If I use gaim-2.0.0_beta5 built from the -r1 or -r2 ebuild, it hangs within libnm_glib_init before any GUI appears and is unusable. Location of hang determined using gdb. strace shows an infinite sequence of nanosleeps. I've used gaim-2.0.0_beta5.ebuild out of the CVS attic and rebuilt with that. In binaries produced using that ebuild libnm_glib_init returns successfully and gaim is usable.
It would be helpful if gaim-2.0.0_beta5.ebuild were returned to the portage tree until this can be resolved.
Created attachment 104597 [details] emerge --info output Attached output of emerge --info
Could ya maybe, just maybe, attach the gdb output?
Entire bug report makes no sense ... If you compare the two ebuild you will see nothing major has changed other then the fact you can build either gtkui or consoleui or both. I see no reason to but a stale ebuild back in tree, when your the only one that is reporting such a problem.
There isn't really helpful gdb output, since the program doesn't exit. I step into libnm_glib_init (which is not in a library I've built with debug symbols) and never return. Was there some specific output you had in mind? I took a look at the network manager sources for that function ( NetworkManager/gnome/libnm_glib/libnm_glib.c) and found that it appears to (1) spawn a thread to set up a dbus connection and (2) loop with a sleep until that set up is complete. That is one explanation for strace outputting the nanosleeps.
It may not make sense, but is easily repeatable for me. I'm happy to provide any information you'd like to help make sense of it.
(In reply to comment #6) > It may not make sense, but is easily repeatable for me. I'm happy to provide > any information you'd like to help make sense of it. > My recommendations to net-im herd is to close as later .. I see no steps to repeat not even a proper backtrace of the problem at hand here. Just seems funny only one user is even able produce this bug with as many users we have using and testing gaim and other packages in portage.
Created attachment 104599 [details] GDB command file This gdb command file was used to produce the subsequent attachment.
Created attachment 104600 [details] Output of gdb for -r2 binary with previously attached commands This gdb output shows stack information at the point of the sleep loop. I killed gdb with an interrup to terminate the loop.
Created attachment 104601 [details] Output of gdb for binary created with removed ebuild using previously attached commands The gdb output in this attachment shows that the loop does terminate in the binary built with the ebuild that is no longer in portage.
A similar report appears upstream as issue 1616747. The gaim_report attached to that issue shows a similar stack. http://sourceforge.net/tracker/index.php?func=detail&aid=1616747&group_id=235&atid=100235
I had the same problem too. An strace with gaim shows only nanosleeps and no gui occurs. I've installed networkmanager too. (it's a masked ebuild) After an USE="-dbus" emerge gaim everything works fine. Merry christmas! ;-)
Thanks for your workaround Matthias. I had the same problem and disabling dbus does the trick. Merry christmas!
Stephan and Matthias, could you two please test dbus with version 1.0.2?
Can you reproduce with net-im/pidgin-2.0_beta7 ?
