[Bug 752746] harness: Keep synchronized events in sync with buffers

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=752746

--- Comment #3 from Olivier Crête <olivier.crete at ocrete.ca> ---
I'm totally favourable to magically forwarding events, the problem is that the
events get forwarded in the streaming thread, so the unit tests is not in
control of when they are forwarded, leading to races. The attached patch
forwards the event automatically on _pull() if you have a sink_forward_pad, so
in that case, it should not change the behaviour.

We could rename the current gst_harness_pull_event() to _pull_event_oob() and
make the synchronized one into the main pull_event() function and in it just
drop all buffers until an event is received, a bit like it drops or forwards
events if you wait for a buffer.

My test sends the original caps event only with the first buffer, not on
gst_harnness_play(), and the synchronized events cause the clock id to be
unscheduled, which breaks the crank, as I have no way to know if those events
have been forwarded or not.

But before we commit to the API for 1.6, I'd really like to fix this thing
somehow. I'll fix any tests that have been upstreamed, tests that haven't don't
count!

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.