[gtk-osx-users] Program crash using GTK-OSX



I got a following crash backtrace from the user of my program:

11:49:30.923 6   libSystem.B.dylib                   0x939646e4 abort + 93 ()
11:49:30.923 7   libintl.8.dylib                     0x013cca7f _nl_load_domain + 95 ()
11:49:30.923 8   libintl.8.dylib                     0x013cbc4d _nl_find_domain + 717 ()
11:49:30.923 9   libintl.8.dylib                     0x013d0dfc libintl_dcigettext + 1884 ()
11:49:30.923 10  libintl.8.dylib                     0x013cb8e9 libintl_dcgettext + 73 ()
11:49:30.923 11  libintl.8.dylib                     0x013cb967 libintl_gettext + 39 ()
11:49:30.923 12  libglib-2.0.0.dylib                 0x0129b221 _g_dgettext_should_translate + 129 ()
11:49:30.923 13  libglib-2.0.0.dylib                 0x0129b31f g_dgettext + 31 ()
11:49:30.923 14  libgtk-quartz-2.0.0.dylib           0x0099d83a gtk_get_option_group + 90 ()
11:49:30.923 15  libgtk-quartz-2.0.0.dylib           0x0099daa6 gtk_parse_args + 150 ()
11:49:30.923 16  libgtk-quartz-2.0.0.dylib           0x0099db64 gtk_init_check + 36 ()
11:49:30.923 17  notecase-bin                        0x002dedb2 _ZN3App7StartupEiPPc + 4258 ()
11:49:30.923 18  notecase-bin                        0x000023a4 main + 36 ()

User uses OS X v10.6.8, and I've built the program using OS X 10.7.5 (32-bit architecture, GTK2 from recent stable moduleset), building with XCode v4.3.3.

The closest crash report to that one I could find is this: https://trac.adium.im/ticket/15996

In that link developer finds a following problem with the code:
"Cross-compiling for 10.6 using the darwin11 gcc42 leaves PTHREAD_RECURSIVE_MUTEX_INITIALIZER defined in libintl's config.h, but that macro isn't defined in the stdc headers in the actual 10.6 SDK so it fails an assert. Apple didn't put a check in there to define that symbol as 0 if the deployment target is less than 10.7, unfortunately. Someone on the gcc team filed a radar on it that is supposedly fixed in 10.8: rdar://10784745
I'm writing a patch to define that symbol to 0 when build.sh builds it, as the default behavior (to initialize the mutex instead of checking the result of setlock first) is safe anyway on Lion (and, for that matter, the recursive POSIX threading stuff as a whole seems to be kind of buggy in darwin11)."

Any tip what can I do to fix the issue?
Is downgrading to XCode 4.1 the solution ? Can it be installed in parallel ?

Regards,
  Miroslav


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]