GNOME Bugzilla – Bug 317043
basesink preroll_queued loses buffer sometimes when bufferprobes are used
Last modified: 2005-11-27 14:24:07 UTC
about 1 out of every 10 runs of gst-python-s testsuite/test_pad.py, the preroll_queued drops from 1 to 0 during the state commit. I changed the g_assert to give an ERROR: /* FIXME: a pad probe could have made us lose the buffer, according * to one of the python tests */ if (length == 0) { GST_ERROR_OBJECT (basesink, "preroll_queued dropped from 1 to 0 while committing state change"); } g_assert (length <= 1); } but this probably needs a better fix.