. It took me a long time to get what it does and why is it useful. Also, history rewriting in general. In simpler terms: I find it especially useful to do this: - Work on feature1 and send it for code review (I don't commit it yet) - While feature1 is under review, work on feature2 which depends on feature1 After completing feature1, I can create a set of patches for feature2, while keeping the history linear. > find hardest to understand? git reset vs git reset --hard, how one doesn't modify the sources and the other does. I consider the words "reset" and "hard" misnomers. > wish you'd learned earlier? Git's internal data structures. > find most dangerous? Using git-svn. E.g. you read [1] and you think "cool! I will now use all of git features! I will be able to move code between hosts using git and then commit to subversion!" And then you try it, and you see the self-resurrecting code conflicts. After half a year, you arrive at [2]. [1] http://andy.delcambre.com/2008/03/04/git-svn-workflow.html [2] http://automatthias.wordpress.com/2011/03/28/a-git-svn-workflow/ From opk at opencsw.org Wed Aug 17 19:18:29 2011 From: opk at opencsw.org (Oliver Kiddle) Date: Wed, 17 Aug 2011 19:18:29 +0200 Subject: [csw-maintainers] [offtopic] git presentation In-Reply-To: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> References: <1313457377-sup-3754@pinkfloyd.chass.utoronto.ca> Message-ID: <6678.1313601509@thecus.kiddle.eu> Ben Walton wrote: > > Hi All, > > I'm going to be putting together a git presentation for a mostly > technical audience. I expect that most of these folks will not have > used git although a few will have. I'm curious what those of you who > use git would present. What features do you: > > find most useful? rebase in general, in particular, rebase -i and pull --rebase Having to add all changed files: it solves the old cvs/svn problem of always forgetting to add new files. bisect and bundle are also great because they are so simple but solve certain problems really well. > find hardest to understand? The options and subcommand naming: differences between fetch/pull/rebase pull --rebase. Along with the options to reset, though the following blog helped on that recently: http://progit.org/blog.html > wish you'd learned earlier? > find most dangerous? Like Maciej I fell foul of git-svn but once you understand the limitations, it's fine and actually one of the most useful things if you're stuck with svn for the master repository. Oliver From bwalton at opencsw.org Fri Aug 19 02:03:08 2011 From: bwalton at opencsw.org (Ben Walton) Date: Thu, 18 Aug 2011 20:03:08 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1313366514-sup-1442@pinkfloyd.chass.utoronto.ca> <1313423990-sup-1384@pinkfloyd.chass.utoronto.ca> Message-ID: <1313712018-sup-1054@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Mon Aug 15 12:07:32 -0400 2011: Hi Maciej, > Yes. There are 2 places where the metadata are stored. One is the > pickled Python data structure. This is everything that has been Cool. That gets me what I needed for that purpose. Now I need more! :) I need either the file list ($md5/files) or the pkgmap as a plain text item or json structure. I'm also going to need the dependencies. Is this easily doable? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bwalton at opencsw.org Fri Aug 19 02:15:09 2011 From: bwalton at opencsw.org (Ben Walton) Date: Thu, 18 Aug 2011 20:15:09 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: <1313712018-sup-1054@pinkfloyd.chass.utoronto.ca> References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1313366514-sup-1442@pinkfloyd.chass.utoronto.ca> <1313423990-sup-1384@pinkfloyd.chass.utoronto.ca> <1313712018-sup-1054@pinkfloyd.chass.utoronto.ca> Message-ID: <1313712734-sup-2029@pinkfloyd.chass.utoronto.ca> Excerpts from Ben Walton's message of Thu Aug 18 20:03:08 -0400 2011: > I need either the file list ($md5/files) or the pkgmap as a plain text > item or json structure. I'm also going to need the dependencies. Oh, and one other teeny, tiny, small little item...a list of all shared libraries that package contains? This is used to provide a reverse mapping of .so.X to packages that use it. Have a look at http://opencsw.org/packages/CSWlibxml2-2/ for an example. I'd be willing to drop this feature as knowing that the dependency exists is enough for my tastes. I don't need to see which packages link to which specific shared object. Others might find this useful though? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bwalton at opencsw.org Fri Aug 19 02:17:11 2011 From: bwalton at opencsw.org (Ben Walton) Date: Thu, 18 Aug 2011 20:17:11 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: <1313712734-sup-2029@pinkfloyd.chass.utoronto.ca> References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1313366514-sup-1442@pinkfloyd.chass.utoronto.ca> <1313423990-sup-1384@pinkfloyd.chass.utoronto.ca> <1313712018-sup-1054@pinkfloyd.chass.utoronto.ca> <1313712734-sup-2029@pinkfloyd.chass.utoronto.ca> Message-ID: <1313712945-sup-401@pinkfloyd.chass.utoronto.ca> Excerpts from Ben Walton's message of Thu Aug 18 20:15:09 -0400 2011: > Oh, and one other teeny, tiny, small little item...a list of all > shared libraries that package contains? Ooops...sorry, this should be "every shared object a package depends on by NEEDED items in the binary/library." This information would have been used by checkpkg so maybe it's in the pickled data structure for extraction? Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bwalton at opencsw.org Fri Aug 19 03:25:51 2011 From: bwalton at opencsw.org (Ben Walton) Date: Thu, 18 Aug 2011 21:25:51 -0400 Subject: [csw-maintainers] web visibility of packages In-Reply-To: <1313712018-sup-1054@pinkfloyd.chass.utoronto.ca> References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1313366514-sup-1442@pinkfloyd.chass.utoronto.ca> <1313423990-sup-1384@pinkfloyd.chass.utoronto.ca> <1313712018-sup-1054@pinkfloyd.chass.utoronto.ca> Message-ID: <1313717110-sup-3843@pinkfloyd.chass.utoronto.ca> Excerpts from Ben Walton's message of Thu Aug 18 20:03:08 -0400 2011: > I need either the file list ($md5/files) or the pkgmap as a plain text > item or json structure. I'm also going to need the dependencies. Actually, dependencies can be taken from the catalog that will be parsed to drive this. Just the pkgmap/file list... Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Fri Aug 19 09:19:39 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Fri, 19 Aug 2011 08:19:39 +0100 Subject: [csw-maintainers] web visibility of packages In-Reply-To: <1313712945-sup-401@pinkfloyd.chass.utoronto.ca> References: <1312668306-sup-2166@pinkfloyd.chass.utoronto.ca> <1313366514-sup-1442@pinkfloyd.chass.utoronto.ca> <1313423990-sup-1384@pinkfloyd.chass.utoronto.ca> <1313712018-sup-1054@pinkfloyd.chass.utoronto.ca> <1313712734-sup-2029@pinkfloyd.chass.utoronto.ca> <1313712945-sup-401@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/19 Ben Walton : > Excerpts from Ben Walton's message of Thu Aug 18 20:15:09 -0400 2011: > >> Oh, and one other teeny, tiny, small little item...a list of all >> shared libraries that package contains? > > Ooops...sorry, this should be "every shared object a package depends > on by NEEDED items in the binary/library." ?This information would > have been used by checkpkg so maybe it's in the pickled data structure > for extraction? I could even return the whole pickled data structure, and you could do whatever you wanted with it, it has all the information. There's one wrinkle, where I was lazy and I pickled a Python object instead of a plain tuple, so I'll have to modify the code to use the tuple instead of a timestamp object. It'll require bumping up the data structure version number, and reindexing of the catalog. Maciej From maciej at opencsw.org Fri Aug 19 19:29:34 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Fri, 19 Aug 2011 18:29:34 +0100 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: References: Message-ID: Progress report: - I repackaged gcc3 shared libraries, to allow migrating to the gcc4 versions of the same SONAMEs - Sparc packages are built (without ada) - i386 (Solaris 9) packages can be built - amd64 (Solaris 10) packages don't build, I'm currently investigating the issue - there has been no progress on the PPL library, so no Ada frontend for gcc-4.6.1 - if anyone wants to help, please check out the sources and observe how the build fails on amd64 If you want to see some evidence of my work (besides the code commits), you can peek at the experimental catalog to see some packages that have been already built (only for sparc right now). http://buildfarm.opencsw.org/experimental.html#gcc4 Also, it is possible to hit the fork limit on the buildfarm: fork: Resource temporarily unavailable Go me! Or should I slow down? Maciej From pfelecan at opencsw.org Sat Aug 20 07:59:03 2011 From: pfelecan at opencsw.org (Peter FELECAN) Date: Sat, 20 Aug 2011 07:59:03 +0200 Subject: [csw-maintainers] [POLICY] Policy-team, policy docs, licenses In-Reply-To: ("Maciej =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Tue, 19 Jul 2011 11:54:12 +0100") References: Message-ID: Maciej Blizi?ski writes: > Em 18/07/2011 14:02, escreveu: >> >> Maciej Blizi?ski writes: >> > Peter, would you like to resume your participation in the policy team? >> >> Of course I wish to. I think that we should start from where we left the >> discussion before it was polluted, i.e., the copyright and the >> abstract. Taking on the Debian policy document and adapting it to our >> specifics but not more. > > Excellent! I have since looked at sphinx, a higher level tool which helps > with structured documentation. It uses reStructured Text as the lightweight > markup language. I will send the patch and a link to an example when I'm > back home at the beginning of August. looked == switched ? -- Peter From maciej at opencsw.org Sun Aug 21 10:42:17 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sun, 21 Aug 2011 09:42:17 +0100 Subject: [csw-maintainers] csw-upload-pkg and _stub packages Message-ID: There was an issue with csw-upload-pkg that made csw-upload-pkg stop and not upload renamed packages. I've modified csw-upload-pkg behavior - during update, the backend removes not only matching catalognames, but also matching pkgnames. For example: Case 1: Catalog: CSWfoo/foo-1.0 Uploaded: CSWfoo/foo-1.1 Backend removes existing CSWfoo/foo-1.0 and replaces it with CSWfoo/foo-1.1. Case 2: Catalog: CSWfoo/foo-1.0 Uploaded: CSWbar/foo-1.0 Backend removes existing CSWbar/foo-1.0 and replaces it with CSWfoo/foo-1.1. Case 3: Catalog: CSWfoo/foo-1.0 Uploaded: CSWfoo/foo_stub-1.0 The backend used to say: "There already is a package with pkgname CSWfoo" and stop. It now automatically removes CSWfoo/CSWfoo-1.0 and replaces it with CSWfoo/foo_stub-1.1. Package renames will be easier now. But remember, more magic means more ways to shoot yourself in the foot! Maciej [1] http://lists.opencsw.org/pipermail/maintainers/2011-July/015023.html From maciej at opencsw.org Sun Aug 21 10:49:42 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sun, 21 Aug 2011 09:49:42 +0100 Subject: [csw-maintainers] csw-upload-pkg and _stub packages In-Reply-To: References: Message-ID: 2011/8/21 Maciej Blizi?ski : > Case 2: > > Catalog: CSWfoo/foo-1.0 > Uploaded: CSWbar/foo-1.0 > > Backend removes existing CSWbar/foo-1.0 and replaces it with CSWfoo/foo-1.1. Errata: Backend removes existing CSWfoo/foo-1.0 and replaces it with CSWbar/foo-1.1. From maciej at opencsw.org Sun Aug 21 13:26:04 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sun, 21 Aug 2011 12:26:04 +0100 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: References: Message-ID: I've built all the packages, and they're available for testing: http://buildfarm.opencsw.org/experimental.html#gcc4 I tried to build ppl with gcc-4.6.1, and got this error: /opt/csw/include/gmpxx-32.h:35:18: fatal error: iosfwd: No such file or directory The iosfwd file is present in the filesystem: maciej at vsol05 ~/opencsw/ppl/trunk $ ls -l /opt/csw/gcc4/include/c++/*/iosfwd -rw-r--r-- 1 root bin 6917 Aug 21 04:39 /opt/csw/gcc4/include/c++/4.6.1/iosfwd Does anyone have an idea why g++ wouldn't find the C++ header file even though it's present on disk? Maciej From dam at opencsw.org Sun Aug 21 20:20:17 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Sun, 21 Aug 2011 20:20:17 +0200 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: References: Message-ID: Hi Maciej, Am 21.08.2011 um 13:26 schrieb Maciej Blizi?ski: > I've built all the packages, and they're available for testing: > > http://buildfarm.opencsw.org/experimental.html#gcc4 > > I tried to build ppl with gcc-4.6.1, and got this error: > > /opt/csw/include/gmpxx-32.h:35:18: fatal error: iosfwd: No such file > or directory > > The iosfwd file is present in the filesystem: > > maciej at vsol05 ~/opencsw/ppl/trunk $ ls -l /opt/csw/gcc4/include/c++/*/iosfwd > -rw-r--r-- 1 root bin 6917 Aug 21 04:39 /opt/csw/gcc4/include/c++/4.6.1/iosfwd > > Does anyone have an idea why g++ wouldn't find the C++ header file > even though it's present on disk? Where does it look for the file? You can try truss or opensnoop to find if gcc looks at the wrong location. Best regards -- Dago -- "You don't become great by trying to be great, you become great by wanting to do something, and then doing it so hard that you become great in the process." - xkcd #896 From maciej at opencsw.org Sun Aug 21 21:50:21 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sun, 21 Aug 2011 20:50:21 +0100 Subject: [csw-maintainers] GCC 4.6 progress In-Reply-To: References: Message-ID: 2011/8/21 Dagobert Michelsen : > Where does it look for the file? You can try truss or opensnoop to find > if gcc looks at the wrong location. Looking at the truss output, gcc tries to locate the header in wrong places: 22203: stat64("/opt/csw/lib/gcc/sparc-sun-solaris2.10/4.6.1/include/iosfwd.gch", 0xFFBFF038) Err#2 ENOENT 22203: open64("/opt/csw/lib/gcc/sparc-sun-solaris2.10/4.6.1/include/iosfwd", O_RDONLY|O_NOCTTY) Err#2 ENOENT 22203: stat64("/opt/csw/gcc4/include/iosfwd.gch", 0xFFBFF038) Err#2 ENOENT 22203: open64("/opt/csw/gcc4/include/iosfwd", O_RDONLY|O_NOCTTY) Err#2 ENOENT 22203: stat64("/opt/csw/lib/gcc/sparc-sun-solaris2.10/4.6.1/include-fixed/iosfwd.gch", 0xFFBFF038) Err#2 ENOENT 22203: open64("/opt/csw/lib/gcc/sparc-sun-solaris2.10/4.6.1/include-fixed/iosfwd", O_RDONLY|O_NOCTTY) Err#2 ENOENT 22203: stat64("/usr/include/iosfwd.gch", 0xFFBFF038) Err#2 ENOENT 22203: open64("/usr/include/iosfwd", O_RDONLY|O_NOCTTY) Err#2 ENOENT My theory is that gcc was confused by the explicit --includedir setting. From bwalton at opencsw.org Mon Aug 22 03:11:41 2011 From: bwalton at opencsw.org (Ben Walton) Date: Sun, 21 Aug 2011 21:11:41 -0400 Subject: [csw-maintainers] progress report Message-ID: <1313975389-sup-4471@pinkfloyd.chass.utoronto.ca> Hi All, A short check in on the web and mantis integration automation. I now have code that can take an old catalog and a new one and then generate a list of high level steps required to make our browseable web interface and mantis 'line up' with the catalog state. I've also trudged through the old ksh code and I think I have a list of sql statements required to turn high level action to low level implementation. Tomorrow I'm going to work on writing tests for the high level action generation code though as there are quite a few rarities and b-sides that can occur in there. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Mon Aug 22 11:02:35 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 22 Aug 2011 10:02:35 +0100 Subject: [csw-maintainers] Rremoving oldies / duplicates from experimental Message-ID: >From the "Maciej's random shell snippets" series: Let's suppose you're working on a project and putting files into /home/experimental/project. You build not just from one recipe, you build from two or three, and you stick all packages you built into your experimental directory. You will likely end up with packages from yesterday you want to remove, but you don't want to remove everything, because there are older packages from the other build recipe that you don't want to remove. Typically it's annoying, because you need to look for two files with the same catalogname, architecture and OS release, but older REV stamp. For example: libstdc++6-4.6.1,REV=2011.08.21-SunOS5.10-sparc-CSW.pkg.gz libstdc++6-4.6.1,REV=2011.08.20-SunOS5.10-sparc-CSW.pkg.gz libstdc++5-3.4.6,REV=2011.08.20-SunOS5.10-sparc-CSW.pkg.gz One of the above packages can be removed. Can you tell which one, in 1 second? I can't. But shell can. Here's a shell snippet that outputs rm commands you can execute to clean up your experimental directory. I broke it up into shorter lines to ensure you can copy and paste it. exp=/home/experimental/gcc4 ls ${exp}/ | gawk -F- '{print $1, $3, $4}' \ | sort | uniq -c \ | gegrep -v '^\s+1' \ | while read number catalogname osrel arch; do \ echo rm \ $(echo $exp/${catalogname}-*-${osrel}-${arch}* \ | gtr ' ' '\n' | ghead -n 1); done If anyone has a cleverer way to clean up experimental directories, hit me! Maciej From maciej at opencsw.org Mon Aug 22 15:29:40 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 22 Aug 2011 14:29:40 +0100 Subject: [csw-maintainers] binutils and the .a files Message-ID: Are the .a files necessary in the binutils package? In some cases, the .a files have to be present, for example libgcc.a and libgcc_eh.a in the gcc packages. The current binutils package contains e.g. libiberty.a: http://buildfarm.opencsw.org/pkgdb/srv4/27cd3802d6e4edf68837e29e708a3bb7/ Does anyone know if the .a files are used? Maciej From maciej at opencsw.org Tue Aug 23 12:09:19 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 23 Aug 2011 11:09:19 +0100 Subject: [csw-maintainers] Ongoing projects Message-ID: There is a number of ongoing efforts that haven't been finished yet. They vary in complexity and importance, but they can all be considered important, as they concern core packages. I'll list the projects here, and I'd like to ask people who know the status of the project (e.g. what's missing), to reply and describe what the projects are blocking on. Also, if the list is incomplete, feel free to add missing items. == krb5_lib == Contributors: Dagobert, Maciej Status: active Missing: The library split recipe is done, but the build links against an old version of libkdb5.so from /opt/csw/lib. There are some binaries depending on libkdb5.so.4, e.g. /opt/csw/sbin/kadmind, so if we aren't updating this binary just yet, we need to split libkdb5.so.4 into a separate package. == PHP 5 == Contributors: Ben Status: used to be active, but no activity in last weeks Missing: == Perl 5.12.4 == Works on the update to 5.12.4 are on the way. There are some failing tests. Dagobert is working on this. Contributors: Dagobert, El_Che (from #opencsw) Status: active Missing: == GCC 4.6.1 == Contributors: Maciej Status: active Missing: gcc-4.6.1 builds, except the Ada frontend. Ada needs the PPL library which doesn't currently compile. Help would be appreciated, as debugging PPL is separate from everything else. Also, g++ was broken, and I'm rebuilding it to see if my guess about what was broken, was right. == MySQL 5.1 == Contributors: Maciej, Dagobert (alternatives support for 5.0) Status: stale Missing: 5.1 itself builds. There are conflicting binaries with 5.0. The 5.0 recipe has been updated to use alternatives; the 5.1 recipe needs updating the same way. There are no other known blockers. == PostgreSQL 9.x == Contributors: Maciej Status: stale Missing: There's a lot of moving of shared libraries to be done. The binaries themselves compile fine, but there are problems with 32/64 bits, where data from 32-bit binaries are incomaptible with the 64-bit binaries. The main issue is how to package PostgreSQL so that multiple versions can be installed at the same time. The builds I worked on use /opt/csw (and not /opt/csw/postgresql). == ImageMagick 64-bit == Contributors: Roger Hakansson Status: stale? Missing: == Ruby 1.9 == Contributors: Ben Status: ? Missing: == Python 2.7, Python 3.1 == Contributors: Maciej Status: stale Missing: No fundamental problems; rebuilds needed, dep refreshment needed, testing needed. == Sendmail with largefile suport == Contributors: Peter B Status: ? Missing: --------------------- That's all I could write from the top of my head. All the work that has been done already, is committed to the repository. As a project, we'll be better off if we focus on one project at a time and get it finished, meaning, completed enough to be pushed out to the world. This way, we deliver to our users. Perhaps some of the projects are very close to completion, and just a little help could get them released. For example, Ben has been working on PHP, but he's now focusing on the release engine. It's clear that the release engine has more priority, but meanwhile, perhaps another person could step in and move the php project forwards. I'm sure Ben would be happy to provide more details on what's the state of the project and what's left to be done. Maciej From bonivart at opencsw.org Tue Aug 23 14:51:02 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Tue, 23 Aug 2011 14:51:02 +0200 Subject: [csw-maintainers] Ongoing projects In-Reply-To: References: Message-ID: 2011/8/23 Maciej Blizi?ski : > == Sendmail with largefile suport == > > Contributors: Peter B > Status: Ready for testing > Missing: Testing. It's been sitting in experimental since July 2nd and I've had no feedback on my latest version (Jake Goerzen responded about file permissions which I fixed). I will probably do some kind of test suite myself and if it passes my needs I will release it. ^^^ > As a project, we'll be better off if we focus on one project at a time > and get it finished, meaning, completed enough to be pushed out to the > world. ?This way, we deliver to our users. I've been pushing that view to several of you for a long time, glad to see someone else supporting it. It's obvious, both by your long list above and by the almost two month long interruption in package release, that we add tasks quicker than we complete them. Maybe we should have some cap on how many tasks we have going at once? Another problem is the size of the tasks and how they tend to grow. Perl is one example where the last step of 64 bit is holding everything up, we have open bugs and an obsolete version of Perl, I think we should have fixed those issues and then continued on with 64 bit after that. Divide the tasks into steps that the users can benefit from. If I'm correct the release process task was previously stuck on signing the catalogs, then on notifying someone when the signing process stopped and now it has sucked in the package browser on the web site and Mantis integration into its black hole. :) I'm wondering if this couldn't have been done in steps so users could have gotten updates by now? Was this discussed on the list or only between those involved? I'm fine with people actually doing the work having more say but by making stuff complex you also exclude people. /peter From bwalton at opencsw.org Tue Aug 23 15:01:09 2011 From: bwalton at opencsw.org (Ben Walton) Date: Tue, 23 Aug 2011 09:01:09 -0400 Subject: [csw-maintainers] Ongoing projects In-Reply-To: References: Message-ID: <1314104325-sup-7792@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Tue Aug 23 06:09:19 -0400 2011: > Contributors: Ben > Status: used to be active, but no activity in last weeks > Missing: It's basically ready to go. It needs to be rebuilt and have mysql deps modified to match the new mysql layout in unstable, but the recipe and build is sound afaik. If somebody wanted to push this forward, I'm happy to hand it off...I took it on as it really needed an update but I have little personal need for it. > Contributors: Ben > Status: ? > Missing: A few tests fail, but the recipe is in good shape. I haven't built it with the latest patch level, so it may be better (or worse) than it was. I'd like to get back to this one. As soon as I clear away the automation tasks, I'll dive into some of my pending package changes. I need to migrate apache config as well. :) Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From maciej at opencsw.org Wed Aug 24 10:59:25 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 24 Aug 2011 09:59:25 +0100 Subject: [csw-maintainers] Ongoing projects In-Reply-To: References: Message-ID: 2011/8/23 Peter Bonivart : >> As a project, we'll be better off if we focus on one project at a time >> and get it finished, meaning, completed enough to be pushed out to the >> world. ?This way, we deliver to our users. > > I've been pushing that view to several of you for a long time, glad to > see someone else supporting it. It's obvious, both by your long list > above and by the almost two month long interruption in package > release, that we add tasks quicker than we complete them. Maybe we > should have some cap on how many tasks we have going at once? Yes. There's one more thing to look at: who is contributing to projects. If we say that we work on 1 project, we need to make sure that all maintainers can contribute to that project. I imagine that Dagobert is able to work on pretty much any project you can think of. That doesn't have to be true for other people. So if we only focus on getting more people involved in the works, we would get a big boost, regardless to whether we cap the number of projects or not. > Another problem is the size of the tasks and how they tend to grow. > Perl is one example where the last step of 64 bit is holding > everything up, we have open bugs and an obsolete version of Perl, I > think we should have fixed those issues and then continued on with 64 > bit after that. Divide the tasks into steps that the users can benefit > from. Yes, and it's best if someone pays attention to that and suggests how to divide ongoing projects into parts. > If I'm correct the release process task was previously stuck on > signing the catalogs, then on notifying someone when the signing > process stopped and now it has sucked in the package browser on the > web site and Mantis integration into its black hole. :) I'm wondering > if this couldn't have been done in steps so users could have gotten > updates by now? Was this discussed on the list or only between those > involved? It's blocking on generating the new key and deploying it on cswsign. I'm talking to Ihsan about it, I think he has the key ready, and we need to escrow it, the way we decided in a vote some time ago. > I'm fine with people actually doing the work having more say > but by making stuff complex you also exclude people. I agree that keeping things simple is beneficial. However, there are cases where it's impossible to keep things simple and be able to do advanced tasks at the same time. Usually, we try to decouple functionality into smaller and easier to understand modules. For example, checkpkg is complex at places, but writing a single check is easy: you get a data structure, you pull out the list or dict you want, you look at the data that interests you and you call a function to report an error. It doesn't get simpler than that, at least I can't think of a simpler way to do it, given how complex are the packages themselves. Perhaps it's the matter of documentation and encouragement? If you have any ideas how to make things easier for people to get involved, please share. Maciej From maciej at opencsw.org Wed Aug 24 11:37:12 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 24 Aug 2011 10:37:12 +0100 Subject: [csw-maintainers] Technical Summercamp in Kiel 2011 In-Reply-To: <4E46BF16.50602@opencsw.org> References: <958B6EC9-5CE1-4B2E-BE39-2C85617D3EDD@opencsw.org> <4E46BF16.50602@opencsw.org> Message-ID: 2011/8/13 ?hsan?Do?an : > Am 11.08.2011 10:16, schrieb Maciej Blizi?ski: > >>> There's a problem with the date for me: there is a work event I need >>> to attend, which showed up after I filled in the doodle poll. ?If we >>> do the camp on the weekend of the 17th, I can't come! >> >> Peter, Dago and I met on IRC this morning to discuss possible >> solutions. ?Almost everyone declared the next weekend (24-25 of >> September) as available. ?Peter figured out a way to make that weekend >> available for him, so all 7 people are available now. >> >> In conclusion, we're changing the date of Summercamp 2011 to 24-25 of September. > > Just booked the hotel. I will arrive on Friday afternoon and leave on > Monday noon. Just booked the flight, arriving Friday afternoon and leaving Sunday afternoon. I also attempted to book the hotel, but the code was rejected, and I'm checking with Dago whether our rate is still on. Maciej From maciej at opencsw.org Thu Aug 25 13:23:26 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Thu, 25 Aug 2011 12:23:26 +0100 Subject: [csw-maintainers] Reasons for gcc to have a separate prefix Message-ID: Hi Peter, I'd like to ask about the prefix for gcc. Current user experience with gcc looks more or less like this[1]: pkgutil -i gcc (fails) pkgutil -a gcc ah. pkgutil -i gcc4core gcc test.c gcc: command not found wtf? Didn't I just install gcc? (bangs head against desk) find /opt/csw -name gcc (gets coffee) /opt/csw/gcc4/bin/gcc is found (raises eyebrows) Why would you do that? (shrugs and grumbles) The discussion about why build gcc in a separate prefix predates me, so perhaps you remember why we do it this way. I tend to think of using separate prefixes as of a sloppy way to support multiple versions of a piece of software. Alternatively, as of an inability to deprecate old versions, or as of poor understanding how shared libraries work. Do you recall the reasons to compile gcc into a different prefix, and do you think these reasons still apply today? Maciej [1] http://serverfault.com/questions/229606/how-to-install-gnu-build-tools-from-blastwave From bwalton at opencsw.org Fri Aug 26 02:21:40 2011 From: bwalton at opencsw.org (Ben Walton) Date: Thu, 25 Aug 2011 20:21:40 -0400 Subject: [csw-maintainers] Reasons for gcc to have a separate prefix In-Reply-To: References: Message-ID: <1314317946-sup-2767@pinkfloyd.chass.utoronto.ca> Excerpts from Maciej Blizi?ski's message of Thu Aug 25 07:23:26 -0400 2011: Hi Maciej, > Do you recall the reasons to compile gcc into a different prefix, > and do you think these reasons still apply today? I'd definitely love to see gcc move into /opt/csw. Alternatives should make this possible. A symlink from /opt/csw/gcc4 -> /opt/csw should be provided if such a move is ever attempted though so that things with the path coded in continue working. -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From pfelecan at opencsw.org Fri Aug 26 12:51:03 2011 From: pfelecan at opencsw.org (Peter FELECAN) Date: Fri, 26 Aug 2011 12:51:03 +0200 Subject: [csw-maintainers] Reasons for gcc to have a separate prefix References: Message-ID: Maciej Blizi?ski writes: > The discussion about why build gcc in a separate prefix predates me, > so perhaps you remember why we do it this way. I tend to think of > using separate prefixes as of a sloppy way to support multiple > versions of a piece of software. Alternatively, as of an inability to > deprecate old versions, or as of poor understanding how shared > libraries work. > > Do you recall the reasons to compile gcc into a different prefix, and > do you think these reasons still apply today? * History: - gcc2: - until the end of 2004 it was maintained by Phil Brown and Markus Gyger - I took it up and provided the packages - at that time, some projects' preferred compiler set was that branch, e.g., MPlayer - gcc3 - same story as for gcc2 - the last supported version for this branch, 3.4.6, is available - gcc4 - I maintained this branch since March 2005, the last version that I delivered was 4.0.2 - after 2006, Mike Waters was the maintainer for this branch * Prefix: the separate prefix was the rule from the beginning and it was enforced by the previous release manager; I proposed at the time an alternatives system but to no avail, BTW, this is why I implemented a specific alternative system for Emacs. * Proposition for the future: Support only one branch, 4, but use alternatives for supporting different releases and make provision for a 5th branch; I think that I can release a branch 3 package using alternatives after it's implemented in the 4th. -- Peter From maciej at opencsw.org Sat Aug 27 13:27:49 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sat, 27 Aug 2011 12:27:49 +0100 Subject: [csw-maintainers] Reasons for gcc to have a separate prefix In-Reply-To: References: Message-ID: Hi Peter, Thanks a lot for the write-up. You're helping us avoid the old mistakes. 2011/8/26 Peter FELECAN : > * History: > - gcc2: > ?- until the end of 2004 it was maintained by Phil Brown and Markus Gyger > ?- I took it up and provided the packages > ?- at that time, some projects' preferred compiler set was that > ? ? ? ?branch, e.g., MPlayer That would be the biggest problem with deprecating old compilers. I think it might occur again when gcc does a major update in the future, so we need to have a plan. > - gcc3 > ?- same story as for gcc2 > ?- the last supported version for this branch, 3.4.6, is available > - gcc4 > ?- I maintained this branch since March 2005, the last version that I > ? ?delivered was 4.0.2 > ?- after 2006, Mike Waters was the maintainer for this branch > * Prefix: > ?the separate prefix was the rule from the beginning and it was > ?enforced by the previous release manager; I proposed at the time an > ?alternatives system but to no avail, BTW, this is why I implemented a > ?specific alternative system for Emacs. > * Proposition for the future: > ?Support only one branch, 4, but use alternatives for supporting > ?different releases and make provision for a 5th branch; I think that > ?I can release a branch 3 package using alternatives after it's > ?implemented in the 4th. The gcc build system supports adding a binary prefix, so we could set it to something like "gcc4" and have gcc4gcc, gcc4g++, etc. That would also start matching the package names. However, alternatives are not a viable solution for the buildfarm. We all work on the same hosts and we need to make it possible to build with gccN and gcc(N+1) at the same time. You can't have buildfarm admins flip symlink every time a maintainer wants to switch between e.g. gcc-3 and gcc-4. How about this plan: - we support one current/blessed version of gcc, which lives in /opt/csw - when there's a need to have an old version of gcc available, we recompile the old version into a separate prefix, as a legacy mode The algorithm goes like this: T = 1 gccN: prefix=/opt/csw T = 2 gccN: prefix=/opt/csw/gccN (recompile) gcc(N+1): prefix=/opt/csw T = 3 gccN: prefix=/opt/csw/gccN gcc(N+1): prefix=/opt/csw/gcc(N+1) gcc(N+2): prefix=/opt/csw etc. The annoying part of this plan is that you have to recompile the old version of gcc to make it available, you can't just let the packages be. When you replace gccN with gcc(N+1), you have to make the extra effort of moving gccN into own prefix. I consider it a reasonable trade off. Maciej From maciej at opencsw.org Sat Aug 27 13:29:32 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sat, 27 Aug 2011 12:29:32 +0100 Subject: [csw-maintainers] Retiring old versions of GCC Message-ID: Hello maintainers, Apart from multi-version support, we can also think about retiring the older versions of gcc. When there's no reason to keep an old version around, I'd suggest removing it, although I understand it's painful to remove something that e.g. once required a lot of work to get in. But keeping old unused software around only clutters the catalog. For example, there might potentially still be a use case for gcc-3, but I doubt that there's any reason to keep gcc2 around. What are your thoughts, and how would you approach the issue? Maciej From dam at opencsw.org Sat Aug 27 14:40:20 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Sat, 27 Aug 2011 14:40:20 +0200 Subject: [csw-maintainers] Reasons for gcc to have a separate prefix In-Reply-To: References: Message-ID: <950E3F36-5F3F-4332-9E70-2635BB75AC4D@opencsw.org> Hi, Am 27.08.2011 um 13:27 schrieb Maciej Blizi?ski: > Hi Peter, > > Thanks a lot for the write-up. You're helping us avoid the old mistakes. > > 2011/8/26 Peter FELECAN : >> * History: >> - gcc2: >> - until the end of 2004 it was maintained by Phil Brown and Markus Gyger >> - I took it up and provided the packages >> - at that time, some projects' preferred compiler set was that >> branch, e.g., MPlayer > > That would be the biggest problem with deprecating old compilers. I > think it might occur again when gcc does a major update in the future, > so we need to have a plan. > >> - gcc3 >> - same story as for gcc2 >> - the last supported version for this branch, 3.4.6, is available >> - gcc4 >> - I maintained this branch since March 2005, the last version that I >> delivered was 4.0.2 >> - after 2006, Mike Waters was the maintainer for this branch >> * Prefix: >> the separate prefix was the rule from the beginning and it was >> enforced by the previous release manager; I proposed at the time an >> alternatives system but to no avail, BTW, this is why I implemented a >> specific alternative system for Emacs. >> * Proposition for the future: >> Support only one branch, 4, but use alternatives for supporting >> different releases and make provision for a 5th branch; I think that >> I can release a branch 3 package using alternatives after it's >> implemented in the 4th. > > The gcc build system supports adding a binary prefix, so we could set > it to something like "gcc4" and have gcc4gcc, gcc4g++, etc. That would > also start matching the package names. > > However, alternatives are not a viable solution for the buildfarm. We > all work on the same hosts and we need to make it possible to build > with gccN and gcc(N+1) at the same time. You can't have buildfarm > admins flip symlink every time a maintainer wants to switch between > e.g. gcc-3 and gcc-4. > > How about this plan: > > - we support one current/blessed version of gcc, which lives in /opt/csw > - when there's a need to have an old version of gcc available, we > recompile the old version into a separate prefix, as a legacy mode > > The algorithm goes like this: > > T = 1 > gccN: prefix=/opt/csw > > T = 2 > gccN: prefix=/opt/csw/gccN (recompile) > gcc(N+1): prefix=/opt/csw > > T = 3 > gccN: prefix=/opt/csw/gccN > gcc(N+1): prefix=/opt/csw/gcc(N+1) > gcc(N+2): prefix=/opt/csw > > etc. > > The annoying part of this plan is that you have to recompile the old > version of gcc to make it available, you can't just let the packages > be. When you replace gccN with gcc(N+1), you have to make the extra > effort of moving gccN into own prefix. I consider it a reasonable > trade off. What about the "jdk" technique to always package with prefix and link CSWgcc in /opt/csw/bin to the latest or as alternative? This would sacve repckaging on update. Best regards -- Dago From maciej at opencsw.org Sat Aug 27 15:14:31 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sat, 27 Aug 2011 14:14:31 +0100 Subject: [csw-maintainers] Reasons for gcc to have a separate prefix In-Reply-To: <950E3F36-5F3F-4332-9E70-2635BB75AC4D@opencsw.org> References: <950E3F36-5F3F-4332-9E70-2635BB75AC4D@opencsw.org> Message-ID: 2011/8/27 Dagobert Michelsen : >> The annoying part of this plan is that you have to recompile the old >> version of gcc to make it available, you can't just let the packages >> be. ?When you replace gccN with gcc(N+1), you have to make the extra >> effort of moving gccN into own prefix. ?I consider it a reasonable >> trade off. > > What about the "jdk" technique to always package with prefix and link CSWgcc in > /opt/csw/bin to the latest or as alternative? This would sacve repckaging on > update. There is a problem with that. We want the libraries to be installed in /opt/csw/lib, even though the rest of the package would be under /opt/csw/gcc4. During compilation, gcc looks for its includes relative to the libdir. The includes live under $(prefix)/include/..., so there's a bit of code in the gcc build system to calculate the relative path going from the libdir to the include dir. This function uses a set of regular expressions, which in our case (prefix=/opt/csw/gcc4, libdir=/opt/csw/lib) do not calculate the right answer. As result, g++ fails to find the include files. We could try to fix this function, but I don't know what their other corner cases are, so there's a lot of potential of breaking builds for other people. I'm suspecting that we could support multiple versions by adding a prefix to the binary names and providing symlinks for them. For example (not literal symlinks, but something to give you the idea): --binary-prefix=gcc4 /opt/csw/bin/gcc ? gcc4gcc /opt/csw/bin/g++ ? gcc4g++ ... Maciej From dam at opencsw.org Sat Aug 27 22:31:06 2011 From: dam at opencsw.org (Dagobert Michelsen) Date: Sat, 27 Aug 2011 22:31:06 +0200 Subject: [csw-maintainers] Problem with pygtk In-Reply-To: References: <365106AE-4065-4AAD-9790-0D1799A17798@opencsw.org> Message-ID: <5343F633-B69B-40F3-BD67-9336711A6112@opencsw.org> Hi Maciej, Am 27.08.2011 um 10:50 schrieb Maciej Blizi?ski: > 1. to build pygtk (the newest), we need a newer pygobject: Requested > 'pygobject-2.0 >= 2.21.3' but version of PyGObject is 2.15.4 > > 2. to build pygobject a newer glib is needed: checking for GLIB - > version >= 2.24.0... no Thanks for the info, I looked at an updated glib for another package as dependency lately and it does not look good, it compiles but lots of tests are failing, at least these: > /gvariant/parser: ** > ERROR:gvariant.c:3730:test_parses: assertion failed (tests[i] == printed): ("inf" == "Infinity") > FAIL > GTester: last random seed: R02S02eadc789816e349295c132201328082 > > ERROR:utils.c:214:test_find_program: assertion failed (res == "/bin/sh"): ("/usr/bin/sh" == "/bin/sh") > /utils/find-program: FAIL > > /unicode/collate/1: ** > ERROR:collate.c:62:do_collate: assertion failed (str == test->sorted[i]): ("event.h" == "eventgenerator.c") > FAIL > GTester: last random seed: R02Sbf664a6dd3bf0b2ef74cba9dca01b2c3 > > /checksum/MD5/2: FAIL > GTester: last random seed: R02S2bc54047877005d2ccb56daf49f3e03a > > /GDateTime/equal: ** > ERROR:gdatetime.c:192:test_GDateTime_equal: assertion failed (g_date_time_get_utc_offset (dt1) / G_USEC_PER_SEC == (-3 * 3600)): (0 == -10800) > FAIL > GTester: last random seed: R02S06f5565a24ef0b341b999017422ab99e > > /GDateTime/get_year: ** > ERROR:gdatetime.c:105:test_GDateTime_new_from_unix: assertion failed (g_date_time_get_hour (dt) == tm.tm_hour): (19 == 21) > OK > /GDateTime/hash: OK > /GDateTime/new_from_unix: FAIL > GTester: last random seed: R02Sc0a7bc1b621aef05955352333574ee7f > > /GDateTime/get_utc_offset: ** > ERROR:gdatetime.c:593:test_GDateTime_new_full: assertion failed ("BRT" == g_date_time_get_timezone_abbreviation (dt)): ("BRT" == "UTC") > OK > /GDateTime/get_year: OK > /GDateTime/hash: OK > /GDateTime/new_from_unix_utc: OK > /GDateTime/new_from_timeval: OK > /GDateTime/new_full: FAIL > GTester: last random seed: R02Sa530a15aed8aa3993babdd7b4839d0a7 > > *** Solaris 10 mit _XPG6 > > /gvariant/invalid-varargs: OK > /gvariant/varargs: > GLib-CRITICAL **: file gvariant.c: line 4750: assertion `first_time || format_string == GVSI(iter)->loop_format' failed > FAIL > GTester: last random seed: R02Sf7f75f608928ec55ec5fc64a5cb1ab29 The tests were from the most recent 2.29.16. Every one of them would need close inspection and a bug report. Lets first get gcc and the release process done. Apart from that I would find having a recent glib/gtk+ important enough to spent some time on it. Best regards -- Dago -- "You don't become great by trying to be great, you become great by wanting to do something, and then doing it so hard that you become great in the process." - xkcd #896 From maciej at opencsw.org Sun Aug 28 00:41:50 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sat, 27 Aug 2011 23:41:50 +0100 Subject: [csw-maintainers] Problem with pygtk In-Reply-To: <5343F633-B69B-40F3-BD67-9336711A6112@opencsw.org> References: <365106AE-4065-4AAD-9790-0D1799A17798@opencsw.org> <5343F633-B69B-40F3-BD67-9336711A6112@opencsw.org> Message-ID: 2011/8/27 Dagobert Michelsen : > The tests were from the most recent 2.29.16. > > Every one of them would need close inspection and a bug report. > Lets first get gcc and the release process done. Apart from that > I would find having a recent glib/gtk+ important enough to spent > some time on it. Yes, I'll focus on gcc4 for now. From maciej at opencsw.org Sun Aug 28 00:41:50 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sat, 27 Aug 2011 23:41:50 +0100 Subject: [csw-maintainers] Problem with pygtk In-Reply-To: <5343F633-B69B-40F3-BD67-9336711A6112@opencsw.org> References: <365106AE-4065-4AAD-9790-0D1799A17798@opencsw.org> <5343F633-B69B-40F3-BD67-9336711A6112@opencsw.org> Message-ID: 2011/8/27 Dagobert Michelsen : > The tests were from the most recent 2.29.16. > > Every one of them would need close inspection and a bug report. > Lets first get gcc and the release process done. Apart from that > I would find having a recent glib/gtk+ important enough to spent > some time on it. Yes, I'll focus on gcc4 for now. From maciej at opencsw.org Sun Aug 28 00:41:50 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Sat, 27 Aug 2011 23:41:50 +0100 Subject: [csw-maintainers] Problem with pygtk In-Reply-To: <5343F633-B69B-40F3-BD67-9336711A6112@opencsw.org> References: <365106AE-4065-4AAD-9790-0D1799A17798@opencsw.org> <5343F633-B69B-40F3-BD67-9336711A6112@opencsw.org> Message-ID: 2011/8/27 Dagobert Michelsen : > The tests were from the most recent 2.29.16. > > Every one of them would need close inspection and a bug report. > Lets first get gcc and the release process done. Apart from that > I would find having a recent glib/gtk+ important enough to spent > some time on it. Yes, I'll focus on gcc4 for now. From pfelecan at opencsw.org Sun Aug 28 15:04:45 2011 From: pfelecan at opencsw.org (Peter FELECAN) Date: Sun, 28 Aug 2011 15:04:45 +0200 Subject: [csw-maintainers] Reasons for gcc to have a separate prefix In-Reply-To: ("Maciej =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Sat, 27 Aug 2011 12:27:49 +0100") References: Message-ID: Maciej Blizi?ski writes: > However, alternatives are not a viable solution for the buildfarm. We > all work on the same hosts and we need to make it possible to build > with gccN and gcc(N+1) at the same time. You can't have buildfarm > admins flip symlink every time a maintainer wants to switch between > e.g. gcc-3 and gcc-4. Look at how Debian maintainers solved this issue and use please alternatives. For a build farm there is a default gcc compiler, the current release; alternative gcc releases should be used directly, without alternatives help or in dedicated zones. To cripple the public packages for a specific build farm requirement seems to me a big compromise. -- Peter From pfelecan at opencsw.org Sun Aug 28 15:06:47 2011 From: pfelecan at opencsw.org (Peter FELECAN) Date: Sun, 28 Aug 2011 15:06:47 +0200 Subject: [csw-maintainers] Retiring old versions of GCC In-Reply-To: ("Maciej =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Sat, 27 Aug 2011 12:29:32 +0100") References: Message-ID: Maciej Blizi?ski writes: > Hello maintainers, > > Apart from multi-version support, we can also think about retiring the > older versions of gcc. When there's no reason to keep an old version > around, I'd suggest removing it, although I understand it's painful to > remove something that e.g. once required a lot of work to get in. But > keeping old unused software around only clutters the catalog. For > example, there might potentially still be a use case for gcc-3, but I > doubt that there's any reason to keep gcc2 around. What are your > thoughts, and how would you approach the issue? I don't have any objection toward its removal. However, is there a conflict with the current or future gcc releases? If there is no conflict don't remove it. -- Peter From bwalton at opencsw.org Sun Aug 28 15:27:42 2011 From: bwalton at opencsw.org (Ben Walton) Date: Sun, 28 Aug 2011 09:27:42 -0400 Subject: [csw-maintainers] Retiring old versions of GCC In-Reply-To: References: Message-ID: <1314537998-sup-2689@pinkfloyd.chass.utoronto.ca> Excerpts from Peter FELECAN's message of Sun Aug 28 09:06:47 -0400 2011: > > Apart from multi-version support, we can also think about retiring the > > older versions of gcc. When there's no reason to keep an old version > > around, I'd suggest removing it, although I understand it's painful to > > remove something that e.g. once required a lot of work to get in. But > > keeping old unused software around only clutters the catalog. For > > example, there might potentially still be a use case for gcc-3, but I > > doubt that there's any reason to keep gcc2 around. What are your > > thoughts, and how would you approach the issue? > > I don't have any objection toward its removal. However, is there a > conflict with the current or future gcc releases? If there is no > conflict don't remove it. I think we need to retain the *rt packages though...either that or also drop the few things that depend on them. I don't have any comment on whether or not that is good/bad as I don't use those packages. In this case, I'd likely err toward leaving the *rt packages around. Thanks -Ben -- Ben Walton Systems Programmer - CHASS University of Toronto C:416.407.5610 | W:416.978.4302 From bonivart at opencsw.org Sun Aug 28 23:27:35 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Sun, 28 Aug 2011 23:27:35 +0200 Subject: [csw-maintainers] Technical Summercamp in Kiel 2011 In-Reply-To: References: <958B6EC9-5CE1-4B2E-BE39-2C85617D3EDD@opencsw.org> <4E46BF16.50602@opencsw.org> Message-ID: 2011/8/24 Maciej Blizi?ski : > Just booked the flight, arriving Friday afternoon and leaving Sunday > afternoon. ?I also attempted to book the hotel, but the code was > rejected, and I'm checking with Dago whether our rate is still on. See you there. :) I will be taking a seven hour drive to Kiel, I will arrive Thursday evening and stay until Monday morning when I will take a detour to Hamburg to buy accessories for my motorcycle (much cheaper in Germany than in Sweden) before heading north. The hotel code still didn't work online so I wrote them an e-mail and they fixed it, no problem. /peter From maciej at opencsw.org Mon Aug 29 09:16:00 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 29 Aug 2011 08:16:00 +0100 Subject: [csw-maintainers] [POLICY] Policy-team, policy docs, licenses In-Reply-To: References: Message-ID: 2011/8/20 Peter FELECAN : >> Excellent! ?I have since looked at sphinx, a higher level tool which helps >> with structured documentation. It uses reStructured Text as the lightweight >> markup language. I will send the patch and a link to an example when I'm >> back home at the beginning of August. > > looked == switched ? I wrote a proof of concept[1], but I haven't committed anything yet. I'm currently focusing on gcc4, and we also need to start signing and pushing out new catalogs. I'll look at the documentation then these two things are completed. Maciej [1] http://buildfarm.opencsw.org/~maciej/policy-sphinx/_build/html/ From pfelecan at opencsw.org Mon Aug 29 19:14:42 2011 From: pfelecan at opencsw.org (Peter FELECAN) Date: Mon, 29 Aug 2011 19:14:42 +0200 Subject: [csw-maintainers] Reasons for gcc to have a separate prefix In-Reply-To: ("Maciej =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Sat, 27 Aug 2011 12:27:49 +0100") References: Message-ID: Maciej Blizi?ski writes: > However, alternatives are not a viable solution for the buildfarm. We > all work on the same hosts and we need to make it possible to build > with gccN and gcc(N+1) at the same time. You can't have buildfarm > admins flip symlink every time a maintainer wants to switch between > e.g. gcc-3 and gcc-4. Just a follow up on my own advice to look how Debian does it... * How Debian manages multiple versions of GCC Useful information is found, on an installed system, in /usr/share/doc/gcc/README.Debian ** Default version Each Debian release has a default set of compilers issued by a generic source package called gcc-defaults (http://ftp.de.debian.org/debian/pool/main/g/gcc-defaults/gcc-defaults_1.107.tar.gz) The generated packages supply a symbolic link toward the default driver, e.g., for the C compiler, there is /usr/bin/gcc which point toward /usr/bin/gcc-4.6 Note that you find the README.Debian in the above mentioned source package. ** Alternatives Alternatives are used at a higher level, e.g., /usr/bin/cc is an generic name, having cc as the alternative name with a possible alternative path of /usr/bin/gcc ** Using a non default version When compiling with a non-default version of the compiler you must explicitly specify the one that you wish to use, e.g. CC=gcc-4.1 ./configure ** File-system organization each version of GCC provides its specific components installed in corresponding directories; e.g., /usr/lib/gcc/i486-linux-gnu/4.3; this organization is directly supported by the GCC build system, i.e., it doesn't need modifications ** Run-time shared libraries The run-time shared libraries conforms with the shared libraries policy and thus supports multiple releases installed simultaneously with one default version. * How Open CSW should manage multiple versions of GCC I don't discuss here the integration of Sun Studio in the alternatives system but that would be a nice feature. ** Default version the most recent version of GCC is considered as the default one and provides the /usr/bin/gcc symbolic link ** Alternatives Use the cc as the generic name ** Using a non default version Same as Debian ** File-system organization Similar with that of Debian ** Run-time shared libraries Similar with that of Debian -- Peter From ja at opencsw.org Mon Aug 29 20:19:10 2011 From: ja at opencsw.org (Juergen Arndt) Date: Mon, 29 Aug 2011 20:19:10 +0200 Subject: [csw-maintainers] Problems with chkpkg Message-ID: Hi all, I have some problems with chkpkg, which I cannot solve: I tuned my gar recipe for nagios that way, that I can build my package without errors with "gmake platforms". When uploading the package with csw-upload-pkg I get the following messages: #> csw-upload-pkg /home/ja/staging/build-29.Aug.2011/nagios-3.2.3\,REV\=2011.08.29-SunOS5.9-*Processing 2 file(s). Please wait. Uploading '/home/ja/staging/build-29.Aug.2011/nagios-3.2.3,REV=2011.08.29-SunOS5.9-i386-CSW.pkg.gz' Uploading '/home/ja/staging/build-29.Aug.2011/nagios-3.2.3,REV=2011.08.29-SunOS5.9-sparc-CSW.pkg.gz' Checking 1 package(s) against catalog unstable sparc SunOS5.9 Checking 1 package(s) against catalog unstable i386 SunOS5.9 Checking 1 package(s) against catalog unstable sparc SunOS5.10 Checking 1 package(s) against catalog unstable i386 SunOS5.10 Checks failed for catalogs: - sparc SunOS5.9 nagios-3.2.3,REV=2011.08.29-SunOS5.9-sparc-CSW.pkg.gz To see errors, run: /opt/csw/bin/checkpkg --catalog-release unstable --os-release SunOS5.9 --architecture sparc 814c15621d0815e4c127e9330362ae22 - i386 SunOS5.9 nagios-3.2.3,REV=2011.08.29-SunOS5.9-i386-CSW.pkg.gz To see errors, run: /opt/csw/bin/checkpkg --catalog-release unstable --os-release SunOS5.9 --architecture i386 4dd45d37e1a11088bf9aeccda8efa6d0 - sparc SunOS5.10 nagios-3.2.3,REV=2011.08.29-SunOS5.9-sparc-CSW.pkg.gz To see errors, run: /opt/csw/bin/checkpkg --catalog-release unstable --os-release SunOS5.10 --architecture sparc 814c15621d0815e4c127e9330362ae22 - i386 SunOS5.10 nagios-3.2.3,REV=2011.08.29-SunOS5.9-i386-CSW.pkg.gz To see errors, run: /opt/csw/bin/checkpkg --catalog-release unstable --os-release SunOS5.10 --architecture i386 4dd45d37e1a11088bf9aeccda8efa6d0 Packages have not been submitted to the unstable catalog. ja at login:/home/ja/mgar/pkg/nagios/trunk:> /opt/csw/bin/checkpkg --catalog-release unstable --os-release SunOS5.9 --architecture sparc 814c15621d0815e4c127e9330362ae22 INFO:root:Unwrapping candies... 100% |#############################################################################################################################| INFO:root:Tasting candies one by one... 100% |#############################################################################################################################| INFO:root:Tasting them all at once... INFO:root:Stuffing the candies under the pillow... 100% |#############################################################################################################################| CSWnagios: .... * Dependency issues of CSWnagios: * CSWlibiconv2 is needed by CSWnagios, because: * - opt/csw/nagios/sbin/histogram.cgi needs the libiconv.so.2 soname * - opt/csw/nagios/sbin/statusmap.cgi needs the libiconv.so.2 soname * - opt/csw/nagios/sbin/trends.cgi needs the libiconv.so.2 soname * RUNTIME_DEP_PKGS_CSWnagios += CSWlibiconv2 * CSWlibpng12-0 is needed by CSWnagios, because: * - opt/csw/nagios/sbin/histogram.cgi needs the libpng12.so.0 soname * - opt/csw/nagios/sbin/statusmap.cgi needs the libpng12.so.0 soname * - opt/csw/nagios/sbin/trends.cgi needs the libpng12.so.0 soname * RUNTIME_DEP_PKGS_CSWnagios += CSWlibpng12-0 * CSWlibz1 is needed by CSWnagios, because: * - opt/csw/nagios/sbin/trends.cgi needs the libz.so.1 soname * - opt/csw/nagios/sbin/statusmap.cgi needs the libz.so.1 soname * - opt/csw/nagios/sbin/histogram.cgi needs the libz.so.1 soname * RUNTIME_DEP_PKGS_CSWnagios += CSWlibz1 * If you don't know of any reasons to include these dependencies, you might remove them: * ? CSWap2modphp5 * ? CSWbdb * ? CSWgd * ? CSWggettextrt * ? CSWglib2 * ? CSWiconv * ? CSWlibtoolrt * ? CSWnagiosp * ? CSWosslrt * ? CSWpng * ? CSWzlib # Checkpkg suggests adding the following lines to the GAR recipe: # This is a summary; see above for details. RUNTIME_DEP_PKGS_CSWnagios += CSWlibiconv2 RUNTIME_DEP_PKGS_CSWnagios += CSWlibpng12-0 RUNTIME_DEP_PKGS_CSWnagios += CSWlibz1 If any of the reported errors were false positives, you can override them pasting the lines below to the GAR recipe. CHECKPKG_OVERRIDES_CSWnagios += surplus-dependency|CSWpng CHECKPKG_OVERRIDES_CSWnagios += surplus-dependency|CSWzlib CHECKPKG_OVERRIDES_CSWnagios += surplus-dependency|CSWiconv CHECKPKG_OVERRIDES_CSWnagios += missing-dependency|CSWlibpng12-0 CHECKPKG_OVERRIDES_CSWnagios += missing-dependency|CSWlibz1 CHECKPKG_OVERRIDES_CSWnagios += missing-dependency|CSWlibiconv2 Well, are there packages named CSWlibiconv2, CSWlibpng12-0 and CSWlibz1? I cannot find them. I thougt, I would be fine with CSWiconv, CSWpng and CSWzlib. Second question: If I define these overrides, I cannot build the packages anymore, because chkpkg throws an error because the overrides don't match any error. What can I do to solve this? Regards, Juergen -- Juergen Arndt From bonivart at opencsw.org Mon Aug 29 20:27:58 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Mon, 29 Aug 2011 20:27:58 +0200 Subject: [csw-maintainers] Problems with chkpkg In-Reply-To: References: Message-ID: On Mon, Aug 29, 2011 at 8:19 PM, Juergen Arndt wrote: > # Checkpkg suggests adding the following lines to the GAR recipe: > # This is a summary; see above for details. > RUNTIME_DEP_PKGS_CSWnagios += CSWlibiconv2 > RUNTIME_DEP_PKGS_CSWnagios += CSWlibpng12-0 > RUNTIME_DEP_PKGS_CSWnagios += CSWlibz1 > If any of the reported errors were false positives, you can override them > pasting the lines below to the GAR recipe. > CHECKPKG_OVERRIDES_CSWnagios += surplus-dependency|CSWpng > CHECKPKG_OVERRIDES_CSWnagios += surplus-dependency|CSWzlib > CHECKPKG_OVERRIDES_CSWnagios += surplus-dependency|CSWiconv > CHECKPKG_OVERRIDES_CSWnagios += missing-dependency|CSWlibpng12-0 > CHECKPKG_OVERRIDES_CSWnagios += missing-dependency|CSWlibz1 > CHECKPKG_OVERRIDES_CSWnagios += missing-dependency|CSWlibiconv2 > > Well, are there packages named CSWlibiconv2, CSWlibpng12-0 and CSWlibz1? I cannot find them. I thougt, I would be fine with CSWiconv, CSWpng and CSWzlib. You will find them here: http://mirror.opencsw.org/opencsw-future/unstable We build against unstable now and you should use those hosts (e.g. unstable9x) to have the right packages installed. /peter From ja at opencsw.org Mon Aug 29 20:31:39 2011 From: ja at opencsw.org (Juergen Arndt) Date: Mon, 29 Aug 2011 20:31:39 +0200 Subject: [csw-maintainers] Problems with chkpkg In-Reply-To: References: Message-ID: On 29.08.2011, at 20:27, Peter Bonivart wrote: > On Mon, Aug 29, 2011 at 8:19 PM, Juergen Arndt wrote: >> # Checkpkg suggests adding the following lines to the GAR recipe: >> # This is a summary; see above for details. >> RUNTIME_DEP_PKGS_CSWnagios += CSWlibiconv2 >> RUNTIME_DEP_PKGS_CSWnagios += CSWlibpng12-0 >> RUNTIME_DEP_PKGS_CSWnagios += CSWlibz1 >> If any of the reported errors were false positives, you can override them >> pasting the lines below to the GAR recipe. >> CHECKPKG_OVERRIDES_CSWnagios += surplus-dependency|CSWpng >> CHECKPKG_OVERRIDES_CSWnagios += surplus-dependency|CSWzlib >> CHECKPKG_OVERRIDES_CSWnagios += surplus-dependency|CSWiconv >> CHECKPKG_OVERRIDES_CSWnagios += missing-dependency|CSWlibpng12-0 >> CHECKPKG_OVERRIDES_CSWnagios += missing-dependency|CSWlibz1 >> CHECKPKG_OVERRIDES_CSWnagios += missing-dependency|CSWlibiconv2 >> >> Well, are there packages named CSWlibiconv2, CSWlibpng12-0 and CSWlibz1? I cannot find them. I thougt, I would be fine with CSWiconv, CSWpng and CSWzlib. > > You will find them here: http://mirror.opencsw.org/opencsw-future/unstable > > We build against unstable now and you should use those hosts (e.g. > unstable9x) to have the right packages installed. Hu, it seems, that I missed something :) Ok, I will try that. Thank you! Juergen -- Juergen Arndt From maciej at opencsw.org Mon Aug 29 22:33:11 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Mon, 29 Aug 2011 21:33:11 +0100 Subject: [csw-maintainers] Reasons for gcc to have a separate prefix In-Reply-To: References: Message-ID: 2011/8/29 Peter FELECAN : > Maciej Blizi?ski writes: > >> However, alternatives are not a viable solution for the buildfarm. We >> all work on the same hosts and we need to make it possible to build >> with gccN and gcc(N+1) at the same time. ?You can't have buildfarm >> admins flip symlink every time a maintainer wants to switch between >> e.g. gcc-3 and gcc-4. > > Just a follow up on my own advice to look how Debian does it... > > * How Debian manages multiple versions of GCC > ?Useful information is found, on an installed system, in > ?/usr/share/doc/gcc/README.Debian > ** Default version > ? Each Debian release has a default set of compilers issued by a > ? generic source package called gcc-defaults > ? (http://ftp.de.debian.org/debian/pool/main/g/gcc-defaults/gcc-defaults_1.107.tar.gz) > ? The generated packages supply a symbolic link toward the default > ? driver, e.g., for the C compiler, there is /usr/bin/gcc which point > ? toward /usr/bin/gcc-4.6 Yes, I found out this can be done by using --program-suffix. > ? Note that you find the README.Debian in the above mentioned source > ? package. > ** Alternatives > ? Alternatives are used at a higher level, e.g., /usr/bin/cc is an > ? generic name, having cc as the alternative name with a possible > ? alternative path of /usr/bin/gcc > ** Using a non default version > ? When compiling with a non-default version of the compiler you must > ? explicitly specify the one that you wish to use, e.g. > ? CC=gcc-4.1 ./configure > ** File-system organization > ? each version of GCC provides its specific components installed in > ? corresponding directories; e.g., /usr/lib/gcc/i486-linux-gnu/4.3; > ? this organization is directly supported by the GCC build system, > ? i.e., it doesn't need modifications > ** Run-time shared libraries > ? The run-time shared libraries conforms with the shared libraries > ? policy and thus supports multiple releases installed simultaneously > ? with one default version. > > * How Open CSW should manage multiple versions of GCC > ?I don't discuss here the integration of Sun Studio in the > ?alternatives system but that would be a nice feature. > ** Default version > ? the most recent version of GCC is considered as the default one and > ? provides the /usr/bin/gcc symbolic link > ** Alternatives > ? Use the cc as the generic name > ** Using a non default version > ? Same as Debian > ** File-system organization > ? Similar with that of Debian > ** Run-time shared libraries > ? Similar with that of Debian Roger that. The plan looks good. I'm compiling a new cut of the packages right now. Maciej From bonivart at opencsw.org Tue Aug 30 09:20:33 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Tue, 30 Aug 2011 09:20:33 +0200 Subject: [csw-maintainers] Fwd: CSW catalog problems (2011-08-30 07:25) In-Reply-To: <201108300525.p7U5PC2e015378@ajax2.int.lio.se> References: <201108300525.p7U5PC2e015378@ajax2.int.lio.se> Message-ID: This is weird, why only that catalog? catalog.opencsw-future.unstable.i386.5.9 =========================== ERROR! Dependency CSWkrb5user of package CSWkrb5kdc is missing. /peter From maciej at opencsw.org Tue Aug 30 09:44:55 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 30 Aug 2011 08:44:55 +0100 Subject: [csw-maintainers] Fwd: CSW catalog problems (2011-08-30 07:25) In-Reply-To: References: <201108300525.p7U5PC2e015378@ajax2.int.lio.se> Message-ID: 2011/8/30 Peter Bonivart : > This is weird, why only that catalog? > > > catalog.opencsw-future.unstable.i386.5.9 > =========================== > ERROR! Dependency CSWkrb5user of package CSWkrb5kdc is missing. Hm. None of these two exist in the catalog as files: maciej at login [login]:~ > ls /home/mirror/opencsw-future/unstable/*/*/krb5kdc* /home/mirror/opencsw-future/unstable/*/*/krb5kdc*: No such file or directory maciej at login [login]:~ > ls /home/mirror/opencsw-future/unstable/*/*/krb5user* /home/mirror/opencsw-future/unstable/*/*/krb5user*: No such file or directory Maciej From bonivart at opencsw.org Tue Aug 30 09:51:06 2011 From: bonivart at opencsw.org (Peter Bonivart) Date: Tue, 30 Aug 2011 09:51:06 +0200 Subject: [csw-maintainers] Fwd: CSW catalog problems (2011-08-30 07:25) In-Reply-To: References: <201108300525.p7U5PC2e015378@ajax2.int.lio.se> Message-ID: 2011/8/30 Maciej Blizi?ski : > Hm. None of these two exist in the catalog as files: > > maciej at login [login]:~ > ls /home/mirror/opencsw-future/unstable/*/*/krb5kdc* > /home/mirror/opencsw-future/unstable/*/*/krb5kdc*: No such file or directory > maciej at login [login]:~ > ls /home/mirror/opencsw-future/unstable/*/*/krb5user* > /home/mirror/opencsw-future/unstable/*/*/krb5user*: No such file or directory krb5kdc is 1.4.4 from 2006 so it should probably not be there but krb5user is 1.9.1 from 2011 so that should be there. From maciej at opencsw.org Tue Aug 30 10:03:58 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Tue, 30 Aug 2011 09:03:58 +0100 Subject: [csw-maintainers] Fwd: CSW catalog problems (2011-08-30 07:25) In-Reply-To: References: <201108300525.p7U5PC2e015378@ajax2.int.lio.se> Message-ID: 2011/8/30 Peter Bonivart : > 2011/8/30 Maciej Blizi?ski : >> Hm. None of these two exist in the catalog as files: >> >> maciej at login [login]:~ > ls /home/mirror/opencsw-future/unstable/*/*/krb5kdc* >> /home/mirror/opencsw-future/unstable/*/*/krb5kdc*: No such file or directory >> maciej at login [login]:~ > ls /home/mirror/opencsw-future/unstable/*/*/krb5user* >> /home/mirror/opencsw-future/unstable/*/*/krb5user*: No such file or directory > > krb5kdc is 1.4.4 from 2006 so it should probably not be there but > krb5user is 1.9.1 from 2011 so that should be there. I found out what was the problem - it's my fault. Fixing now. Checking files on disk was silly - I should have been checking the database directly. I started uploading new packages with the intention to remove both krb5kdc and krb5user. At some point, I saw that I'm uploading a new version of krb5user, so I hit ctrl-C and then used csw-upload-pkg to remove the new krb5user from the catalog. But krb5user already replaced the 1.4.4 one in the i386/5.9 catalog, and not in other catalogs. When I ran csw-upload-pkg --remove, I effectively removed krb5user from i386/5.9 only. I'm now removing both krb5user and krb5kdc from the catalog. I need to do this by also specifying --os-release, because the old package is for 5.8, and if I didn't specify that I want to remove that package from 5.9/5.10/5.11 only, csw-upload-pkg would have removed it from 5.8 too. maciej at login [login]:~ > csw-upload-pkg --remove --os-release SunOS5.9 /home/mirror/opencsw-future/allpkgs/krb5_kdc-1.4.4\,REV\=2006.12.27-SunOS5.8- krb5_kdc-1.4.4,REV=2006.12.27-SunOS5.8-i386-CSW.pkg.gz krb5_kdc-1.4.4,REV=2006.12.27-SunOS5.8-sparc-CSW.pkg.gz maciej at login [login]:~ > csw-upload-pkg --remove --os-release SunOS5.9 /home/mirror/opencsw-future/allpkgs/krb5_kdc-1.4.4\,REV\=2006.12.27-SunOS5.8-* Removing krb5_kdc (i386 SunOS5.8) from catalog unstable i386 SunOS5.9 Removing krb5_kdc (sparc SunOS5.8) from catalog unstable sparc SunOS5.9 maciej at login [login]:~ > csw-upload-pkg --remove --os-release SunOS5.10 /home/mirror/opencsw-future/allpkgs/krb5_kdc-1.4.4\,REV\=2006.12.27-SunOS5.8-* Removing krb5_kdc (i386 SunOS5.8) from catalog unstable i386 SunOS5.10 Removing krb5_kdc (sparc SunOS5.8) from catalog unstable sparc SunOS5.10 maciej at login [login]:~ > csw-upload-pkg --remove --os-release SunOS5.11 /home/mirror/opencsw-future/allpkgs/krb5_kdc-1.4.4\,REV\=2006.12.27-SunOS5.8-* Removing krb5_kdc (i386 SunOS5.8) from catalog unstable i386 SunOS5.11 Removing krb5_kdc (sparc SunOS5.8) from catalog unstable sparc SunOS5.11 maciej at login [login]:~ > csw-upload-pkg --remove --os-release SunOS5.9 /home/mirror/opencsw-future/allpkgs/krb5_user-1.4.4\,REV\=2006.12.27-SunOS5.8-* Removing krb5_user (i386 SunOS5.8) from catalog unstable i386 SunOS5.9 Removing krb5_user (sparc SunOS5.8) from catalog unstable sparc SunOS5.9 maciej at login [login]:~ > csw-upload-pkg --remove --os-release SunOS5.10 /home/mirror/opencsw-future/allpkgs/krb5_user-1.4.4\,REV\=2006.12.27-SunOS5.8-* Removing krb5_user (i386 SunOS5.8) from catalog unstable i386 SunOS5.10 Removing krb5_user (sparc SunOS5.8) from catalog unstable sparc SunOS5.10 maciej at login [login]:~ > csw-upload-pkg --remove --os-release SunOS5.11 /home/mirror/opencsw-future/allpkgs/krb5_user-1.4.4\,REV\=2006.12.27-SunOS5.8-* Removing krb5_user (i386 SunOS5.8) from catalog unstable i386 SunOS5.11 Removing krb5_user (sparc SunOS5.8) from catalog unstable sparc SunOS5.11 Maciej From maciej at opencsw.org Wed Aug 31 09:11:39 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 31 Aug 2011 08:11:39 +0100 Subject: [csw-maintainers] gpg key for signing In-Reply-To: <1313366233-sup-2282@pinkfloyd.chass.utoronto.ca> References: <1312938121-sup-613@pinkfloyd.chass.utoronto.ca> <1313366233-sup-2282@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/15 Ben Walton : > Excerpts from Ben Walton's message of Tue Aug 09 21:07:08 -0400 2011: > >> Any objections? > > Maciej and Ihsan, would you please generate a new key and install it > into the catalogsign account on cswsign in the buildfarm? ?I'll ensure > that you're notified when the passphrase times out. ?We should maybe > add a third person with the ability to reset the passphrase, but that > can be a separate discussion. I've generated and installed a new key. The key ID is 9306CC77. The key can be downloaded here: http://buildfarm.opencsw.org/~maciej/opencsw-signing-pub.asc I will distribute the passphrase so that Ben and Ihsan are also able to restart the signing daemon. When and how do we do the big move? Maciej From maciej at opencsw.org Wed Aug 31 10:21:51 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 31 Aug 2011 09:21:51 +0100 Subject: [csw-maintainers] gpg key for signing In-Reply-To: References: <1312938121-sup-613@pinkfloyd.chass.utoronto.ca> <1313366233-sup-2282@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/31 Maciej Blizi?ski : > The key can be downloaded here: > > http://buildfarm.opencsw.org/~maciej/opencsw-signing-pub.asc I've updated our mirrors page to include the new key: http://www.opencsw.org/get-it/mirrors/ Maciej From maciej at opencsw.org Wed Aug 31 16:47:56 2011 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_Blizi=C5=84ski?=) Date: Wed, 31 Aug 2011 15:47:56 +0100 Subject: [csw-maintainers] Reasons for gcc to have a separate prefix In-Reply-To: <1314317946-sup-2767@pinkfloyd.chass.utoronto.ca> References: <1314317946-sup-2767@pinkfloyd.chass.utoronto.ca> Message-ID: 2011/8/26 Ben Walton : > Excerpts from Maciej Blizi?ski's message of Thu Aug 25 07:23:26 -0400 2011: > > Hi Maciej, > >> Do you recall the reasons to compile gcc into a different prefix, >> and do you think these reasons still apply today? > > I'd definitely love to see gcc move into /opt/csw. ?Alternatives > should make this possible. ?A symlink from /opt/csw/gcc4 -> /opt/csw > should be provided if such a move is ever attempted though so that > things with the path coded in continue working. I've got now a working set of packages with binaries in /opt/csw and alternatives, just the way Peter Felecan has suggested. Currently, there's no /opt/csw/gcc4 compatibility layer. It's time to tackle this issue. I'm afraid that we cannot create the /opt/csw/gcc4 ? /opt/csw symlink. Imagine a scenario where someone puts a file under /opt/csw/gcc4 - pkgrm will not remove this directory. I'm not sure what happens when you try to install a symlink over a path which is already a directory, but I'm fairly certain that you won't make pkgadd happy that way. We'd better provide symlinks for individual files. For example: /opt/csw/gcc4/bin/gcc ? ../../bin/gcc-4.6 All other links accordingly. Does that look good? Maciej