[libgadu-devel] Cleanup of -Wall flags

2012-08-27 Thread Daniel Macks
Now that configure.ac puts -Wall in CFLAGS for the whole build process 
conditional on GCC, that same flag should not also be hardcoded in any specific 
makefiles. There are several instances of it in test/automatic/Makefile.am and 
test/manual/Makefile.am

While working on that, I checked the automake documentation for CFLAGS, and I 
think test/automatic/Makefile.am has another problem as well. Defining that 
variable there overwrites CFLAGS that the user might set earlier (propagated 
from ./configure). AM_CFLAGS is the automake variable that the package-author 
controls instead:

-CFLAGS = -Wall -ggdb -DGG_IGNORE_DEPRECATED
+AM_CFLAGS = -ggdb -DGG_IGNORE_DEPRECATED

message2 has its own _CFLAGS variable that overrides the more general AM_ one, 
so they need to be merged for that case:

-message2_CFLAGS = @LIBXML2_CFLAGS@
+message2_CFLAGS = @LIBXML2_CFLAGS@ $(AM_CFLAGS)

dan

 --
Daniel Macks
dma...@netspace.org

___
libgadu-devel mailing list
libgadu-devel@lists.ziew.org
http://lists.ziew.org/mailman/listinfo/libgadu-devel


Re: [libgadu-devel] Kolejne zawodzące testy

2012-08-27 Thread Marcin Owsiany
On Mon, Aug 27, 2012 at 12:33:07AM +0200, Bartosz Brachaczek wrote:
 2012/8/18 Marcin Owsiany mar...@owsiany.pl:
  .. w r1314, na kfreebsd-i386 i hurd-i386, ale w różny sposób:
 
  https://buildd.debian.org/status/fetch.php?pkg=libgaduarch=kfreebsd-i386ver=1%3A1.12.0%7Epre%2Br1314-1stamp=1345247660
 
  https://buildd.debian.org/status/fetch.php?pkg=libgaduarch=hurd-i386ver=1%3A1.12.0%7Epre%2Br1314-1stamp=1345256162
 
 Mam poprawkę na problem z Test error (ten z Hurda). Brakowało
 continue po złapaniu EINTR w select(). Natomiast nie wiem jeszcze, co
 powoduje problem na kFreeBSD. Być może też gdzieś jest EINTR, a my
 tego nie obsługujemy. Wygląda na to, że na Linuksie nigdy nie
 dostajemy EINTR, jeśli nie mamy własnej obsługi sygnałów, ale gdzie
 indziej może się to zdarzyć (dziwne...).
 
 Wrzucę poprawki do repozytorium, jak tylko będzie to możliwe. Bo
 chwilowo serwer ma problemy, już napisałem Wojtkowi.

Jeśli wyślesz mi łatkę wcześniej, to mogę potestować.

-- 
Marcin Owsiany mar...@owsiany.pl  http://marcin.owsiany.pl/
GnuPG: 2048R/02F946FC  35E9 1344 9F77 5F43 13DD  6423 DBF4 80C6 02F9 46FC

Every program in development at MIT expands until it can read mail.
  -- Unknown
___
libgadu-devel mailing list
libgadu-devel@lists.ziew.org
http://lists.ziew.org/mailman/listinfo/libgadu-devel


Re: [libgadu-devel] Cleanup of -Wall flags

2012-08-27 Thread Bartosz Brachaczek
2012/8/27 Daniel Macks dma...@netspace.org:
 Now that configure.ac puts -Wall in CFLAGS for the whole build process 
 conditional on GCC, that same flag should not also be hardcoded in any 
 specific makefiles. There are several instances of it in 
 test/automatic/Makefile.am and test/manual/Makefile.am

 While working on that, I checked the automake documentation for CFLAGS, and I 
 think test/automatic/Makefile.am has another problem as well. Defining that 
 variable there overwrites CFLAGS that the user might set earlier (propagated 
 from ./configure). AM_CFLAGS is the automake variable that the package-author 
 controls instead:

 -CFLAGS = -Wall -ggdb -DGG_IGNORE_DEPRECATED
 +AM_CFLAGS = -ggdb -DGG_IGNORE_DEPRECATED

 message2 has its own _CFLAGS variable that overrides the more general AM_ 
 one, so they need to be merged for that case:

 -message2_CFLAGS = @LIBXML2_CFLAGS@
 +message2_CFLAGS = @LIBXML2_CFLAGS@ $(AM_CFLAGS)

 dan

That particular issue with CFLAGS in test/automatic/Makefile.am is
fixed in current svn trunk for some time already. The checkout url is
http://toxygen.net/svn/libgadu/trunk if you want to have a look at
this. I'm not saying the fixes are like super-correct or examples of
good automake practices (they're not and I'll probably fix them
sometime soon) but at least they work.

Thanks,
Bartosz
___
libgadu-devel mailing list
libgadu-devel@lists.ziew.org
http://lists.ziew.org/mailman/listinfo/libgadu-devel


Re: [libgadu-devel] Kolejne zawodzące testy

2012-08-27 Thread Bartosz Brachaczek
W dniu 27 sierpnia 2012 09:17 użytkownik Marcin Owsiany
mar...@owsiany.pl napisał:
 Jeśli wyślesz mi łatkę wcześniej, to mogę potestować.

Racja. Łatka jest bardzo prosta. Załączyłem ją w pliku 0005-*.patch.
Łatka 0006-*.patch może przy odrobinie szczęścia nieco rozjaśnić powód
problemów na kFreeBSD.

Bartosz


0005-Popraw-test-connect-pod-GNU-Hurd.patch
Description: Binary data


0006-B-d-paranoiczny-je-li-chodzi-o-potencjalne-b-dy-w-te.patch
Description: Binary data
___
libgadu-devel mailing list
libgadu-devel@lists.ziew.org
http://lists.ziew.org/mailman/listinfo/libgadu-devel


Re: [libgadu-devel] Kolejne zawodzące testy

2012-08-27 Thread Bartosz Brachaczek
Dobra. Chyba poprawiłem też ten test na kFreeBSD. Załączam łatki.


0008-Zawsze-zeruj-struktur-sockaddr_in-valgrind-FreeBSD.patch
Description: Binary data


0009-Popraw-implementacj-timeoutu-w-connect.patch
Description: Binary data
___
libgadu-devel mailing list
libgadu-devel@lists.ziew.org
http://lists.ziew.org/mailman/listinfo/libgadu-devel