New issue 11 by Christia...@gmail.com: Parallel build is broken
http://code.google.com/p/gmtk/issues/detail?id=11

The fact that libgmtk requires libgmlib is not shown in the Makefile,  
making the build sometimes fail when building in parallel.

Patch attached.

Attachments:
        gmtk-1.0.6-parallel_build.patch  1.2 KB

Comment #1 on issue 11 by kdeko...@gmail.com: Parallel build is broken
http://code.google.com/p/gmtk/issues/detail?id=11

I'm not sure I can use this patch. I want libgmtk to be dynamically linked  
to libgmlib. I believe your patch statically links them.

I put in another patch that may fix this so that it does what I want.  
However, even without the patch make -j4 always worked on my setup. So can  
you see if gmtk (SVN) compiles correctly for you now.

Comment #2 on issue 11 by Christia...@gmail.com: Parallel build is broken
http://code.google.com/p/gmtk/issues/detail?id=11

It dynamically links to libgmlib if a .so version of the library is  
available.

Well, it never failed to me neither until today it did once ;-) So I can't  
really test. But your patch should also do the trick.
I would close the bug if I could see how to...

Comment #3 on issue 11 by kdeko...@gmail.com: Parallel build is broken
http://code.google.com/p/gmtk/issues/detail?id=11

(No comment was entered for this change.)

Comment #4 on issue 11 by Christia...@gmail.com: Parallel build is broken
http://code.google.com/p/gmtk/issues/detail?id=11

"gcc: error: /home/abuild/rpmbuild/BUILD/gmtk-1.0.7/src/.libs/libgmlib.so: 
No such file or directory"

I guess the .la file is created before the .so one.

PS. Seriously, I can't reopen/close a bug in this thing??

Comment #5 on issue 11 by kdeko...@gmail.com: Parallel build is broken
http://code.google.com/p/gmtk/issues/detail?id=11

I opened this back up.. but I have to say that on my quad core machine I  
can't duplicate this issue you are having...

[kdekorte@quad gmtk]$ make clean
Making clean in po
make[1]: Entering directory `/home/kdekorte/svn/gmtk/po'
rm -f *.pox gmtk.pot *.old.po cat-id-tbl.tmp
rm -f .intltool-merge-cache
make[1]: Leaving directory `/home/kdekorte/svn/gmtk/po'
Making clean in src
make[1]: Entering directory `/home/kdekorte/svn/gmtk/src'
test -z "libgmlib.la libgmtk.la " || rm -f libgmlib.la libgmtk.la
rm -f "./so_locations"
rm -f "./so_locations"
rm -rf .libs _libs
rm -f *.o
rm -f *.lo
make[1]: Leaving directory `/home/kdekorte/svn/gmtk/src'
Making clean in .
make[1]: Entering directory `/home/kdekorte/svn/gmtk'
rm -rf .libs _libs
rm -f *.lo
make[1]: Leaving directory `/home/kdekorte/svn/gmtk'
[kdekorte@quad gmtk]$ make -j4
make  all-recursive
make[1]: Entering directory `/home/kdekorte/svn/gmtk'
Making all in src
make[2]: Entering directory `/home/kdekorte/svn/gmtk/src'
   CC     libgmlib_la-gm_common.lo
   CC     libgmlib_la-gm_audio.lo
   CC     libgmlib_la-gm_file.lo
   CC     libgmlib_la-gm_log.lo
   CC     libgmlib_la-gm_parse.lo
   CC     libgmlib_la-gm_pref_store.lo
   CC     libgmlib_la-gm_strfuncs.lo
   CC     libgmtk_la-gmtk_common.lo
   CC     libgmtk_la-gmtk_audio_meter.lo
   CC     libgmtk_la-gmtk_media_player.lo
   CC     libgmtk_la-gmtk_media_tracker.lo
   CC     libgmtk_la-gmtk_output_combo_box.lo
   CCLD   libgmlib.la
   CCLD   libgmtk.la
make[2]: Leaving directory `/home/kdekorte/svn/gmtk/src'
Making all in po
make[2]: Entering directory `/home/kdekorte/svn/gmtk/po'
make[2]: Nothing to be done for `all'.
make[2]: Leaving directory `/home/kdekorte/svn/gmtk/po'
make[2]: Entering directory `/home/kdekorte/svn/gmtk'
make[2]: Leaving directory `/home/kdekorte/svn/gmtk'
make[1]: Leaving directory `/home/kdekorte/svn/gmtk'

Comment #6 on issue 11 by Christia...@gmail.com: Parallel build is broken
http://code.google.com/p/gmtk/issues/detail?id=11

I know, it's not easy to trigger, bug the logic dependency problem is clear.

Test the original patch and you will see libgmtk still has a DT_NEEDED  
entry for libgmlib (it's dynamically linked).

Comment #7 on issue 11 by kdeko...@gmail.com: Parallel build is broken
http://code.google.com/p/gmtk/issues/detail?id=11

Ok, I went and added your patch to the code. But I would have liked a test  
case that reproduced the original problem 100% of the time.