atom feed13 messages in org.gnome.ekiga-devel-listRe: [Ekiga-devel-list] Win32 build br...
FromSent OnAttachments
Torsten SchlabachApr 11, 2008 4:11 am 
Matthias SchneiderApr 11, 2008 4:31 am 
Torsten SchlabachApr 11, 2008 5:29 am 
Torsten SchlabachApr 11, 2008 5:38 am 
Matthias SchneiderApr 11, 2008 5:43 am 
Torsten SchlabachApr 11, 2008 6:14 am 
Matthias SchneiderApr 11, 2008 6:40 am 
Matthias SchneiderApr 13, 2008 1:22 am.patch
Torsten SchlabachApr 14, 2008 5:14 am 
Torsten SchlabachApr 14, 2008 6:46 am 
Torsten SchlabachApr 14, 2008 7:23 am 
Matthias SchneiderApr 14, 2008 7:29 am 
Matthias SchneiderApr 14, 2008 7:32 am 
Subject:Re: [Ekiga-devel-list] Win32 build broken - status before the weekend
From:Torsten Schlabach (TSch@gmx.net)
Date:Apr 11, 2008 6:14:56 am
List:org.gnome.ekiga-devel-list

Hi Mattias!

Once I get it running, maybe we should try to get the same revisions running on your setup first? So we know at least we are talking about the same code? Its easy, we jsut have to put the revision number in the makefile...

Sounds good. I had tried that with the last revisions of OPAL and PTLib which
worked for me, probably from ~1 week back. But then I found that we're getting
way more than just those two off SVN in the meanwhile.

So if you can name any combination of revisions of all dependencies which are
fetched from SVN and which compiles, send it and I will try it on my system.

We'll get there ...

BTW: I think what we're doing here clearly makes the case for continous
integration, i.e. having automated builds (and maybe even tests). Right now, we
do these nightly build kind of manually, but as soon as we will stop doing that,
chances are that incompatible changes are committed into all the dependency
repositories so that next time someone tries to build, he'll start at squre #1.
It's always easier to fix it if feedback is timely.

I haven't understood the Ekiga / OPAL / any other dependencies version
management concept yet anyway, but that's a whole separate issue.

Enjoy the weekend. Unfortunately, I will not be able to continue until monday.

Regards, Torsten

-------- Original-Nachricht --------

Datum: Fri, 11 Apr 2008 14:43:59 +0200 Von: Matthias Schneider <matt@TU-Ilmenau.DE> An: Ekiga development mailing list <ekig@gnome.org> Betreff: Re: [Ekiga-devel-list] Win32 build broken - status before the weekend

Quoting Torsten Schlabach <TSch@gmx.net>:

So for anyone wondering what exactly might be wrong ...

I have

Debian lenny (installed yesterday, April 10th) it's a 32 bit VServer on an AMD64 host machine with a 32 bit userland

Freshly installed packages as per the Wiki page.

I do

# svn co http://svn.gnome.org/svn/ekiga/trunk/win32 ekiga-win32 cd ekiga-win32

Checked out revision 6170.

# cd ekiga-win32 # make update-sources

Works fine, except for some

svn: Write error: Broken pipe

I get that as well. Actually the update process should be improved, I already have some ideas, but not enough time...

Here's the OPAL / PTLib version I got:

OPAL / PTLib: rev 19973

Replaced -1.9 with -1.10 in Makefile, then

# make

Breaks here:

./src/ptclib/asner.cxx: In member function âPBoolean PASN_BMPString::IsLegalCharacter(WORD)â: ./src/ptclib/asner.cxx:1490: error: invalid conversion from âconst short unsigned int*â to âconst wchar_t*â In file included from ./src/ptclib/asner.cxx:2498: ./src/ptclib/asnber.cxx: In member function âvoid PASN_BMPString::EncodeBER(PBER_Stream&) constâ: ./src/ptclib/asnber.cxx:246: error: invalid cast from type âconst PWCharArrayâ to type âconst wchar_t*â make[2]: *** [asner.lo] Error 1 make[2]: Leaving directory `/usr/local/src/ekiga-win32/ptlib' make[1]: *** [all] Error 2 make[1]: Leaving directory `/usr/local/src/ekiga-win32/ptlib' make: *** [/usr/local/src/ekiga-win32/lib/libpt.a] Error 2 Ok, this was expected...

Let's try again, with the two OPAL and PTLib patches.

As discussed before, I am deleting the entire src folder to be on the save side, then:

# make update-sources

apply both patches

# make

make[2]: Leaving directory `/usr/local/src/ekiga-win32/ffmpeg/libavdevice' : make -C libpostproc install-headers make -C libswscale install-headers make[2]: Entering directory `/usr/local/src/ekiga-win32/ffmpeg/libswscale' Makefile:20: ../subdir.mak: No such file or directory make[2]: *** No rule to make target `../subdir.mak'. Stop. make[2]: Leaving directory `/usr/local/src/ekiga-win32/ffmpeg/libswscale' make[1]: *** [install-headers] Error 2 make[1]: Leaving directory `/usr/local/src/ekiga-win32/ffmpeg' make: *** [/usr/local/src/ekiga-win32/bin/avcodec.dll] Error 2

Now it's getting interesting ...

Now thats strange since I fixed the revision being checked out by ffmpeg. I will try to reproduce.

Without doing anything else, let's do

# make

again and watch. It goes on compiling for some 5 minutes. Then, after that

/bin/sh ./libtool --tag=CXX --mode=compile i586-mingw32msvc-g++ -mms-bitfields -g -DHAVE_CONFIG_H -I. -I./include/opal -I./include -I/usr/local/src/ekiga-win32/include -march=pentium-mmx -DPTRACING -DPTRACING -MT h323.lo -MD -MP -MF .deps/h323.Tpo -c -o h323.lo `test -f './src/h323/h323.cxx' || echo './'`./src/h323/h323.cxx i586-mingw32msvc-g++ -mms-bitfields -g -DHAVE_CONFIG_H -I. -I./include/opal -I./include -I/usr/local/src/ekiga-win32/include -march=pentium-mmx -DPTRACING -DPTRACING -MT h323.lo -MD -MP -MF .deps/h323.Tpo -c ./src/h323/h323.cxx -DDLL_EXPORT -DPIC -o .libs/h323.o ./src/h323/h323.cxx:2926: error: no âbool H323Connection::TransferConnection(const PString&)â member function declared in class âH323Connectionâ ./src/h323/h323.cxx: In member function âbool H323Connection::TransferConnection(const PString&)â: ./src/h323/h323.cxx:2930: error: âTransferCallâ was not declared in this scope ./src/h323/h323.cxx:2935: error: âTransferCallâ was not declared in this scope make[2]: *** [h323.lo] Error 1 make[2]: Leaving directory `/usr/local/src/ekiga-win32/opal' make[1]: *** [all] Error 2 make[1]: Leaving directory `/usr/local/src/ekiga-win32/opal' make: *** [/usr/local/src/ekiga-win32/lib/libopal.a] Error 2

This may be due to latest changes in opal. Maybe the commit was not finished and/or only reverted incompletely? There was a lot going on in opal last two days.

This is when I gave up.

Once I get it running, maybe we should try to get the same revisions running on your setup first? So we know at least we are talking about the same code? Its easy, we jsut have to put the revision number in the makefile...

Regards, Torsten

Matthias

---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.