From kde-multimedia  Sat Feb 22 21:54:42 2003
From: Stefan Westerfeld <stefan () space ! twc ! de>
Date: Sat, 22 Feb 2003 21:54:42 +0000
To: kde-multimedia
Subject: Re: aRts vs JACK
X-MARC-Message: https://marc.info/?l=kde-multimedia&m=104595096330384

   Hi!

On Fri, Feb 21, 2003 at 09:22:01PM +0100, Jozef Kosoru wrote:
> A long time ago I wrote ALSA-0.5.x output support for the aRts and forced to
> make an abstraction layer for the output routines (at those times, there
> was OSS support only). I guess it shouldn't be a problem for me to write
> the JACK output plugin. It'll be an additional layer for the aRts - but
> that's ok; aRts has a huge latency anyway and for playing kaboodle it
> really doesn't matter.
>  
> ...but my opinion is still same: aRts is overdesigned, too slow, 
> unnecessary complex and has an unacceptable latency even on fast CPUs. 

It depends on what you want to do with it. If you want to play a single
audio stream from an application, thats certainly true. If you want to play
two audio streams from two applications, yes, its also true. If you want
to play a wavefile serverside, mix it with six audio streams from four
applications and have appropriate routing and effects inbetween, aRts fits.

For noatun, I think its not harmful, but greatly helpful, because most of
what you need to do for a media player is done in artsd, and not much less
efficient at that.

Still: please do go for CSL, then. Its written for people like you who do
want to keep every bit of efficiency.

  http://www.arts-project.org/doc/csl-0.2.0.html

> Ok, replacement is not possible but I would vote for cutting the aRts
> sources to the minimum. How many KDE users will ever need aRts builder and
> all those things? Is there any logic to have a realtime sythesizer as a
> default sound server in the desktop enviroment? It's like to have a 3D
> modeling tool instead of KPaint or MATLAB instead of KCalc.

Cutting away features only some people need doesn't help. You need to ensure
that

(a) features that only some people need will be available and integrate
    smoothly
(b) features that all people need will not suffer (too much) under additional
    flexibility

Look at the size of the kernel source tree. You would not throw out some
drivers, just to reduce the size of the code. But you might make things
dynamically loadable that only some people need, and ensure that they do not
consume resources for those that don't need them.

   Cu... Stefan
-- 
  -* Stefan Westerfeld, stefan@space.twc.de (PGP!), Hamburg/Germany
     KDE Developer, project infos at http://space.twc.de/~stefan/kde *-         
_______________________________________________
kde-multimedia mailing list
kde-multimedia@mail.kde.org
http://mail.kde.org/mailman/listinfo/kde-multimedia