Bug#428674: glib2.0 testsuite failures on kfreebsd-*

[Petr Salinger]

> glib2.0 doesn't pass its testsuite on kfreebsd in unstable and in
> experimental versions, on kfreebsd-i386 and on -amd64:
>    <http://experimental.ftbfs.de/fetch.php?&pkg=glib2.0&ver=2.12.12-1&arch=kfreebsd-amd64&stamp=1178469721&file=log&as=raw>
>    <http://experimental.ftbfs.de/fetch.php?&pkg=glib2.0&ver=2.13.4-1&arch=kfreebsd-i386&stamp=1181732513&file=log&as=raw>
>
> I don't plan looking into this, but wanted to document this in a bug
> report so that we don't forget about it.  It would be nice if a porter
> could look into this; glib2.0 is used in a lot of apps, and perhaps
> this is the source of more than this failure.

I looked on this on kfreebsd-i386, it works for me without failure.

**************
    DEBUG_MSG (("[unused] stopping unused threads"));
    g_thread_pool_stop_unused_threads ();
    DEBUG_MSG (("[unused] waiting ONE second for threads to die"));
    /* Some time for threads to die. */
    g_usleep (G_USEC_PER_SEC);
    DEBUG_MSG (("[unused] stopped idle threads, %d remain, %d threads still exist",
                g_thread_pool_get_num_unused_threads (),
                test_count_threads ()));

    g_assert (g_thread_pool_get_num_unused_threads () == test_count_threads ());
    g_assert (g_thread_pool_get_num_unused_threads () == 0);
**************

"ONE second" might be enough for my system, but may be not for loaded 
system on a buildd. Also, some thread might ended during evaluation of
each assert side. Definitely, it needs further investigation.

Petr