[Bug 739281] video-blend: correct improper logic calculations

[GStreamer (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=739281
  GStreamer | gst-plugins-base | unspecified

Tim-Philipp Müller <t.i.m> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #289864|none                        |reviewed
             status|                            |

--- Comment #30 from Tim-Philipp Müller <t.i.m at zen.co.uk> 2014-11-03 11:10:02 UTC ---
(From update of attachment 289864)
Thanks for writing these Vinneth! Definitely goes in the right direction, but
is not quite complete yet.

One question: you create overlays of 200x50 pixels, but then in
gst_video_overlay_rectangle_new_raw() scale them to target width/height 50x50 -
is that on purpose? (It's fine of course, was just wondering if it was
intentional)

Currently the tests just make sure that we don't crash with the various
parameter combinations given. What's missing in the test is then that we fill
the overlay rectangles with actual data (ideally with alpha=1 and not alpha=0),
and then later check that blending actually took place, in the way that we
expected, by looking at some of the "edge pixels" in the video surface we blend
the rectangle onto.

This requires two things: (a) filling the overlay rectangle strategically with
values, and (b) using a clean background image for every blend operation so we
start with a new slate.

I think perhaps a small helper function that creates the background surface,
and the overlay rectangle, and blends it onto the x/y/width/height provided
would be useful, then the checks can just call that and afterwards check if the
video surface was changed or not changed as expected, and then unmap/free it.

-- 
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.