Re: Status of Echo/Event support

[Takashi Iwai <tiwai@suse.de>]

At Mon, 9 Dec 2002 15:20:21 +0100,
David Olofson wrote:
> 
> > when i write a new driver, it would take
> >
> > - ca. 1 week if there is an OSS driver,
> > - ca. 2 weeks if it's a normal PCI (or ISA) device and there is
> > enough technical info, and
> > - more for exotic environment :)
> >
> > until the first beta release.  a test version may be ealier,
> > though.
> 
> Well, I'll see if I can match those times... ;-)
> 
> BTW, we're talking about som 650 kB of C++ code. Any estimates on 
> what that actually means?

hmm, sorry, difficult to say only from the code size.
the difficulty lies on how exotic the hardware design is.
if there is a similar design, you can reuse the existing driver code,
and it will reduce really much time.

> Anyway, I'll get started as soon as I get an official statement on 
> this "sublicensing as pure GPL" issue. That is, whether I should make 
> this a real driver, or an unofficial "C++ in kernel" hack. I'm not 
> converting all that code if it's never going into ALSA or Linux 
> anyway. (*Maybe* I could use some C++ to C converter, but that's 
> probably even worse than starting with the C++! ;-)

c++ on kernel is yet taboo.
and as long as the driver is written in c, it would be much better to
rewrite it by yourself rather than using some conversion tools.
otherwise you'll get lost what the code does.

although it depends on the architecture, usually the difficult
parts of most of pci sound drivers are

- initialization, especially if a micro-code is required,
- interrupt handling,
- conversion of units (bytes, frames, samples!)


the mixer stuff is usually easy to implement.


Takashi


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf