Re: GNOME Sound Server, was Re: GNOME sounds - which component

[sopwith redhat com (Elliot Lee)]

On 27 Jul 1999 00:08:07 -0400, Colin Davis <cdavis@thepentagon.com> wrote:
>  Would it not be possible to create a new sound server, but use the
>Esound API, thus a wrapper is not needed, and people will coninue to
>write to Esound, instead of having to choose.
>  If this is not done, I fear people will just continue writing to
>Esound, as that will work with this proposed server, and maintain
>compatibility with Esound.
>
>Again, you said that the Esound code is poor. Is there anything that is
>so terrible wrong it cannot be overlooked about it's API?

Yes.
	No synchronization support.

	The method of specifying stream format isn't extensible
	to arbitrary bit depths.

	Lack of data transfer abstraction means we will have
	to do horrible hacks to make a wrapper.

	Doesn't provide a sufficiently useful method of examining
	and controlling client connections.

	Isn't sufficiently object-oriented to be what I consider a
	clean API. (not a reason in itself, but it adds weight to the
	argument.)

Is there any further convincing you need? I could read the code aloud for you
at the local library's storytime, but then I'd get sued by parents for
frightening their children... ;-)

Ignoring implementation issues, the libesd API does the job for maybe 80% of
the uses out there, and the other 20% are utterly impossible and desparately
needed.
-- Elliot