[PATCH weston v2 4/6] tests: Convert keyboard-test to new test extension

[Pekka Paalanen]

On Sun, 9 Dec 2012 15:43:18 +0000
"Eoff, Ullysses A" <ullysses.a.eoff at intel.com> wrote:

> >-----Original Message-----
> >From: Pekka Paalanen [mailto:ppaalanen at gmail.com]
> >
> >On Fri,  7 Dec 2012 13:50:32 -0800
> >"U. Artie Eoff" <ullysses.a.eoff at intel.com> wrote:
> >
> >> +void
> >> +yield(struct client *client)
> >> +{
> >> +	/*
> >> +	 * FIXME: ugh! how do we ensure all events have finished
> >> +	 * propagating to the client.  The calls to usleep seem to do a
> >> +	 * pretty reasonable job... and without them, tests can fail
> >> +	 * intermittently.
> >> +	 */
> >> +	usleep(0.02 * 1e6);
> >> +	wl_display_flush(client->wl_display);
> >> +	wl_display_roundtrip(client->wl_display);
> >> +	usleep(0.02 * 1e6);
> >> +}
> >
> >This FIXME is still here. Did I guess the cause wrong in my earlier
> >comment, or have you just not had the opportunity to look into it
> >yet?
> >
> 
> Yes.  I just haven't had time to look into it yet.  Since it works, I was
> hoping for this patch series to make it into upstream first
> so that it doesn't bit rot in my tree.  I'm not sure how much the
> event propagation plays into the repaint mechanism as you may
> suspect.  Those mechanisms are still a bit unfamiliar to me. When
> time permits I plan to investigate.

Ah ok, cool. I was just wondering since I couldn't find any comment
about it.

Repaint should have quite a big effect. For instance, window geometry
is not updated until repaint, so input events will be routed according
to the old geometry, IIRC. It's not really about event propagation as
it is about when new state is applied.


Thanks,
pq