[Bug 735660] v4l2: fix new v4l2 code not working with certain devices (regression)

[GStreamer (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=735660
  GStreamer | gst-plugins-good | 1.4.1

--- Comment #11 from Hans de Goede <jwrdegoede at fedoraproject.org> 2014-08-29 19:37:53 UTC ---
(In reply to comment #7)
> Note the other patch seems incomplete fixed around bug 726195 and 726194. We
> need to write proper interlacing detection code, it's not true that for 1
> resolution driver only do mixed or progressive.
> Would it be possible to move patches to proper bugs instead of having all unrelated issue in the same bug ?

Note the subject of this bug: "v4l2: fix new v4l2 code not working with certain
devices (regression)", so if you wonder why all the patches are together in a
single bug, because the entire set is needed to get things work again / to fix
the regression.

Frankly, I'm quite unhappy about all the flack I'm getting here, don't shoot
the messenger and all that. I just spend a couple of hours this morning to fix
a regression I did not cause, and instead of a thank you I'm getting a lot of
push-back, and no indication of any willingness to actually care about and fix
the regression. Not a great way to win harts and minds IMHO.

No about the technical argument you make:

> Note the other patch seems incomplete fixed around bug 726195 and 726194. We
> need to write proper interlacing detection code, it's not true that for 1
> resolution driver only do mixed or progressive.

Yes some devices may support both progressive and interlaced for a single
resolution. But currently the code is not structured to deal with this, dealing
with this requires at least some refactoring. My patch is not about this, my
patch is a bug fix to the current code, which follows a model where only one of
progessive or interlaced can be supported, and tries to get progressive first.

This current code, which is currently shipping in a stable release, is broken
not at a design level which you're talking about now (it could use improvements
there too). but in the implementation of the chosen design. Even following the
chosen design of a mode being either progressive or interlaced it get things
wrong, with as end result that things do not work at all.

So rather then talking about design issues, which seems to me to be something
for 1.5.x, lets first focus on fixing the regression (ah there is that word
from the summary again), and get the current code to at least work properly for
what it does do, and that is *exactly* what my patch does.

-- 
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.