[Freevo-devel] v4l_mga changes

[Jason B. <jas...@ce...>]

Hi all,

I have been looking at re-developing a number of the existing freevo
components with a thought to portability, making it easier to support a
variety of hardware and display models (x11, sdl etc).  The sdl osd was the
first part of this and will undergoe a number of enhancements, namely
support for various display resolutions in full screen mode.  (This works
already, you just need to change the resolution in the code and ensure your
framebuffer/x11 config has the relevent modelines defined.)

The second peice of core-functionality in freevo is the TV app, namely
v4l_mga which as named is specific to the matrox chiipset.  I am looking at
changing this app to support a variety of interfaces, hopefully with an
autodetect for the best (fastest) mode.

I have already started re-coding this app and have the following (almost)
working:

1. Autodetect MGA, and use if if the device is available.

2. Failing MGA, test for x11/xv support.  If xv is supported, check if the
video card supports the required codecs and functions and use it.

3. Failing MGA and XV, use SDL.

SDL obviously provides support for x11, framebuffer, SVGALib etc.  So far
this app is fairly simple and I am trying to maintain compatibility with the
existing app for changing channels etc.  Then comes the fun part, combining
it with the osds for text-over-video displays!  (Why SDL?  Simple, it
provides YUV surface and RGB->YUV conversion capabilities.  Most image
libraries don't.)

I'm curious to hear feedback on wanted/desired functionality in the TV app
as I may be able to include some extras while I'm at it.  This version
should be available for testing in a week or so.

Now for the change management - if we are not going to use branching (which
can work very well, but requires a SINGLE change manager) we should create a
directory structure for contribs and work in progress similar to the
following:

	freevo/WIP
	freevo/WIP/Jasonb 	(me)
	freevo/WIP/Alan
	freevo/WIP/Dirk
	freevo/WIP/Krister
	.
	.
	.

That way we can commit changes to CVS, and advertise for functionality to be
tested PIOR to integrating it into the core application.

Any thoughts ?

Cheers
Jason.