iPAQ port

[Jean-Paul Saman]

Christophe Massiot wrote:
> 
> À (At) 11:08 +0200 17/10/01, Jean-Paul Saman écrivait (wrote) :
> 
> >Other things that are needed are:
> >- The default screen size should match the iPaq LCD screen
> 
> In the latest cvs the default screen size is the video size, so you
> don't need scaling.
> 
> >- A landscape mode for vlc would be nice
> 
> X configuration is your friend :-).
> 
> >- The default userinterface gtk should be sized with in the LCD screen
> >size. For easier access to the controls and dialogs.
> 
> Yep.
> 
> >- No floating point arithmic, because the strongArm does not support
> >hardware floating points calculations.
> 
> Yep.
> 
> >A little history of what I'm doing now.
> >For my employer I am creating a demonstration for a wireless streaming
> >video solution.
> 
> Welcome to the club :-))).
> 
> >I use an iPaq, WLAN card (3COM AirConnect), WLAN AP also
> >3Com. The software I'am using for streaming video is from VideoLan (the
> >vlms and vlc).
> >The Wireless LAN can cope with a throughput of about 4 MB/s, so I had to
> >scale down some DVD to less then that. The overall image quality is
> >acceptable for such a small screen.
> 
> Well personnally I use MPEG-1, you do not need the DVD resolution.

I know but vlms did only support MPEG-2, so stuck with that for the time
being.

> 
> >The demo works now, but is not stable enough and lacks audio. Currently
> >I'm trying to get the WLAN drivers working better and after that I'll
> >have some time on my hand to help out for vlc.
> >
> >The time I got allocated to work on this (pet) project is limited to one
> >day a week. If I do not have other assignments, then I can spare more
> >time to help out.
> >
> >Where can I help?
> 
> As I said there are several items to do. I know why audio isn't
> working, the current audio decoder is FPU-based and is waaaay too
> slow on iPAQ. The solution I see is to use Henri's new plug-in
> architecture to create a new audio decoder based on libmad.
> 
> As for the optimizations, the YUV can be easily optimized (the code
> has already been written for BBPlay, but lacks scaling). IDCT could
> be too, but we have to write it from scratch.
> 
> Finally, we need a new interface.
> 
> What do you want to do ?

I could start working on creating the libmad based on the new plug-in
architecture.  
Where can I find code for libmad? Is it base on madplay?
How do the interface specs look for the new plug-in architecture?

> 
> --
> Christophe Massiot.