Re: stk and the Dreamcast

On Tue, 2003-07-22 at 12:36, Mike Melanson wrote:
> On 22 Jul 2003, Darren Vincent Hart wrote:
> 
> > I have written an experimental video output plugin for a set-top toolkit
> > I have been working on (www.libstk.org) and would like to present it for
> > distribution with xine.
> 
> 	Hmm, okay, so you've got my attention with this stk stuff. I am
> one of the principle xine developers (been dormant for some time). I have
> also been working on a Sega Dreamcast media player on-and-off for some
> time now. The player leverages code from xine (mostly demuxers and
> decoders) but is not a straight port of the whole app. The part I am stuck
> on right now is making some kind of UI.

I think we can help you there *g*.

> 	A little more information: For amateur Sega Dreamcast development,
> many developers base their projects on the free, open source KOS platform:
>   http://cadcdev.sourceforge.net/
> The Sega DC has a 200 MHz Hitachi SH-4 CPU running its show, and a PowerVR
> graphics chip to render 3D graphics into a framebuffer for display.
> Straight pixel drawing is allowed. But for drawing, e.g., text, it makes
> more sense to load a font map as a texture into video RAM and ask the chip
> to render 8x8 or 8x16 blocks from the map.
> 
> 	I wonder what are the chances of getting stk to run on the DC?
> Does stk do everything pixel-by-pixel or is it aware (or made aware) of
> certain underlying specialty hardware?

Libstk has an abstract base surface classes which all backend surface
classes derive from.  So pixel-by-pixel drawing routines are used only
if the backend class hasn't overridden them.  We would have to write a
stk::surface_dreamcast backend class (we currently have
stk::surface_sdl, stk::surface_dfb, and stk::surface_fbdev).

We would be interested in helping where we can.  Let us know what we can
do (besided writing better libstk documentation, we are working on that
:-) ).

If you are interested, feel free to join the libstk mailing list,
subscription link on the contact page on the website
(http://www.libstk.org ).  It is currently under heavy development, but
there are several running test_apps and the design is pretty much nailed
down.  We have one developer using it to control nitride embedding
furnaces for automobile manufacturer with a industrial pc and a
touchscreen - not our intended target, but hey.  The DC sounds like a
great platform for libstk.

Darren

> 	BTW, after today, I am out of town for the next week, in case you
> respond soon...
> --
> 	-Mike Melanson
>