Re: GNOME 2 Sound Architecture and APIs?

[Vlad Harchev <hvv hippo ru>]

On Thu, 12 Apr 2001, Elliot Lee wrote:

 Hi, 

> On Thu, 12 Apr 2001, Alan Cox wrote:
> 
> > esound is junk. The only thing esd has is a good client API for going boing
> > at approximately the right time. Anything else is beyond it.
> 
> The API has stylistic problems as well (e.g. passing fd's around
> directly), although this isn't a strike against the esound design so much
> as against the present implementation.

 Also AFAIR esd seems to allow only one sample rate for all streams. That's
very bad IMO.

 Have anybody considered NAS? That sound system is very mature, and it's even
implemented in various X Terminals (i.e. in hardware) and X Servers for
Windows! And it's written in C. At least it should be considered - it has very
rich API and functionality, server-side caching of samples, non-trivial
resampling algorithms, support for sound input (writing from a microphone).
Yes, API is slightly complex - but we can make wrappers with gnomefied
function names for common cases. It's how esd should look like :)


PS: I don't have post-permissions to gnome-hackers, so alas this mail will
appear on gnome-hackers with a reasonable delay :(


 Best regards,
  -Vlad