From dam at opencsw.org Tue Oct 1 09:09:21 2013 From: dam at opencsw.org (Dagobert Michelsen) Date: Tue, 1 Oct 2013 09:09:21 +0200 Subject: [csw-maintainers] guile packaging Message-ID: Hi Peter, the upcoming GNU make will come with optional guile support and while looking at the guile packaging I noticed that CSWlibguile2-0-22 only contains the library itself. From my understanding (which is far from knowledgable in terms of guile) the files in /opt/csw/lib/guile are also needed when using the library. If this is really the case these files should be relocated to the library package. Thoughts? 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 -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 2351 bytes Desc: not available URL: From pfelecan at opencsw.org Tue Oct 1 09:14:25 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Tue, 01 Oct 2013 09:14:25 +0200 Subject: [csw-maintainers] guile packaging In-Reply-To: (Dagobert Michelsen's message of "Tue, 1 Oct 2013 09:09:21 +0200") References: Message-ID: Dagobert Michelsen writes: > Hi Peter, > > the upcoming GNU make will come with optional guile support and while looking at the guile > packaging I noticed that CSWlibguile2-0-22 only contains the library itself. From my > understanding (which is far from knowledgable in terms of guile) the files in /opt/csw/lib/guile > are also needed when using the library. If this is really the case these files should > be relocated to the library package. Thoughts? You're probably right. I'll explore and act results wise. -- Peter From markp at opencsw.org Tue Oct 1 15:38:12 2013 From: markp at opencsw.org (Mark Phillips) Date: Tue, 1 Oct 2013 14:38:12 +0100 Subject: [csw-maintainers] It's time to give up Puppet Message-ID: Hey Folks, Would anybody like to pick up ownership of the CSW puppet & puppet3 (+master) packages? I haven't been using Solaris for a good year or so now, and I no longer have any Solaris hosts myself. So I can't properly test this - or, more honestly, I don't have the drive to test properly ;-) Generally speaking repacking each time a new release comes out is a doddle - the gar builds work a treat. If nobody from the immediate CSW community steps forward, I'll throw it out to puppet-users to see if anybody there would like to take it. Cheers, --Mark -------------- next part -------------- An HTML attachment was scrubbed... URL: From pfelecan at opencsw.org Thu Oct 3 11:18:08 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Thu, 03 Oct 2013 11:18:08 +0200 Subject: [csw-maintainers] guile packaging In-Reply-To: (Dagobert Michelsen's message of "Tue, 1 Oct 2013 09:09:21 +0200") References: Message-ID: Dagobert Michelsen writes: > the upcoming GNU make will come with optional guile support and while > looking at the guile packaging I noticed that CSWlibguile2-0-22 only > contains the library itself. From my understanding (which is far from > knowledgable in terms of guile) the files in /opt/csw/lib/guile are > also needed when using the library. If this is really the case these > files should be relocated to the library package. Thoughts? There is a new release of the guile packages. The main feature is the relocation of the relevant component in the library package. Now you're ready for the new GNU Make... -- Peter From pfelecan at opencsw.org Thu Oct 3 13:57:50 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Thu, 03 Oct 2013 13:57:50 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: ("Maciej (Matchek) =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Fri, 27 Sep 2013 09:45:59 +0100") References:

Message-ID: I started the build-farm documentation verification and made some corrections and clarification, visible on the devel mailing list. Unfortunately, for the moment, I'm stuck with an issue in libssl: https://www.opencsw.org/mantis/view.php?id=5112 -- Peter From dam at opencsw.org Thu Oct 3 14:43:04 2013 From: dam at opencsw.org (Dagobert Michelsen) Date: Thu, 3 Oct 2013 14:43:04 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

Message-ID: Hi Peter, Am 03.10.2013 um 13:57 schrieb Peter FELECAN : > I started the build-farm documentation verification and made some > corrections and clarification, visible on the devel mailing list. > > Unfortunately, for the moment, I'm stuck with an issue in libssl: > https://www.opencsw.org/mantis/view.php?id=5112 Looks like your emulation has an quite old processor. Can you powerup the emulation for some newer isa? 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 -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 2351 bytes Desc: not available URL: From pfelecan at opencsw.org Thu Oct 3 14:51:43 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Thu, 03 Oct 2013 14:51:43 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: (Dagobert Michelsen's message of "Thu, 3 Oct 2013 14:43:04 +0200") References:

Message-ID: Dagobert Michelsen writes: > Hi Peter, > > Am 03.10.2013 um 13:57 schrieb Peter FELECAN : >> I started the build-farm documentation verification and made some >> corrections and clarification, visible on the devel mailing list. >> >> Unfortunately, for the moment, I'm stuck with an issue in libssl: >> https://www.opencsw.org/mantis/view.php?id=5112 > > Looks like your emulation has an quite old processor. Can you powerup > the emulation for some newer isa? This is not an emulation but real "metal". No so old either, well not in real historical time ... 10 years. BTW, it supports SSE, see http://en.wikipedia.org/wiki/AMD_Athlon_XP_3200%2B#Athlon_XP_.22Barton.22_.28Model_10.2C_130_nm.29 , consequently I don't get what the message conveys. -- Peter From pfelecan at opencsw.org Thu Oct 3 15:14:06 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Thu, 03 Oct 2013 15:14:06 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: (Peter FELECAN's message of "Thu, 03 Oct 2013 13:57:50 +0200") References:

Message-ID: Why do we install, even optionally, the following: - DB2 client - IBM Informix Client SDK Are they needed for a feature of the build-farm? -- Peter From dam at opencsw.org Thu Oct 3 15:19:35 2013 From: dam at opencsw.org (Dagobert Michelsen) Date: Thu, 3 Oct 2013 15:19:35 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

Message-ID: <64A9C94C-5CC7-4A3C-A6AE-43FEEFDD2153@opencsw.org> Hi Peter, Am 03.10.2013 um 15:14 schrieb Peter FELECAN : > Why do we install, even optionally, the following: > > - DB2 client > - IBM Informix Client SDK There is also the Oracle Instantclient. > Are they needed for a feature of the build-farm? Yes, I built DBD Perl modules for all of these. They are not super-important, I just added the how-to-install as a reminder for me and others who want to redo the work. If you don't intend to rebuilt the respective Perl modules you can skip the install. 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 -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 2351 bytes Desc: not available URL: From laurent at opencsw.org Thu Oct 3 15:21:13 2013 From: laurent at opencsw.org (Laurent Blume) Date: Thu, 03 Oct 2013 15:21:13 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

Message-ID: <524D6F49.1060506@opencsw.org> On 03/10/13 14:51, Peter FELECAN wrote: > This is not an emulation but real "metal". No so old either, well not in > real historical time ... 10 years. BTW, it supports SSE, see > http://en.wikipedia.org/wiki/AMD_Athlon_XP_3200%2B#Athlon_XP_.22Barton.22_.28Model_10.2C_130_nm.29 > , consequently I don't get what the message conveys. > It supports SSE, but it is really obsolete and does not support SSe2, and that library needs it: $ file /opt/csw/lib/libcrypto.so.1.0.0 /opt/csw/lib/libcrypto.so.1.0.0: ELF 32-bit LSB dynamic lib 80386 Version 1 [SSE2 SSE MMX CMOV FPU], dynamically linked, stripped I think that's probably a mistake, since so far, I've seen even SSE avoided for the 32 bit binaries. Laurent From dam at opencsw.org Thu Oct 3 15:26:05 2013 From: dam at opencsw.org (Dagobert Michelsen) Date: Thu, 3 Oct 2013 15:26:05 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: <524D6F49.1060506@opencsw.org> References:

<524D6F49.1060506@opencsw.org> Message-ID: <813F7CAE-7DEB-4127-9309-01DE49D8A557@opencsw.org> Hi folks, Am 03.10.2013 um 15:21 schrieb Laurent Blume : > On 03/10/13 14:51, Peter FELECAN wrote: >> This is not an emulation but real "metal". No so old either, well not in >> real historical time ... 10 years. BTW, it supports SSE, see >> http://en.wikipedia.org/wiki/AMD_Athlon_XP_3200%2B#Athlon_XP_.22Barton.22_.28Model_10.2C_130_nm.29 >> , consequently I don't get what the message conveys. >> > > It supports SSE, but it is really obsolete and does not support SSe2, and that library needs it: > $ file /opt/csw/lib/libcrypto.so.1.0.0 > /opt/csw/lib/libcrypto.so.1.0.0: ELF 32-bit LSB dynamic lib 80386 Version 1 [SSE2 SSE MMX CMOV FPU], dynamically linked, stripped > > I think that's probably a mistake, since so far, I've seen even SSE avoided for the 32 bit binaries. Or if necessary we could build more hwcap versions: http://www.oracle.com/technetwork/server-storage/solaris/hwcap-modification-139536.html The needed aux-linkage has already been done for full-features versions of neon and curl. 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 -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 2351 bytes Desc: not available URL: From laurent at opencsw.org Thu Oct 3 15:27:45 2013 From: laurent at opencsw.org (Laurent Blume) Date: Thu, 03 Oct 2013 15:27:45 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: <813F7CAE-7DEB-4127-9309-01DE49D8A557@opencsw.org> References:

<524D6F49.1060506@opencsw.org> <813F7CAE-7DEB-4127-9309-01DE49D8A557@opencsw.org> Message-ID: <524D70D1.6000807@opencsw.org> On 03/10/13 15:26, Dagobert Michelsen wrote: > Or if necessary we could build more hwcap versions: The best way to go :-) From pfelecan at opencsw.org Thu Oct 3 15:30:32 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Thu, 03 Oct 2013 15:30:32 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: <813F7CAE-7DEB-4127-9309-01DE49D8A557@opencsw.org> (Dagobert Michelsen's message of "Thu, 3 Oct 2013 15:26:05 +0200") References:

<524D6F49.1060506@opencsw.org> <813F7CAE-7DEB-4127-9309-01DE49D8A557@opencsw.org> Message-ID: Dagobert Michelsen writes: > Hi folks, > > Am 03.10.2013 um 15:21 schrieb Laurent Blume : >> On 03/10/13 14:51, Peter FELECAN wrote: >>> This is not an emulation but real "metal". No so old either, well not in >>> real historical time ... 10 years. BTW, it supports SSE, see >>> http://en.wikipedia.org/wiki/AMD_Athlon_XP_3200%2B#Athlon_XP_.22Barton.22_.28Model_10.2C_130_nm.29 >>> , consequently I don't get what the message conveys. >>> >> >> It supports SSE, but it is really obsolete and does not support SSe2, and that library needs it: >> $ file /opt/csw/lib/libcrypto.so.1.0.0 >> /opt/csw/lib/libcrypto.so.1.0.0: ELF 32-bit LSB dynamic lib 80386 Version 1 [SSE2 SSE MMX CMOV FPU], dynamically linked, stripped >> >> I think that's probably a mistake, since so far, I've seen even SSE avoided for the 32 bit binaries. > > > Or if necessary we could build more hwcap versions: > http://www.oracle.com/technetwork/server-storage/solaris/hwcap-modification-139536.html > The needed aux-linkage has already been done for full-features versions of > neon and curl. Thank you for my enlightenment folks. I hope that this will be corrected either by not requiring SSE2 or by hwcap. For the moment I'm not investing in bare metal just for the sake of build-farm documentation verification. -- Peter From pfelecan at opencsw.org Thu Oct 3 15:33:00 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Thu, 03 Oct 2013 15:33:00 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: <64A9C94C-5CC7-4A3C-A6AE-43FEEFDD2153@opencsw.org> (Dagobert Michelsen's message of "Thu, 3 Oct 2013 15:19:35 +0200") References:

<64A9C94C-5CC7-4A3C-A6AE-43FEEFDD2153@opencsw.org> Message-ID: Dagobert Michelsen writes: > Hi Peter, > > Am 03.10.2013 um 15:14 schrieb Peter FELECAN : >> Why do we install, even optionally, the following: >> >> - DB2 client >> - IBM Informix Client SDK > > There is also the Oracle Instantclient. > >> Are they needed for a feature of the build-farm? > > Yes, I built DBD Perl modules for all of these. They are not super-important, > I just added the how-to-install as a reminder for me and others who want > to redo the work. If you don't intend to rebuilt the respective Perl modules > you can skip the install. Alright. Not installing avoids the capability to rebuild the totality of the stack which is, I remind you, one of the purposes of the build-farm documentation verification. Can you, please, provide a little bit more information about these additional components, e.g. from where to get the packages, &c. -- Peter From pfelecan at opencsw.org Thu Oct 3 15:40:56 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Thu, 03 Oct 2013 15:40:56 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: <524D70D1.6000807@opencsw.org> (Laurent Blume's message of "Thu, 03 Oct 2013 15:27:45 +0200") References:

<524D6F49.1060506@opencsw.org> <813F7CAE-7DEB-4127-9309-01DE49D8A557@opencsw.org> <524D70D1.6000807@opencsw.org> Message-ID: Laurent Blume writes: > On 03/10/13 15:26, Dagobert Michelsen wrote: >> Or if necessary we could build more hwcap versions: > > The best way to go :-) Well, after reading the documentation I'm not sure that work wise is the easier. Finally, it depends on the resources than Yann wishes to allocate to this, if any. -- Peter From yann at pleiades.fr.eu.org Thu Oct 3 15:41:31 2013 From: yann at pleiades.fr.eu.org (Yann Rouillard) Date: Thu, 3 Oct 2013 15:41:31 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: <524D6F49.1060506@opencsw.org> References:

<524D6F49.1060506@opencsw.org> Message-ID: openssl has it own build system with specific build flags, that must be the reason why it requires SSE2 and SSE and not most other binaries. I will look for the options at fault. But BTW what is our policy concerning enabling of SSE/SSE2 ? I didn't have a complain so far about it and I suspect that this set of instructions helps to speed up some cryptographic codes in openssl, so I wonder if this is a good idea to remove them. Yann 2013/10/3 Laurent Blume > On 03/10/13 14:51, Peter FELECAN wrote: > >> This is not an emulation but real "metal". No so old either, well not in >> real historical time ... 10 years. BTW, it supports SSE, see >> http://en.wikipedia.org/wiki/**AMD_Athlon_XP_3200%2B#Athlon_** >> XP_.22Barton.22_.28Model_10.**2C_130_nm.29 >> , consequently I don't get what the message conveys. >> >> > It supports SSE, but it is really obsolete and does not support SSe2, and > that library needs it: > $ file /opt/csw/lib/libcrypto.so.1.0.**0 > /opt/csw/lib/libcrypto.so.1.0.**0: ELF 32-bit LSB dynamic lib > 80386 Version 1 [SSE2 SSE MMX CMOV FPU], dynamically linked, stripped > > I think that's probably a mistake, since so far, I've seen even SSE > avoided for the 32 bit binaries. > > Laurent > > ______________________________**_________________ > 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 dam at opencsw.org Thu Oct 3 15:45:06 2013 From: dam at opencsw.org (Dagobert Michelsen) Date: Thu, 3 Oct 2013 15:45:06 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

<524D6F49.1060506@opencsw.org> Message-ID: <2E5E8148-EEEF-40CE-97F7-6CA601149C3E@opencsw.org> Hi Yann, Am 03.10.2013 um 15:41 schrieb Yann Rouillard : > openssl has it own build system with specific build flags, that must be the reason why it requires SSE2 and SSE and not most other binaries. > I will look for the options at fault. > > But BTW what is our policy concerning enabling of SSE/SSE2 ? I guess we don't have one yet. > I didn't have a complain so far about it and I suspect that this set of instructions helps to speed up some cryptographic codes in openssl, so I wonder if this is a good idea to remove them. Adding another ISA as option will probably take quite some time. Peter, do you think your use case is common enough to justify the extra work for Yann? Would using a VM on a more modern machine be also possible? Best regards -- Dago > > Yann > > > > > 2013/10/3 Laurent Blume > On 03/10/13 14:51, Peter FELECAN wrote: > This is not an emulation but real "metal". No so old either, well not in > real historical time ... 10 years. BTW, it supports SSE, see > http://en.wikipedia.org/wiki/AMD_Athlon_XP_3200%2B#Athlon_XP_.22Barton.22_.28Model_10.2C_130_nm.29 > , consequently I don't get what the message conveys. > > > It supports SSE, but it is really obsolete and does not support SSe2, and that library needs it: > $ file /opt/csw/lib/libcrypto.so.1.0.0 > /opt/csw/lib/libcrypto.so.1.0.0: ELF 32-bit LSB dynamic lib 80386 Version 1 [SSE2 SSE MMX CMOV FPU], dynamically linked, stripped > > I think that's probably a mistake, since so far, I've seen even SSE avoided for the 32 bit binaries. > > Laurent > > _______________________________________________ > maintainers mailing list > maintainers at lists.opencsw.org > https://lists.opencsw.org/mailman/listinfo/maintainers > .:: This mailing list's archive is public. ::. > > _______________________________________________ > maintainers mailing list > maintainers at lists.opencsw.org > https://lists.opencsw.org/mailman/listinfo/maintainers > .:: This mailing list's archive is public. ::. -- "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 -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 2351 bytes Desc: not available URL: From yann at pleiades.fr.eu.org Thu Oct 3 15:46:18 2013 From: yann at pleiades.fr.eu.org (Yann Rouillard) Date: Thu, 3 Oct 2013 15:46:18 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: <813F7CAE-7DEB-4127-9309-01DE49D8A557@opencsw.org> References:

<524D6F49.1060506@opencsw.org> <813F7CAE-7DEB-4127-9309-01DE49D8A557@opencsw.org> Message-ID: > > Or if necessary we could build more hwcap versions: > > http://www.oracle.com/technetwork/server-storage/solaris/hwcap-modification-139536.html > The needed aux-linkage has already been done for full-features versions of > neon and curl. > > Great ! Yet another modulation for openssl ! :) I will have a look at neon and curl to see how it's done. Yann -------------- next part -------------- An HTML attachment was scrubbed... URL: From dam at opencsw.org Thu Oct 3 15:51:48 2013 From: dam at opencsw.org (Dagobert Michelsen) Date: Thu, 3 Oct 2013 15:51:48 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

<524D6F49.1060506@opencsw.org> <813F7CAE-7DEB-4127-9309-01DE49D8A557@opencsw.org> Message-ID: <7C37D6D0-BBA4-47F0-887A-94BE2313635F@opencsw.org> Hi Yann, Am 03.10.2013 um 15:46 schrieb Yann Rouillard : >> Or if necessary we could build more hwcap versions: >> http://www.oracle.com/technetwork/server-storage/solaris/hwcap-modification-139536.html >> The needed aux-linkage has already been done for full-features versions of >> neon and curl. > > Great ! Yet another modulation for openssl ! :) > I will have a look at neon and curl to see how it's done. Well, it is quite ugly. The idea is to make one build with simple dependencies and another with more dependencies. Then on post-build remove the respective library which should have the AUX filter, add the flag and redo the build. Make then picks up just the missing files and rebuilds them with the additional flag. If there were libraries the removal-set-compile would also have to be done several times: http://sourceforge.net/apps/trac/gar/browser/csw/mgar/pkg/libcurl4/trunk/Makefile http://sourceforge.net/apps/trac/gar/browser/csw/mgar/pkg/libneon27/trunk/Makefile It is incredibly ugly, but after fiddling for probably half a year and trying different patch methods this was by far the least untrusive, easiest to understand and most flexible. 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 -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 2351 bytes Desc: not available URL: From pfelecan at opencsw.org Thu Oct 3 15:57:31 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Thu, 03 Oct 2013 15:57:31 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: <2E5E8148-EEEF-40CE-97F7-6CA601149C3E@opencsw.org> (Dagobert Michelsen's message of "Thu, 3 Oct 2013 15:45:06 +0200") References:

<524D6F49.1060506@opencsw.org> <2E5E8148-EEEF-40CE-97F7-6CA601149C3E@opencsw.org> Message-ID: Dagobert Michelsen writes: > Hi Yann, > > Am 03.10.2013 um 15:41 schrieb Yann Rouillard : >> openssl has it own build system with specific build flags, that must be the reason why it requires SSE2 and SSE and not most other binaries. >> I will look for the options at fault. >> >> But BTW what is our policy concerning enabling of SSE/SSE2 ? > > I guess we don't have one yet. Currently we use as a default Pentium Pro. Maybe it's not adequate. But, I'm not aware, for the moment, of the consequences. >> I didn't have a complain so far about it and I suspect that this set >> of instructions helps to speed up some cryptographic codes in >> openssl, so I wonder if this is a good idea to remove them. > > Adding another ISA as option will probably take quite some time. > Peter, do you think your use case is common enough to justify the > extra work for Yann? Would using a VM on a more modern machine > be also possible? Using a VM, even on modern machine, has it's own inconvenients and, frankly, who can imagine a build farm on VMs? It requires bare metal for a lot of reasons. -- Peter From dam at opencsw.org Thu Oct 3 16:01:46 2013 From: dam at opencsw.org (Dagobert Michelsen) Date: Thu, 3 Oct 2013 16:01:46 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

<524D6F49.1060506@opencsw.org> <2E5E8148-EEEF-40CE-97F7-6CA601149C3E@opencsw.org> Message-ID: <15166493-F769-4CCF-AF1A-7D14082771EE@opencsw.org> Hi Peter, Am 03.10.2013 um 15:57 schrieb Peter FELECAN : > Dagobert Michelsen writes: >> Adding another ISA as option will probably take quite some time. >> Peter, do you think your use case is common enough to justify the >> extra work for Yann? Would using a VM on a more modern machine >> be also possible? > > Using a VM, even on modern machine, has it's own inconvenients and, > frankly, who can imagine a build farm on VMs? It requires bare metal for > a lot of reasons. Umh, the buildfarm you use runs on VMs ;-) That include all *x hosts. What reasons do you have in mind preventing running a farm on virtualized environments? 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 -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 2351 bytes Desc: not available URL: From pfelecan at opencsw.org Thu Oct 3 16:12:34 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Thu, 03 Oct 2013 16:12:34 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: <15166493-F769-4CCF-AF1A-7D14082771EE@opencsw.org> (Dagobert Michelsen's message of "Thu, 3 Oct 2013 16:01:46 +0200") References:

<524D6F49.1060506@opencsw.org> <2E5E8148-EEEF-40CE-97F7-6CA601149C3E@opencsw.org> <15166493-F769-4CCF-AF1A-7D14082771EE@opencsw.org> Message-ID: Dagobert Michelsen writes: > Hi Peter, > > Am 03.10.2013 um 15:57 schrieb Peter FELECAN : >> Dagobert Michelsen writes: >>> Adding another ISA as option will probably take quite some time. >>> Peter, do you think your use case is common enough to justify the >>> extra work for Yann? Would using a VM on a more modern machine >>> be also possible? >> >> Using a VM, even on modern machine, has it's own inconvenients and, >> frankly, who can imagine a build farm on VMs? It requires bare metal for >> a lot of reasons. > > Umh, the buildfarm you use runs on VMs ;-) That include all *x hosts. > What reasons do you have in mind preventing running a farm on > virtualized environments? I know that. And suffer from that, don't get me wrong but at least on i386 the build times are often one order of magnitude slower than on my similar bare metal on which, of course, I'm alone. It's often more easier to get a bunch of oldish bare metal than have a server capable of well supporting VMs at a reasonable performance. And I'm not even mentioning SPARC bare metal versus virtualization for which I have not enough experience. Anyhow, the point is not this, the point is that we have stuff which doesn't run on the lowest denominator that we support and that can be quite surprising, even for me. -- Peter From dam at opencsw.org Thu Oct 3 19:27:13 2013 From: dam at opencsw.org (Dagobert Michelsen) Date: Thu, 3 Oct 2013 19:27:13 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

<524D6F49.1060506@opencsw.org> <2E5E8148-EEEF-40CE-97F7-6CA601149C3E@opencsw.org> <15166493-F769-4CCF-AF1A-7D14082771EE@opencsw.org> Message-ID: Hi Peter, Am 03.10.2013 um 16:12 schrieb Peter FELECAN : > Dagobert Michelsen writes: >> Am 03.10.2013 um 15:57 schrieb Peter FELECAN : >>> Dagobert Michelsen writes: >>>> Adding another ISA as option will probably take quite some time. >>>> Peter, do you think your use case is common enough to justify the >>>> extra work for Yann? Would using a VM on a more modern machine >>>> be also possible? >>> >>> Using a VM, even on modern machine, has it's own inconvenients and, >>> frankly, who can imagine a build farm on VMs? It requires bare metal for >>> a lot of reasons. >> >> Umh, the buildfarm you use runs on VMs ;-) That include all *x hosts. >> What reasons do you have in mind preventing running a farm on >> virtualized environments? > > I know that. And suffer from that, don't get me wrong but at least on > i386 the build times are often one order of magnitude slower than on my > similar bare metal on which, of course, I'm alone. It's often more easier > to get a bunch of oldish bare metal than have a server capable of > well supporting VMs at a reasonable performance. And I'm not even > mentioning SPARC bare metal versus virtualization for which I have not > enough experience. Ah yes, I can understand that, especially that you are often working on very large sets of packages like texlive or qt4. > Anyhow, the point is not this, the point is that we have stuff which > doesn't run on the lowest denominator that we support and that can be > quite surprising, even for me. This is also very true. IIRC we raised the general build level to PentiumPro, but as we didn't have negative feedback on openssl before I would like to hear some more feedback. Maybe we can have some more thoughts on the base cpu level for x86 from other maintainers? 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 -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 2351 bytes Desc: not available URL: From maciej at opencsw.org Thu Oct 3 19:57:24 2013 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_=28Matchek=29_Blizi=C5=84ski?=) Date: Thu, 3 Oct 2013 18:57:24 +0100 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

Message-ID: 2013/10/3 Peter FELECAN > I started the build-farm documentation verification and made some > corrections and clarification, visible on the devel mailing list. Very cool. It's a long document and it'll take us a number of iterations to make it semi-right. As far as 'sudo' - I assumed that the instructions would be performed as an unprivileged user. Otherwise people might think that these instructions are meant for the root user, while they're really not. What do you think? Maciej From pfelecan at opencsw.org Fri Oct 4 09:26:12 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Fri, 04 Oct 2013 09:26:12 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: ("Maciej (Matchek) =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Thu, 3 Oct 2013 18:57:24 +0100") References:

Message-ID: "Maciej (Matchek) Blizi?ski" writes: > 2013/10/3 Peter FELECAN >> I started the build-farm documentation verification and made some >> corrections and clarification, visible on the devel mailing list. > > Very cool. It's a long document and it'll take us a number of > iterations to make it semi-right. Oh, I prefer to make it at least semi-left. That is, I prefer the laevogyration. > As far as 'sudo' - I assumed that the instructions would be performed > as an unprivileged user. Otherwise people might think that these > instructions are meant for the root user, while they're really not. > What do you think? I understood the sudo usage but there is not the slightest indication about how to activate it. I know that Google is your friend, at least when it's not your employer. What about using the discreet and quai-subliminal hints of # and $ ? -- Peter From maciej at opencsw.org Fri Oct 4 11:11:32 2013 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_=28Matchek=29_Blizi=C5=84ski?=) Date: Fri, 4 Oct 2013 10:11:32 +0100 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

Message-ID: 2013/10/4 Peter FELECAN > > > As far as 'sudo' - I assumed that the instructions would be performed > > as an unprivileged user. Otherwise people might think that these > > instructions are meant for the root user, while they're really not. > > What do you think? > > I understood the sudo usage but there is not the slightest indication > about how to activate it. What about this? https://sourceforge.net/apps/trac/gar/browser/csw/mgar/pkg/opencsw-manual/trunk/files/for-maintainers/buildfarm-setup.rst#L47 > I know that Google is your friend, at least > when it's not your employer. What about using the discreet and > quai-subliminal hints of # and $ ? I'm not a fan. With Sphinx it's even worse, because it thinks that # is a comment and screws up the rest of the line; Sphinx highlights shell syntax and PS1 is not part of it, so you can't blame Sphinx. If you insist on including instructions how to set up sudo, you can copy the relevant snippet from my Usable Solaris howto. http://usable-solaris.googlecode.com/svn/trunk/docs/solaris-10-preliminary-setup.html#_regular_user_setup YOUR_REGULAR_USER=joe pkgutil -y -i sudo echo "${YOUR_REGULAR_USER} ALL =(ALL) ALL" >> /etc/opt/csw/sudoers chmod 0440 /etc/opt/csw/sudoers Maciej From pfelecan at opencsw.org Fri Oct 4 15:00:46 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Fri, 04 Oct 2013 15:00:46 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: ("Maciej (Matchek) =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Thu, 3 Oct 2013 18:57:24 +0100") References:

Message-ID: I "unstuck" myself from the crypto SSE2 issue by creating a more up to date package and installing it. Now I'm attacking the "checkpkg" section and I'm discovering with great displeasure what a mess is to start using MySQL from our unstable distribution! Even if you have experience in setting up this beast it is a travel through a fuzzy labyrinth finding out all the places and components. I'll try to fill issues in our bug tracker as much as I can without distracting me from the main goal and not shaving the yak... What's the most convenient template for the data base server ? Small, medium or large? Maybe some hints on what's the procedure on enabling MySQL on a pristine system? -- Peter From maciej at opencsw.org Fri Oct 4 15:08:11 2013 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_=28Matchek=29_Blizi=C5=84ski?=) Date: Fri, 4 Oct 2013 14:08:11 +0100 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

Message-ID: 2013/10/4 Peter FELECAN : > What's the most convenient template for the data base server ? Small, > medium or large? Maybe some hints on what's the procedure on enabling > MySQL on a pristine system? I don't think you need to do any of that. The only thing you need is the max_allowed_packet setting, you can just run these two commands: echo >>/etc/opt/csw/my.cnf "[mysqld]" echo >>/etc/opt/csw/my.cnf "max_allowed_packet=64M" The rest is taken care of by the postinstall script. The database daemon is running and you can connect as the root user. Maciej From pfelecan at opencsw.org Fri Oct 4 15:18:56 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Fri, 04 Oct 2013 15:18:56 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: ("Maciej (Matchek) =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Fri, 4 Oct 2013 14:08:11 +0100") References:

Message-ID: "Maciej (Matchek) Blizi?ski" writes: > 2013/10/4 Peter FELECAN : >> What's the most convenient template for the data base server ? Small, >> medium or large? Maybe some hints on what's the procedure on enabling >> MySQL on a pristine system? > > I don't think you need to do any of that. The only thing you need is > the max_allowed_packet setting, you can just run these two commands: > > echo >>/etc/opt/csw/my.cnf "[mysqld]" > echo >>/etc/opt/csw/my.cnf "max_allowed_packet=64M" > > The rest is taken care of by the postinstall script. The database > daemon is running and you can connect as the root user. It is not taken care. This is why I'm whining: # svcs -l svc:/network/cswmysql5:default fmri svc:/network/cswmysql5:default enabled false state disabled next_state none state_time Fri Oct 04 12:17:34 2013 restarter svc:/system/svc/restarter:default dependency require_all/none svc:/system/filesystem/local (online) dependency require_all/none svc:/network/loopback (online) bash-3.2# svcs -l svc:/network/cswmysql5:default fmri svc:/network/cswmysql5:default enabled false state disabled next_state none state_time Fri Oct 04 12:17:34 2013 restarter svc:/system/svc/restarter:default dependency require_all/none svc:/system/filesystem/local (online) dependency require_all/none svc:/network/loopback (online) BTW, can you tell me if the template to create /etc/opt/csw/my.cnf as required for checkpkg is the small, medium or large. TIA -- Peter From maciej at opencsw.org Fri Oct 4 15:26:37 2013 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_=28Matchek=29_Blizi=C5=84ski?=) Date: Fri, 4 Oct 2013 14:26:37 +0100 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

Message-ID: 2013/10/4 Peter FELECAN > It is not taken care. This is why I'm whining: > > # svcs -l svc:/network/cswmysql5:default > fmri svc:/network/cswmysql5:default > enabled false > state disabled Hm. Maybe you need to something to start it... but I've done about 3 fresh installations in the last half a year and don't recall anything MySQL related, so the installation must have been easy for me. Maybe you need to run something like 1 command to set up the database files and then svc enable. We can include that in the instructions. > BTW, can you tell me if the template to create /etc/opt/csw/my.cnf as > required for checkpkg is the small, medium or large. No template. Start with empty file, add two lines as instructed, and you're done. Maciej From pfelecan at opencsw.org Fri Oct 4 17:58:42 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Fri, 04 Oct 2013 17:58:42 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: ("Maciej (Matchek) =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Fri, 4 Oct 2013 14:26:37 +0100") References:

Message-ID: "Maciej (Matchek) Blizi?ski" writes: > 2013/10/4 Peter FELECAN >> It is not taken care. This is why I'm whining: >> >> # svcs -l svc:/network/cswmysql5:default >> fmri svc:/network/cswmysql5:default >> enabled false >> state disabled > > Hm. Maybe you need to something to start it... but I've done about 3 > fresh installations in the last half a year and don't recall anything > MySQL related, so the installation must have been easy for me. Maybe > you need to run something like 1 command to set up the database files > and then svc enable. We can include that in the instructions. > >> BTW, can you tell me if the template to create /etc/opt/csw/my.cnf as >> required for checkpkg is the small, medium or large. > > No template. Start with empty file, add two lines as instructed, and > you're done. The ball is rolling further... I documented the data base setup and now I'm stuck with the tables creation: $ pkgdb initdb ERROR:root:Could not create table : Row size too large. The maximum row size for the used table type, not counting BLOBs, is 65535. You have to change some columns to TEXT or BLOBs Traceback (most recent call last): File "/opt/csw/bin/pkgdb", line 709, in main() File "/opt/csw/bin/pkgdb", line 457, in main dbc.CreateTables() File "/opt/csw/lib/python/csw/database.py", line 164, in CreateTables table.createTable(ifNotExists=True) File "/opt/csw/lib/python/site-packages/sqlobject/main.py", line 1414, in createTable constraints = conn.createTable(cls) File "/opt/csw/lib/python/site-packages/sqlobject/dbconnection.py", line 510, in createTable self.query(createSql) File "/opt/csw/lib/python/site-packages/sqlobject/dbconnection.py", line 414, in query return self._runWithConnection(self._query, s) File "/opt/csw/lib/python/site-packages/sqlobject/dbconnection.py", line 327, in _runWithConnection val = meth(conn, *args) File "/opt/csw/lib/python/site-packages/sqlobject/dbconnection.py", line 411, in _query self._executeRetry(conn, conn.cursor(), s) File "/opt/csw/lib/python/site-packages/sqlobject/mysql/mysqlconnection.py", line 121, in _executeRetry raise OperationalError(ErrorMessage(e)) sqlobject.dberrors.OperationalError: Row size too large. The maximum row size for the used table type, not counting BLOBs, is 65535. You have to change some columns to TEXT or BLOBs I have seen hour note on the cswutils issue with missing dependencies. You need to state which is the correct context of usage for pkgdb. -- Peter From laurent at opencsw.org Fri Oct 4 18:44:55 2013 From: laurent at opencsw.org (Laurent Blume) Date: Fri, 04 Oct 2013 18:44:55 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

Message-ID: <524EF087.8020504@opencsw.org> On 04/10/2013 15:18, Peter FELECAN wrote: > It is not taken care. This is why I'm whining: Would you please explain what you are referring to? I saw your bug about some incorrect old changelog files, I'm cleaning that up. But I would not qualify that as ?major? in any way, as it's not preventing anything. Removing a couple of old files and changing the directories of others will wait until 5.5.35. As for the docs, everything MySQL provides is included, so feel free to open a bug upstream if that's a problem with you. Personnally, I use Google to search their website. > # svcs -l svc:/network/cswmysql5:default > fmri svc:/network/cswmysql5:default > enabled false > state disabled > next_state none > state_time Fri Oct 04 12:17:34 2013 > restarter svc:/system/svc/restarter:default > dependency require_all/none svc:/system/filesystem/local (online) > dependency require_all/none svc:/network/loopback (online) > bash-3.2# svcs -l svc:/network/cswmysql5:default > fmri svc:/network/cswmysql5:default > enabled false > state disabled > next_state none > state_time Fri Oct 04 12:17:34 2013 > restarter svc:/system/svc/restarter:default > dependency require_all/none svc:/system/filesystem/local (online) > dependency require_all/none svc:/network/loopback (online) I have no clue what's supposed to be wrong in those lines. Can you point it out? Thanks in advance, Laurent From maciej at opencsw.org Fri Oct 4 19:08:59 2013 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_=28Matchek=29_Blizi=C5=84ski?=) Date: Fri, 4 Oct 2013 18:08:59 +0100 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

Message-ID: 2013/10/4 Peter FELECAN > sqlobject.dberrors.OperationalError: Row size too large. The maximum row size for the used table type, not counting BLOBs, is 65535. You have to change some columns to TEXT or BLOBs Looks like the missing max_allowed_packet setting. > I have seen hour note on the cswutils issue with missing > dependencies. You need to state which is the correct context of usage > for pkgdb. This is done this way: cd .../path/to/gar/sources bin/pkgdb Maciej From pfelecan at opencsw.org Fri Oct 4 19:54:25 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Fri, 04 Oct 2013 19:54:25 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: ("Maciej (Matchek) =?utf-8?Q?Blizi=C5=84ski=22's?= message of "Fri, 4 Oct 2013 18:08:59 +0100") References:

Message-ID: "Maciej (Matchek) Blizi?ski" writes: > 2013/10/4 Peter FELECAN >> sqlobject.dberrors.OperationalError: Row size too large. The maximum >> row size for the used table type, not counting BLOBs, is 65535. You >> have to change some columns to TEXT or BLOBs > > Looks like the missing max_allowed_packet setting. But: $ cat /etc/opt/csw/my.cnf [mysqld] max_allowed_packet=64M how do you explain that? And, of course, the database server was started after the insertion of these lines. >> I have seen hour note on the cswutils issue with missing >> dependencies. You need to state which is the correct context of usage >> for pkgdb. > > This is done this way: > > cd .../path/to/gar/sources > bin/pkgdb It wasn't clear at all ant, BTW I dare to disagree with the idea that cswutils should not contain pkgdb and it must be run from mgar. The use case for which I care is when the database server is not the building machine. -- Peter From pfelecan at opencsw.org Fri Oct 4 20:06:22 2013 From: pfelecan at opencsw.org (Peter FELECAN) Date: Fri, 04 Oct 2013 20:06:22 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: <524EF087.8020504@opencsw.org> (Laurent Blume's message of "Fri, 04 Oct 2013 18:44:55 +0200") References:

<524EF087.8020504@opencsw.org> Message-ID: Laurent Blume writes: > On 04/10/2013 15:18, Peter FELECAN wrote: Unfortunately you omitted to cite/read what Maciej said: >> The rest is taken care of by the postinstall script. The database >> daemon is running and you can connect as the root user. >> It is not taken care. This is why I'm whining: > > Would you please explain what you are referring to? Aside of what Maciej incorrectly assume, i.e. that the postinstall script takes care of that, I'm refering to the incorrect message displayed by the postinstall script: ### The following messages are from /var/sadm/pkg/CSWmysql5/install/postinstall. See /opt/csw/mysql5/share/mysql/doc/README.CSW for packaging changes. Please ignore references to starting mysqld_safe in the messages above. These messages are from mysql_install_db. See the following for starting CSWmysql5. - the file /opt/csw/mysql5/share/mysql/doc/README.CSW does not exist - the message does not help someone who installs this, on the contrary it sows confusion > I saw your bug about > some incorrect old changelog files, I'm cleaning that up. Read again the issue's rapport. The changelog is a qualification for what I found in the documentation directory. In general the README.CSW contains specific instructions for configuring and running our package when things are different from what people get elsewhere. Those wo suypply a changelog just name it like that, i.e. ChangeLog. > But I would not qualify that as ?major? in any way, as it's not > preventing anything. Removing a couple of old files and changing the > directories of others will wait until 5.5.35. The qualification is subjective isn't it? I feel it as major because it's not only non useful but also misleading. > As for the docs, everything MySQL provides is included, so feel free to > open a bug upstream if that's a problem with you. Personnally, I use > Google to search their website. > >> # svcs -l svc:/network/cswmysql5:default >> fmri svc:/network/cswmysql5:default >> enabled false >> state disabled >> next_state none >> state_time Fri Oct 04 12:17:34 2013 >> restarter svc:/system/svc/restarter:default >> dependency require_all/none svc:/system/filesystem/local (online) >> dependency require_all/none svc:/network/loopback (online) >> bash-3.2# svcs -l svc:/network/cswmysql5:default >> fmri svc:/network/cswmysql5:default >> enabled false >> state disabled >> next_state none >> state_time Fri Oct 04 12:17:34 2013 >> restarter svc:/system/svc/restarter:default >> dependency require_all/none svc:/system/filesystem/local (online) >> dependency require_all/none svc:/network/loopback (online) > > I have no clue what's supposed to be wrong in those lines. Can you point > it out? Nothing. It was included only to show that the postinstall script has not take care of something. And I'm not saying that it should. -- Peter From laurent at opencsw.org Fri Oct 4 20:14:31 2013 From: laurent at opencsw.org (Laurent Blume) Date: Fri, 04 Oct 2013 20:14:31 +0200 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

<524EF087.8020504@opencsw.org> Message-ID: <524F0587.6010304@opencsw.org> On 04/10/2013 20:06, Peter FELECAN wrote: > Nothing. It was included only to show that the postinstall script has > not take care of something. And I'm not saying that it should. Whatever. Laurent From maciej at opencsw.org Fri Oct 4 21:27:06 2013 From: maciej at opencsw.org (=?UTF-8?Q?Maciej_=28Matchek=29_Blizi=C5=84ski?=) Date: Fri, 4 Oct 2013 20:27:06 +0100 Subject: [csw-maintainers] Buildfarm setup - documentation In-Reply-To: References:

Message-ID: