From dam at opencsw.org  Tue May  2 09:38:12 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Tue, 2 May 2017 09:38:12 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
Message-ID: <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>

Hi Riccarado,

Am 30.04.2017 um 19:44 schrieb Riccardo Mottola <rmottola at opencsw.org>:
> Jan Holzh?ter wrote:
>> It builds and works at least the tests I did.
>> It just needs update to new version and cross check with oracle pachtes.
>> I wrote I users list a few months/years? back and ask for tests.
>> Did not get much feedback.
>> So basically it just needs more tests (as the testsuite brakes).
>> AES-NI Intel support and Sparc T4+ support does work.
>> I tested that. (Which was broken with 1.0.1)
> 
> I imported some patches from oracle.
> 
> On solaris 10x, tests pass fine, on  solaris 10s, unfortunately they fail, as per the error I sent yesterday.
> 
> I re-tested and the older receipe of Jan passes also on 10s :( So new breakage
> 
> Riccardo
> 
> PS: currently the receipe is set to
> GARCOMPILER = SOS12U4
> 
> can I set something different for solaris9? or a value that suits both? I'd like to try there too but it is trying to use a compiler and a target that don't exist.

Sure, just use
  GARCOMPILER_5.9 = SOS12
  GARCOMPILER_5.10 = SUS12U4
  GARCOMPILER = $(GARCOMPILER_$(GAROSREL))


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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170502/3f3a845b/attachment.asc>

From rmottola at opencsw.org  Wed May  3 08:21:40 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Wed, 3 May 2017 08:21:40 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
Message-ID: <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>

Hi Dago,

Dagobert Michelsen wrote:
>> >can I set something different for solaris9? or a value that suits both? I'd like to try there too but it is trying to use a compiler and a target that don't exist.
> Sure, just use
>    GARCOMPILER_5.9 = SOS12
>    GARCOMPILER_5.10 = SUS12U4
>    GARCOMPILER = $(GARCOMPILER_$(GAROSREL))

thanks, I tweaked the lts102 receipe with that - this allows a bit more 
progress on solaris 9, although build fails, read below.

I added some further small tweaks and there is now some progress to 
report, IT allows me some platform checking:

solaris 9x -> unable to finish build + i386 vs amd64 issues
solaris 9s -> build begins but fails (differently than 9x: pk11 issues)
solaris 10x -> build + test OK
solaris 10s -> build OK, test fail
solaris 11x -> build + test OK
solaris 11s -> buils OK, test fail

So my take is: the code is not SPARC safe, this means in my humble 
experience, bad code in terms of structures.. but I see no relevant 
patches in the oracle stuff.


Riccardo

PS: regarding solaris9, I see that at least for 9s the issue is pk11, I 
noticed that for the old 0.9.8 the configuration appeared optional:

# support for pkcs11 engine 
http://blogs.sun.com/chichang1/entry/how_to_integrate_pkcs11_engine
ifdef PKCS11
         PACKAGING_PLATFORMS = solaris9-sparc solaris9-i386 
solaris10-sparc solaris10-i386
ifneq ($(shell /usr/bin/uname -r),5.9)
         PATCHFILES += pkcs11_engine-0.9.8o.patch.2010-09-03
         CONFIGURE_ARGS += --pk11-libname=$(abspath 
/usr/lib/$(MM_LIBDIR)/libpkcs11.so)
endif
endif


now it is always, has pk11 known issues on solaris9 and should it be 
disabled? or am I reading this ifneq(5.9) wrong?



From rmottola at opencsw.org  Sat May  6 10:02:06 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Sat, 6 May 2017 10:02:06 +0200
Subject: pagealign linking issues
Message-ID: <4fade955-d787-6ea4-f22e-b789078e60d7@opencsw.org>

Hi all.

after some tweaks, I am able to start configure and build of the new 
OpenSSL 1.0.2 on Solaris 9.
On both x86 and SPARC, however, build fails with:

ld: fatal: file /usr/lib/ld/map.pagealign: stat failed: No such file or 
directory

Indeed, that linker file does not exist. Have you already seen such 
issue? it appears to be a feature of newer SUN compiler/linker, but what 
does call it and can I fix that?

Thanks,
Riccardo

PS: no news on the fact that new OpenSSL fails tests on SPARC

From rmottola at opencsw.org  Sat May  6 17:22:26 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Sat, 6 May 2017 17:22:26 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
Message-ID: <552213d3-f838-4027-5587-2b6071857553@opencsw.org>

Hi,

Riccardo Mottola via maintainers wrote:
> solaris 9x -> unable to finish build + i386 vs amd64 issues
> solaris 9s -> build begins but fails (differently than 9x: pk11 issues)
> solaris 10x -> build + test OK
> solaris 10s -> build OK, test fail
> solaris 11x -> build + test OK
> solaris 11s -> buils OK, test fail
>
> So my take is: the code is not SPARC safe, this means in my humble 
> experience, bad code in terms of structures.. but I see no relevant 
> patches in the oracle stuff. 

I found this reference of exactly that error: it says it has been 
solved, but I see no reference to a solution and/or a patch!

https://github.com/openssl/openssl/issues/2540

Riccardo

From rmottola at opencsw.org  Tue May  9 19:09:26 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Tue, 9 May 2017 19:09:26 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
Message-ID: <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>

Hi all,

some updates on the OpenSSL 1.0.2 update process.

I was able to update and clean up the build process.
It now builds on solaris 9, 10, 11 on x86 and sparc.

I set for solaris9 pentium_pro as a minimum.

I stopped using the "tweaked" targets
PATCHFILES += optimized_configure_targets.patch

and most issues I reported the past days went away. Possibly we should 
tweak them for different compilers or something like that.

On solaris 9, 10 & 11 pentium_pro or amd64, test fail
On SPARC we have the same test failure on 9, 10 & 11!

I found this reference of exactly that error: it says it has been 
solved, but I see no reference to a solution and/or a patch!
https://github.com/openssl/openssl/issues/2540


Input on that would be nice.

Everything is comited in /branches/lts102 so you can test yourself.
There is already newer stuff than 1.0.2 so openssl1 might evolve, but 
1.0.2 has been designated as LTS


Riccardo

From dam at opencsw.org  Tue May  9 22:03:06 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Tue, 9 May 2017 22:03:06 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
 <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
Message-ID: <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>

Hi Riccardo,

Am 09.05.2017 um 19:09 schrieb Riccardo Mottola <rmottola at opencsw.org>:
> some updates on the OpenSSL 1.0.2 update process.
> 
> I was able to update and clean up the build process.
> It now builds on solaris 9, 10, 11 on x86 and sparc.
> 
> I set for solaris9 pentium_pro as a minimum.
> 
> I stopped using the "tweaked" targets
> PATCHFILES += optimized_configure_targets.patch
> 
> and most issues I reported the past days went away. Possibly we should tweak them for different compilers or something like that.
> 
> On solaris 9, 10 & 11 pentium_pro or amd64, test fail
> On SPARC we have the same test failure on 9, 10 & 11!
> 
> I found this reference of exactly that error: it says it has been solved, but I see no reference to a solution and/or a patch!
> https://github.com/openssl/openssl/issues/2540
> 
> Input on that would be nice.

Sure, I just asked :-)
  https://github.com/openssl/openssl/issues/2540


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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170509/365d22e5/attachment.asc>

From dam at opencsw.org  Tue May  9 22:05:29 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Tue, 9 May 2017 22:05:29 +0200
Subject: Current issues
Message-ID: <8703D2B9-D2DA-4C11-B328-04FF4B836A8E@opencsw.org>

Hi Riccardo,

the updates lead to a number of hidden issues: the update of freetype introduced
a cyclic dependency between freetype and harfbuzz as explained here:
  https://github.com/Homebrew/legacy-homebrew/issues/27847
This lead to stop catalog generation and seduced me to push the catalog
which was unfortunately not signed.

Riccardo, would you mind having a look at harfbuzz and if the issue with
the cyclic dependency cannot be solved remove the dependency from freetype
to harfbuzz again? This is a bit pressing as the current catalog on the
mirrors is broken.


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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170509/6d27f944/attachment.asc>

From rmottola at opencsw.org  Tue May  9 22:47:39 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Tue, 9 May 2017 22:47:39 +0200
Subject: Current issues
In-Reply-To: <8703D2B9-D2DA-4C11-B328-04FF4B836A8E@opencsw.org>
References: <8703D2B9-D2DA-4C11-B328-04FF4B836A8E@opencsw.org>
Message-ID: <54ef5617-459f-4b70-0b57-90ffa9df29d7@opencsw.org>

Hi,

Dagobert Michelsen wrote:
> the updates lead to a number of hidden issues: the update of freetype introduced
> a cyclic dependency between freetype and harfbuzz as explained here:
>    https://github.com/Homebrew/legacy-homebrew/issues/27847
> This lead to stop catalog generation and seduced me to push the catalog
> which was unfortunately not signed.

Indeed, on solaris 10 freetype picks up harfbuzz automatically. I don't 
think the suggested procedure of building freetype, then harfbuzz then 
freetype again can be done in mgar.

Luckily there is --without-harfbuzz, which I quickly added and commited.
>
> Riccardo, would you mind having a look at harfbuzz and if the issue with
> the cyclic dependency cannot be solved remove the dependency from freetype
> to harfbuzz again? This is a bit pressing as the current catalog on the
> mirrors is broken.

I hope I fixed the receipe. However I cannot check the packages, all the 
dependencies seem wrong, I get overrides suggested for things which are 
there. Strangely, only on 10, on 9 not...
Could the broken catalog cause that? What can we do?

I commited everything if you need to try.

Riccardo

PS: could I have noticed this somehow? checks and upload went fine?

From dam at opencsw.org  Tue May  9 23:19:21 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Tue, 9 May 2017 23:19:21 +0200
Subject: Current issues
In-Reply-To: <54ef5617-459f-4b70-0b57-90ffa9df29d7@opencsw.org>
References: <8703D2B9-D2DA-4C11-B328-04FF4B836A8E@opencsw.org>
 <54ef5617-459f-4b70-0b57-90ffa9df29d7@opencsw.org>
Message-ID: <6FAC8258-B5C2-4333-AF6A-D08F108AB22F@opencsw.org>

Hi Riccardo,

Am 09.05.2017 um 22:47 schrieb Riccardo Mottola <rmottola at opencsw.org>:
> Dagobert Michelsen wrote:
>> the updates lead to a number of hidden issues: the update of freetype introduced
>> a cyclic dependency between freetype and harfbuzz as explained here:
>>   https://github.com/Homebrew/legacy-homebrew/issues/27847
>> This lead to stop catalog generation and seduced me to push the catalog
>> which was unfortunately not signed.
> 
> Indeed, on solaris 10 freetype picks up harfbuzz automatically. I don't think the suggested procedure of building freetype, then harfbuzz then freetype again can be done in mgar.

This won?t help, unfortunately, as you would still get cyclic dependency on shared libs.

> Luckily there is --without-harfbuzz, which I quickly added and commited.

That seems to be the right action for now.

>> Riccardo, would you mind having a look at harfbuzz and if the issue with
>> the cyclic dependency cannot be solved remove the dependency from freetype
>> to harfbuzz again? This is a bit pressing as the current catalog on the
>> mirrors is broken.
> 
> I hope I fixed the receipe. However I cannot check the packages, all the dependencies seem wrong, I get overrides suggested for things which are there. Strangely, only on 10, on 9 not...
> Could the broken catalog cause that? What can we do?
> 
> I commited everything if you need to try.

I?ll have a look and push the fixed packages.

> PS: could I have noticed this somehow? checks and upload went fine?

I am not sure. I could have noticed by having a better look at the cron logs.


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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170509/7e651515/attachment.asc>

From rmottola at opencsw.org  Tue May  9 23:59:33 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Tue, 9 May 2017 23:59:33 +0200
Subject: Current issues
In-Reply-To: <6FAC8258-B5C2-4333-AF6A-D08F108AB22F@opencsw.org>
References: <8703D2B9-D2DA-4C11-B328-04FF4B836A8E@opencsw.org>
 <54ef5617-459f-4b70-0b57-90ffa9df29d7@opencsw.org>
 <6FAC8258-B5C2-4333-AF6A-D08F108AB22F@opencsw.org>
Message-ID: <efcea6af-c963-5908-9560-aa09a804f8c9@opencsw.org>

Hi Dago,

Dagobert Michelsen wrote:

> Indeed, on solaris 10 freetype picks up harfbuzz automatically. I don't think the suggested procedure of building freetype, then harfbuzz then freetype again can be done in mgar.
> This won?t help, unfortunately, as you would still get cyclic dependency on shared libs.

so there is a limitation in depdendencies actually.. a big issue, let's 
see if withhout harfbuzz everything works.
I hate this king od packages, it isnot the only one :(

This would be a reason to "Build" always in a very small chroot. If we 
had a way to install (or make visibile) only the tree of the 
dependencies actually listed in the package, so configure scripts could 
not "pick up" anything by change
>
>> Luckily there is --without-harfbuzz, which I quickly added and commited.
> That seems to be the right action for now.
>
>>> Riccardo, would you mind having a look at harfbuzz and if the issue with
>>> the cyclic dependency cannot be solved remove the dependency from freetype
>>> to harfbuzz again? This is a bit pressing as the current catalog on the
>>> mirrors is broken.
>> I hope I fixed the receipe. However I cannot check the packages, all the dependencies seem wrong, I get overrides suggested for things which are there. Strangely, only on 10, on 9 not...
>> Could the broken catalog cause that? What can we do?
>>
>> I commited everything if you need to try.
> I?ll have a look and push the fixed packages.

I actually noticed the issue also with other packages, where there are a 
lot of errors by just rebuilding.
I hope the breakage is not too big!

>
>> PS: could I have noticed this somehow? checks and upload went fine?
> I am not sure. I could have noticed by having a better look at the cron logs.

I think It might have been broken since quite a while? I was about to 
ask you right today why

https://www.opencsw.org/get-it/packages/

still lists old packages, while I uploaded several new days ago.

Rccardo

From dam at opencsw.org  Wed May 10 09:09:13 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Wed, 10 May 2017 09:09:13 +0200
Subject: Power Outage
Message-ID: <863F0283-3CDE-4191-9C44-052817F50555@opencsw.org>

Hi folks,

we are currently experiencing an unplanned power outage in our datacenter.
It looks like the UPS failed in a way which prevented the automatic bypass to work.
We bypassed it now manually and will restart all machines during the day.

The buildfarm, buildbot and catalog generation are currently down.
The primary mirror and Mail are not affected.


Sorry for the inconvenience

  ? 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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170510/2a226a4a/attachment.asc>

From dam at opencsw.org  Wed May 10 15:43:03 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Wed, 10 May 2017 15:43:03 +0200
Subject: Power Outage
In-Reply-To: <863F0283-3CDE-4191-9C44-052817F50555@opencsw.org>
References: <863F0283-3CDE-4191-9C44-052817F50555@opencsw.org>
Message-ID: <59E7F0BD-CECA-4399-8429-B3B5310FEE52@opencsw.org>

Hi,

Am 10.05.2017 um 09:09 schrieb Dagobert Michelsen via maintainers <maintainers at lists.opencsw.org>:
> we are currently experiencing an unplanned power outage in our datacenter.
> It looks like the UPS failed in a way which prevented the automatic bypass to work.
> We bypassed it now manually and will restart all machines during the day.
> 
> The buildfarm, buildbot and catalog generation are currently down.
> The primary mirror and Mail are not affected.

Power has been restored and everything should work again.
If you encounter anything suspicious just let me know.


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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170510/7cbea6ea/attachment.asc>

From dam at opencsw.org  Wed May 10 15:52:53 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Wed, 10 May 2017 15:52:53 +0200
Subject: Current issues
In-Reply-To: <efcea6af-c963-5908-9560-aa09a804f8c9@opencsw.org>
References: <8703D2B9-D2DA-4C11-B328-04FF4B836A8E@opencsw.org>
 <54ef5617-459f-4b70-0b57-90ffa9df29d7@opencsw.org>
 <6FAC8258-B5C2-4333-AF6A-D08F108AB22F@opencsw.org>
 <efcea6af-c963-5908-9560-aa09a804f8c9@opencsw.org>
Message-ID: <F726271F-AF0C-476D-ACF5-DDD73C9BD722@opencsw.org>

Hi Riccardo.

Am 09.05.2017 um 23:59 schrieb Riccardo Mottola <rmottola at opencsw.org>:
> Dagobert Michelsen wrote:
> 
>> Indeed, on solaris 10 freetype picks up harfbuzz automatically. I don't think the suggested procedure of building freetype, then harfbuzz then freetype again can be done in mgar.
>> This won?t help, unfortunately, as you would still get cyclic dependency on shared libs.
> 
> so there is a limitation in depdendencies actually.. a big issue, let's see if withhout harfbuzz everything works.
> I hate this king od packages, it isnot the only one :(
> 
> This would be a reason to "Build" always in a very small chroot. If we had a way to install (or make visibile) only the tree of the dependencies actually listed in the package, so configure scripts could not "pick up" anything by change

I fixed freetype and updated harfbuzz and pushed everything and all should be fine now.
Also unstable/ has been updated to have the latest pushed packages.


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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170510/91785a15/attachment.asc>

From rmottola at opencsw.org  Wed May 10 18:26:24 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Wed, 10 May 2017 18:26:24 +0200
Subject: Current issues
In-Reply-To: <F726271F-AF0C-476D-ACF5-DDD73C9BD722@opencsw.org>
References: <8703D2B9-D2DA-4C11-B328-04FF4B836A8E@opencsw.org>
 <54ef5617-459f-4b70-0b57-90ffa9df29d7@opencsw.org>
 <6FAC8258-B5C2-4333-AF6A-D08F108AB22F@opencsw.org>
 <efcea6af-c963-5908-9560-aa09a804f8c9@opencsw.org>
 <F726271F-AF0C-476D-ACF5-DDD73C9BD722@opencsw.org>
Message-ID: <c0c6e11d-89a8-482c-71a2-d34996195be2@opencsw.org>

Hi,

Dagobert Michelsen wrote:
> I fixed freetype and updated harfbuzz and pushed everything and all should be fine now.
> Also unstable/ has been updated to have the latest pushed packages.

Fine!

I see that your fix, apart from the update, is:

-CONFIGURE_ARGS += --without-harfbuzz
+EXTRA_CONFIGURE_ARGS += --without-harfbuzz

What's the difference? I copied my line from another project and it 
"worked" because I checked that the parameter was correctly passed and 
freetype was built without harfbuzz (although I couldn't verify the 
package yesterday).

Does your update mean dependent packages need to be rebuilt? I 
configured gnustep against the installed "broken" freetype.

Emergency situation solved?

Riccardo

From dam at opencsw.org  Wed May 10 18:41:15 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Wed, 10 May 2017 18:41:15 +0200
Subject: Current issues
In-Reply-To: <c0c6e11d-89a8-482c-71a2-d34996195be2@opencsw.org>
References: <8703D2B9-D2DA-4C11-B328-04FF4B836A8E@opencsw.org>
 <54ef5617-459f-4b70-0b57-90ffa9df29d7@opencsw.org>
 <6FAC8258-B5C2-4333-AF6A-D08F108AB22F@opencsw.org>
 <efcea6af-c963-5908-9560-aa09a804f8c9@opencsw.org>
 <F726271F-AF0C-476D-ACF5-DDD73C9BD722@opencsw.org>
 <c0c6e11d-89a8-482c-71a2-d34996195be2@opencsw.org>
Message-ID: <C6E1F7B4-A2AF-460B-B58F-21B5CE7ABB90@opencsw.org>

Hi Riccardo,

Am 10.05.2017 um 18:26 schrieb Riccardo Mottola <rmottola at opencsw.org>:
> Dagobert Michelsen wrote:
>> I fixed freetype and updated harfbuzz and pushed everything and all should be fine now.
>> Also unstable/ has been updated to have the latest pushed packages.
> 
> Fine!
> 
> I see that your fix, apart from the update, is:
> 
> -CONFIGURE_ARGS += --without-harfbuzz
> +EXTRA_CONFIGURE_ARGS += --without-harfbuzz
> 
> What's the difference? I copied my line from another project and it "worked" because I checked that the parameter was correctly passed and freetype was built without harfbuzz (although I couldn't verify the package yesterday).

That is because of the conditional assignment:
  CONFIGURE_ARGS ?= $(DIRPATHS) $(EXTRA_CONFIGURE_ARGS)
  https://buildfarm.opencsw.org/source/xref/opencsw/csw/mgar/gar/v2/gar.lib.mk#863

If you set CONFIGURE_ARGS before the include the ?= does nothing.
Essentially += is adding to the existing value, so it doesn?t work because of the
ordering.

> Does your update mean dependent packages need to be rebuilt? I configured gnustep against the installed "broken" freetype.
> 
> Emergency situation solved?

Yes, everything solved now.


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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170510/4ede7d88/attachment-0001.asc>

From rmottola at opencsw.org  Wed May 10 19:07:15 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Wed, 10 May 2017 19:07:15 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
 <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
 <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>
Message-ID: <963e50c2-447c-9783-e52b-d137834d7699@opencsw.org>

Hi,

I saw the resposne.

Dagobert Michelsen wrote:
> Sure, I just asked :-)
>   https://github.com/openssl/openssl/issues/2540
>

I gradually upgraded from j to k, testing and indeed, k seems to work! I 
can build on all platforms and tests pass on sparc 9, 10 and 11.

I have some packaging errors I will investigate, but the outlook is 
promising.

I will let you know if I can fix the packaging issues or if I need help, 
then we can decide the next steps.

Riccardo


From dam at opencsw.org  Wed May 10 19:11:31 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Wed, 10 May 2017 19:11:31 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <963e50c2-447c-9783-e52b-d137834d7699@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
 <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
 <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>
 <963e50c2-447c-9783-e52b-d137834d7699@opencsw.org>
Message-ID: <74B6902F-26B2-4D92-8B97-5676EDC0859C@opencsw.org>

Hi Riccardo,

Am 10.05.2017 um 19:07 schrieb Riccardo Mottola <rmottola at opencsw.org>:
> Dagobert Michelsen wrote:
>> Sure, I just asked :-)
>>  https://github.com/openssl/openssl/issues/2540
>> 
> 
> I gradually upgraded from j to k, testing and indeed, k seems to work! I can build on all platforms and tests pass on sparc 9, 10 and 11.
> 
> I have some packaging errors I will investigate, but the outlook is promising.
> 
> I will let you know if I can fix the packaging issues or if I need help, then we can decide the next steps.

Sure, if in doubt just ask :-)


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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170510/269098cd/attachment.asc>

From rmottola at opencsw.org  Thu May 11 00:17:01 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Thu, 11 May 2017 00:17:01 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <74B6902F-26B2-4D92-8B97-5676EDC0859C@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
 <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
 <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>
 <963e50c2-447c-9783-e52b-d137834d7699@opencsw.org>
 <74B6902F-26B2-4D92-8B97-5676EDC0859C@opencsw.org>
Message-ID: <4d6a4592-13d8-3f47-1097-5c79a922fb59@opencsw.org>

Hi,

Dagobert Michelsen wrote:
> Sure, if in doubt just ask:-)
>
I can package on 9x, 10s, 10x.
However on 9s, I get these:


CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/libcrypto.so.1.0.0|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/libssl.so.1.0.0|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libaep.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libchil.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libsureware.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libpadlock.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libnuron.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libatalla.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libubsec.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libgmp.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/lib4758cca.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libcapi.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libgost.so|arch_id=18|arch_name=sparcv8+
CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += 
binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libcswift.so|arch_id=18|arch_name=sparcv8+


since I do not see overrides for others.

Might it have do with ISA_DEFAULT = sparcv8plus ?
If the overrides are legit, is there a better solution to have them go 
away than overriding?

Riccardo

From dam at opencsw.org  Thu May 11 07:27:35 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Thu, 11 May 2017 07:27:35 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <4d6a4592-13d8-3f47-1097-5c79a922fb59@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
 <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
 <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>
 <963e50c2-447c-9783-e52b-d137834d7699@opencsw.org>
 <74B6902F-26B2-4D92-8B97-5676EDC0859C@opencsw.org>
 <4d6a4592-13d8-3f47-1097-5c79a922fb59@opencsw.org>
Message-ID: <C50F2EFB-64CF-48F1-87FF-00F24B826F21@opencsw.org>

Hi Riccardo,

Am 11.05.2017 um 00:17 schrieb Riccardo Mottola <rmottola at opencsw.org>:
> Dagobert Michelsen wrote:
>> Sure, if in doubt just ask:-)
>> 
> I can package on 9x, 10s, 10x.
> However on 9s, I get these:
> 
> 
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/libcrypto.so.1.0.0|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/libssl.so.1.0.0|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libaep.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libchil.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libsureware.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libpadlock.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libnuron.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libatalla.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libubsec.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libgmp.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/lib4758cca.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libcapi.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libgost.so|arch_id=18|arch_name=sparcv8+
> CHECKPKG_OVERRIDES_CSWlibssl1-0-0 += binary-architecture-does-not-match-placement|file=opt/csw/lib/openssl-1.0.0/engines/libcswift.so|arch_id=18|arch_name=sparcv8+
> 
> since I do not see overrides for others.
> 
> Might it have do with ISA_DEFAULT = sparcv8plus ?

Yes.

> If the overrides are legit, is there a better solution to have them go away than overriding?

In this case I would just override with a comment explaining why
the ISA was raised.


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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170511/18bc0dc0/attachment.asc>

From rmottola at opencsw.org  Fri May 12 19:42:19 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Fri, 12 May 2017 19:42:19 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <C50F2EFB-64CF-48F1-87FF-00F24B826F21@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
 <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
 <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>
 <963e50c2-447c-9783-e52b-d137834d7699@opencsw.org>
 <74B6902F-26B2-4D92-8B97-5676EDC0859C@opencsw.org>
 <4d6a4592-13d8-3f47-1097-5c79a922fb59@opencsw.org>
 <C50F2EFB-64CF-48F1-87FF-00F24B826F21@opencsw.org>
Message-ID: <0bcf2f64-56bd-54ed-7dd3-3f69f6a8938a@opencsw.org>

Hi!

Heads up!

Dagobert Michelsen wrote:
> In this case I would just override with a comment explaining why
> the ISA was raised.

1.0.2k compiled on all platforms and test suite passed.

Should I push my packages?

How would you test this further, Jan?

Riccardo

From dam at opencsw.org  Fri May 12 19:56:47 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Fri, 12 May 2017 19:56:47 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <0bcf2f64-56bd-54ed-7dd3-3f69f6a8938a@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
 <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
 <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>
 <963e50c2-447c-9783-e52b-d137834d7699@opencsw.org>
 <74B6902F-26B2-4D92-8B97-5676EDC0859C@opencsw.org>
 <4d6a4592-13d8-3f47-1097-5c79a922fb59@opencsw.org>
 <C50F2EFB-64CF-48F1-87FF-00F24B826F21@opencsw.org>
 <0bcf2f64-56bd-54ed-7dd3-3f69f6a8938a@opencsw.org>
Message-ID: <B7712FCD-8F78-43C7-A83F-947117FBE945@opencsw.org>

Hi Riccardo,

Am 12.05.2017 um 19:42 schrieb Riccardo Mottola <rmottola at opencsw.org>:
> 
> Dagobert Michelsen wrote:
>> In this case I would just override with a comment explaining why
>> the ISA was raised.
> 
> 1.0.2k compiled on all platforms and test suite passed.
> 
> Should I push my packages?
> 
> How would you test this further, Jan?

I suggest to
  mkdir /home/experimental/openssl
  mv <package> /home/experimental/openssl
so we have a catalog to test these in experimental first before pushing.


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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170512/4dae57c1/attachment.asc>

From jh at opencsw.org  Mon May 15 08:26:53 2017
From: jh at opencsw.org (Jan Holzhueter)
Date: Mon, 15 May 2017 08:26:53 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <0bcf2f64-56bd-54ed-7dd3-3f69f6a8938a@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
 <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
 <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>
 <963e50c2-447c-9783-e52b-d137834d7699@opencsw.org>
 <74B6902F-26B2-4D92-8B97-5676EDC0859C@opencsw.org>
 <4d6a4592-13d8-3f47-1097-5c79a922fb59@opencsw.org>
 <C50F2EFB-64CF-48F1-87FF-00F24B826F21@opencsw.org>
 <0bcf2f64-56bd-54ed-7dd3-3f69f6a8938a@opencsw.org>
Message-ID: <ac62a105-811f-9059-dd4f-ee3ac7b97cf0@opencsw.org>

Hi,

Am 12.05.17 um 19:42 schrieb Riccardo Mottola:
> Hi!
> 
> Heads up!
> 
> Dagobert Michelsen wrote:
>> In this case I would just override with a comment explaining why
>> the ISA was raised.
> 
> 1.0.2k compiled on all platforms and test suite passed.
> 
> Should I push my packages?
> 
> How would you test this further, Jan?


yes please push it to experimental.
I will try to find some time to test it this week and review it.

Greetings
Jan


From rmottola at opencsw.org  Mon May 15 08:56:55 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Mon, 15 May 2017 08:56:55 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <B7712FCD-8F78-43C7-A83F-947117FBE945@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
 <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
 <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>
 <963e50c2-447c-9783-e52b-d137834d7699@opencsw.org>
 <74B6902F-26B2-4D92-8B97-5676EDC0859C@opencsw.org>
 <4d6a4592-13d8-3f47-1097-5c79a922fb59@opencsw.org>
 <C50F2EFB-64CF-48F1-87FF-00F24B826F21@opencsw.org>
 <0bcf2f64-56bd-54ed-7dd3-3f69f6a8938a@opencsw.org>
 <B7712FCD-8F78-43C7-A83F-947117FBE945@opencsw.org>
Message-ID: <99c92fc4-73ab-d1e1-d642-9f08c2fcec3c@opencsw.org>

Hi all,


Dagobert Michelsen wrote:
> I suggest to
>   mkdir /home/experimental/openssl
>   mv <package> /home/experimental/openssl
> so we have a catalog to test these in experimental first before pushing.

Just did it, sorry for the delay over the weekend.

One immediate test would be OpenSSH I guess? And try things like sftp.
I remember when a bad openssl version broke our build system, was it due 
to ssh? I wonder if we can test that without risking a block again.

Can I help in some tests?

Riccardo

From jh at opencsw.org  Mon May 15 15:35:11 2017
From: jh at opencsw.org (Jan Holzhueter)
Date: Mon, 15 May 2017 15:35:11 +0200
Subject: OpenSSL 1.0.2 transition
In-Reply-To: <99c92fc4-73ab-d1e1-d642-9f08c2fcec3c@opencsw.org>
References: <7ade65ca-7c51-3a07-b9c1-e5558d7e3ed1@opencsw.org>
 <1d8b31cb-4f3e-fb00-4154-c7609285cc95@opencsw.org>
 <4985a69b-2dd0-dc81-2755-10ace535a4e3@opencsw.org>
 <C8093C1C-21A1-474A-B2E1-12D09168C3E7@opencsw.org>
 <1aba6a52-d01c-97cb-5831-b3064eaed45b@opencsw.org>
 <552213d3-f838-4027-5587-2b6071857553@opencsw.org>
 <b24481d3-ba04-77ba-23ac-b45ca333335d@opencsw.org>
 <6BFF0A75-689E-429D-8D72-8F5801235618@opencsw.org>
 <963e50c2-447c-9783-e52b-d137834d7699@opencsw.org>
 <74B6902F-26B2-4D92-8B97-5676EDC0859C@opencsw.org>
 <4d6a4592-13d8-3f47-1097-5c79a922fb59@opencsw.org>
 <C50F2EFB-64CF-48F1-87FF-00F24B826F21@opencsw.org>
 <0bcf2f64-56bd-54ed-7dd3-3f69f6a8938a@opencsw.org>
 <B7712FCD-8F78-43C7-A83F-947117FBE945@opencsw.org>
 <99c92fc4-73ab-d1e1-d642-9f08c2fcec3c@opencsw.org>
Message-ID: <65869b51-cc73-d9ca-48f7-e8cf61f90051@opencsw.org>

Am 15.05.17 um 08:56 schrieb Riccardo Mottola:
> Hi all,
> 
> 
> Dagobert Michelsen wrote:
>> I suggest to
>>   mkdir /home/experimental/openssl
>>   mv <package> /home/experimental/openssl
>> so we have a catalog to test these in experimental first before pushing.
> 
> Just did it, sorry for the delay over the weekend.
> 
> One immediate test would be OpenSSH I guess? And try things like sftp.
> I remember when a bad openssl version broke our build system, was it due
> to ssh? I wonder if we can test that without risking a block again.

First we can roll out it to experimental*
I would start with openssl speed to test if nothing obvious is broken.
I did not check what patches you pulled from oracle yet.
But if the openssl speed is like 10x slower then the old one we shoud
look into the patches form oracle what might help etc.

I have to dig out the test cases for Intel und T5 aes enc support tests.
other test case would be setting up some apache lightty and nginx to
test tsl stuff there if anything is broken there.

Again I hope I have some time this week to look.


From rmottola at opencsw.org  Tue May 16 15:40:11 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Tue, 16 May 2017 15:40:11 +0200
Subject: SF.net SVN: gar:[26210] csw/mgar/pkg/gs_ftp/trunk/Makefile
In-Reply-To: <d603a985-4c94-b16d-ffad-88d345e326b4@opencsw.org>
References: <1494545500.638134.24839@sfp-scm-3.v30.ch3.sourceforge.com>
 <3A1EC5B0-420D-40C0-B1DC-3F7C8AB53771@opencsw.org>
 <d603a985-4c94-b16d-ffad-88d345e326b4@opencsw.org>
Message-ID: <5b7d2ac7-7738-cc37-281d-ebb79815beac@opencsw.org>

Hi,

let me add to this topic, since I have seen that also "gnome" apps, for 
example, have different prefixes.

One think is of course the original application name, it may already 
contain the library/framework/desktop name.

Riccardo Mottola via devel wrote:
> Dagobert Michelsen via devel wrote:
>> please use CSWgnustep-ftp as ?gs? usually means Ghostscript.
>> Please also add
>>   OBSOLETED_BY_CSWgnustep-ftp += CSWgs-ftp
>
> Do you insist on that? it has been called that way since 2 years.
> Actually the program is just called "FTP" and I deemed the name too
> generic.
>
> I used the gs prefix for other GNUstep applications too: terminal,
> price, & zipper.
>
> What policy shoud we use? Keep the original name and use a prefix only
> when necessary? Always use it?
>
> gs- is much shorter and handier than gnustep- :)
> The "core" gnustep packages have gnustep_ so we could stick to that too.
> In that case it is however directly in the upstream package name,
> gnustep-base/gnustep-gui/gnustep-back

at the moment we do have

Core framework:
CSWgnustep-back
CSWgnustep-base
CSWgnustep-gui
CSWgnustep-make

No prefix at all:
CSWgorm (graphical designer: dev tool)
CSWprojectcenter (IDE: dev tool)
CSWperformance (framework)

Assorted with the "gs" prefix
CSWgs-zipper
CSWgs-terminal
CSWgs-price
CSWgs-ftp


All the issue started because I deemed "ftp" and "terminal" too generic 
as names.

Do we want to rename "everything" to CSWgnustep- ? just the ambiguous 
one? keep CSWgs- instead but apply it to the ones which have no prefix?


Waiting for you,
Riccardo


From dam at opencsw.org  Tue May 16 15:48:42 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Tue, 16 May 2017 15:48:42 +0200
Subject: SF.net SVN: gar:[26210] csw/mgar/pkg/gs_ftp/trunk/Makefile
In-Reply-To: <5b7d2ac7-7738-cc37-281d-ebb79815beac@opencsw.org>
References: <1494545500.638134.24839@sfp-scm-3.v30.ch3.sourceforge.com>
 <3A1EC5B0-420D-40C0-B1DC-3F7C8AB53771@opencsw.org>
 <d603a985-4c94-b16d-ffad-88d345e326b4@opencsw.org>
 <5b7d2ac7-7738-cc37-281d-ebb79815beac@opencsw.org>
Message-ID: <B6EC336B-9E49-4F1A-BBB4-9E258D2D6A5D@opencsw.org>

Hi Riccardo,

Am 16.05.2017 um 15:40 schrieb Riccardo Mottola <rmottola at opencsw.org>:
> let me add to this topic, since I have seen that also "gnome" apps, for example, have different prefixes.
> 
> One think is of course the original application name, it may already contain the library/framework/desktop name.

Let?s just keep the names as they are now, I thought this was a new package.


Best regards

  ? Dago

> 
> Riccardo Mottola via devel wrote:
>> Dagobert Michelsen via devel wrote:
>>> please use CSWgnustep-ftp as ?gs? usually means Ghostscript.
>>> Please also add
>>>  OBSOLETED_BY_CSWgnustep-ftp += CSWgs-ftp
>> 
>> Do you insist on that? it has been called that way since 2 years.
>> Actually the program is just called "FTP" and I deemed the name too
>> generic.
>> 
>> I used the gs prefix for other GNUstep applications too: terminal,
>> price, & zipper.
>> 
>> What policy shoud we use? Keep the original name and use a prefix only
>> when necessary? Always use it?
>> 
>> gs- is much shorter and handier than gnustep- :)
>> The "core" gnustep packages have gnustep_ so we could stick to that too.
>> In that case it is however directly in the upstream package name,
>> gnustep-base/gnustep-gui/gnustep-back
> 
> at the moment we do have
> 
> Core framework:
> CSWgnustep-back
> CSWgnustep-base
> CSWgnustep-gui
> CSWgnustep-make
> 
> No prefix at all:
> CSWgorm (graphical designer: dev tool)
> CSWprojectcenter (IDE: dev tool)
> CSWperformance (framework)
> 
> Assorted with the "gs" prefix
> CSWgs-zipper
> CSWgs-terminal
> CSWgs-price
> CSWgs-ftp
> 
> 
> All the issue started because I deemed "ftp" and "terminal" too generic as names.
> 
> Do we want to rename "everything" to CSWgnustep- ? just the ambiguous one? keep CSWgs- instead but apply it to the ones which have no prefix?
> 
> 
> Waiting for you,
> Riccardo
> 

--
"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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170516/99b91b49/attachment.asc>

From rmottola at opencsw.org  Wed May 31 14:22:22 2017
From: rmottola at opencsw.org (Riccardo Mottola)
Date: Wed, 31 May 2017 14:22:22 +0200
Subject: GNUtls3 build issues
Message-ID: <5c469b42-2c4d-f6c5-b5df-a82d72bdb43d@opencsw.org>

Hi,

I am trying to rebuild the current recipe of GNUtls3 as a start to bring 
things up-to-date.
I need it on Solaris 9 too. After the "respin" we can decide a path to 
upgrade to a different soname version (as I understand it we have 2.8, 
current would be 3.4, correct me if I am wrong)

I just changed the compiler from "GCC4" to "GNU" since it would not work 
else on solaris 10.

On Solaris 10 I get this error:
copying selected object files to avoid basename conflicts...
   CC       psk.o
   CCLD     psktool
ld: fatal: file /opt/csw/lib/libopts.so: wrong ELF class: ELFCLASS32
ld: fatal: file processing errors. No output written to .libs/psktool
collect2: error: ld returned 1 exit status
gmake[4]: *** [Makefile:1831: psktool] Error 1

which totally puzzles me!

On solaris 9, I get:
   CCLD     psktool
Undefined                       first referenced
  symbol                             in file
inet_aton                           ../lib/.libs/libgnutls.so
ld: fatal: Symbol referencing errors. No output written to .libs/psktool
collect2: ld returned 1 exit status

Interesting that it is the same file. Do we have a workaround there? 
Perhaps it is actually the same issue as above just with a different error?

Riccardo

From dam at opencsw.org  Wed May 31 16:43:23 2017
From: dam at opencsw.org (Dagobert Michelsen)
Date: Wed, 31 May 2017 16:43:23 +0200
Subject: [openssl-announce] OpenSSL version 1.0.2l published
In-Reply-To: <81776028-30f8-96f5-2685-34bbd04d4bbd@opencsw.org>
References: <20170525135756.GA25645@openssl.org>
 <F6EC4B81-154B-4BDE-AD06-043A9F203BD6@opencsw.org>
 <af7f6c81-ceab-f322-2874-0dfc35fe035e@opencsw.org>
 <C80B5DD2-25AB-49A2-8A6C-6379016AE270@opencsw.org>
 <81776028-30f8-96f5-2685-34bbd04d4bbd@opencsw.org>
Message-ID: <D190BA7E-35CB-4255-B1FB-9F199AFE84A4@opencsw.org>

Hi Riccardo,

> Am 30.05.2017 um 15:22 schrieb Riccardo Mottola <rmottola at opencsw.org>:
> 
> Hi guys,

The Solaris 11 version is lacking the 64 bit libssl.so.1.0.0 library!

These lines were missing:

dam at unstable11s [unstable11s]:/home/dam/mgar/pkg/openssl1/branches/lts102 > /opt/csw/bin/svn diff
Index: Makefile
===================================================================
--- Makefile    (Revision 26218)
+++ Makefile    (Arbeitskopie)
@@ -184,6 +184,8 @@
 BUILD64-sparc-5.9 = 1
 BUILD64-sparc-5.10 = 1
 BUILD64-i386-5.10 = 1
+BUILD64-sparc-5.11 = 1
+BUILD64-i386-5.11 = 1
 BUILD64 = $(BUILD64-$(GARCH)-$(GAROSREL))

 ISA_DEFAULT_i386-5.9 = pentium_pro


I committed, so you can just ?svn up? and ?mgar remerge && mgar repackage?.


Best regards

  ? Dago

> 
> 
> Dagobert Michelsen wrote:
>> We can continue testing with what we have and build the new ones in parallel.
> 
> just finished updating.. everything still builds and tests continue to pass!
> 
> I commited and moved all the packages to /home/experimental/openssl like last time
> Let me know!
> 
> Do you confirm that the suggested compilers I am using are the best option for each OS?
> 
> # This recipe can be used to compile openssl with Sun Studio or GCC
> # Just define the GARCOMPILER to the GNU or SUN
> GARCOMPILER_5.9 = SOS12
> GARCOMPILER_5.10 = SOS12U4
> GARCOMPILER_5.11 = SOS12U4
> GARCOMPILER = $(GARCOMPILER_$(GAROSREL))
> GARCOMPILER_TYPE = $(if $(findstring GCC,$(GARCOMPILER)),GNU,SUN)
> 
> 
> Right now everything works because I disabled most "extra" optimizations supplied with the patch:
> PATCHFILES += optimized_configure_targets.patch
> 
> Because this caused build failures, link failures on most platforms and did not even work at all on 9.
> 
> In case, we probably need to tweak this matrix:
> # The corresponding os/compiler to pass to the
> # openssl Configure script
> i386_SUN_CONFIGURE_TARGET               = no-sse2 solaris-x86-cc-sunw
> pentium_pro_SUN_CONFIGURE_TARGET        = solaris-x86-cc-sunw
> amd64_SUN_CONFIGURE_TARGET              = solaris64-x86_64-cc-sunw
> 
> i386_GNU_CONFIGURE_TARGET               = no-sse2 solaris-x86-gcc
> pentium_pro_GNU_CONFIGURE_TARGET        = solaris-x86-gcc
> amd64_GNU_CONFIGURE_TARGET              = solaris64-x86_64-gcc
> 
> sparcv8_SUN_CONFIGURE_TARGET            = solaris-sparcv8-cc-sunw
> sparcv8plus_SUN_CONFIGURE_TARGET        = solaris-sparcv9-cc-sunw
> sparcv8plus+vis_SUN_CONFIGURE_TARGET    = solaris-sparcv9+vis-cc-sunw
> sparcv9_SUN_CONFIGURE_TARGET_5.11               = solaris64-sparcv9-cc-sunw
> sparcv9_SUN_CONFIGURE_TARGET_5.10               = solaris64-sparcv9-cc-sunw10
> sparcv9_SUN_CONFIGURE_TARGET            = $(sparcv9_SUN_CONFIGURE_TARGET_$(GAROSREL))
> 
> depending on OS and Compiler (if that makes sense... I think that OS should use the corresponding compiler, am I right?)
> 
> But maybe this makes sense only if Jan discovers a bad performance drop.
> 
> Riccardo

--
"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: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20170531/a96dbc7c/attachment.asc>