Subject: Re: [linux-audio-dev] What about debian tasks? [was: Audio distribution revisited]
From: Frank Neumann (franky_AT_viona.de)
Date: Mon Sep 18 2000 - 21:25:17 EEST
Hi,
Guenther Geiger wrote:
> > Now so far there have be 3 debian users that have spoken up and agreed that a debian-task
> > would be a useful thing and would like to work on it. Which is plenty enough for me to feel
> > that its creation is justified.
> >
>
> Here's number four.
> Let's go ahead with it.
I don't want to sound like the AOL-style "me too" stereotype, but personally
I think reusing an existing distribution like Debian is the more wise way.
My own experience in this direction is that I started the creation of the
Debian/m68k port (Amiga, Atari etc). That was back in 1995 - I think - when
all we had at that point was a handful of .tar.gz images plus a few outdated
readme's to install the beast. No packaging system, no dependencies, nothing.
A nightmare. So we started discussing what way to go. At first I was planning
on doing something on my own, and made a list of packages "my" distribution
should contain under all circumstances. This list quickly grew due to others'
suggestions, then we came to dependencies, conflict situations, file system
hierarchy discussions, and eventually someone fortunately pointed me at Debian.
We decided to go this way. I believe it was a good idea. Still, it took more
than 3 years until we had a distribution that could "easily" (from our
point of view) be installed. Imagine there were only a handful of developers,
and the user base was only slowly growing.
Ok, to make a possibly long story short: IMHO deciding to create your own
distribution from scratch nowadays is suicide. It's just too many things
out there to watch for.
The "task" package idea sounds very reasonable, and with the addition of
apt one could e.g. set up a SourceForge project where the "AudioDebian"
packages would be stored (of course excluding those that are free anyway and
can become part of the core distribution). Even different architectures
like PPC could be handled easily, as is the case today at the main Debian
ftp/web servers.
Also, some critical packages could be upgraded easily. What came to my mind
first is the kernel itself - right now there are frequent new LL patches
from either Ingo Molnar or Andrew Morton, and for sure not many people who
basically want to make music would like to apply patches, compile, install
modules etc. It would be rather easy for them to e.g. say
# apt-get install kernel-image-ll-2.2.18
and after a few seconds (or minutes, depending on your connectivity)
all you would have to do is reboot the system to yield even lower latencies
than with the previous kernel version.
Ok, dreaming aside, my current belief is what others have said here too:
First we need applications, stable APIs, and once again good (killer?)
applications. Right now I feel like there's too much diversity with regard
to look and feel, and too many promising, but only half-finished products
(this includes myself, unfortunately).
I don't want to discourage the M-station team - I guess they very well know
what musicians need and what to do, and I wish them all the best of luck
in their efforts - but for now creating applications is my main focus.
After that, going for a Linux audio distribution based on an existing
one like Debian sounds like a swell idea to me.
(My dream is a set of programs that all have the same GUI toolkit below
them - GTK+. And among these programs I mostly need a good MIDI sequencer
and a good sample editor - Cubase&Soundforge).
Enough rant,
Frank
This archive was generated by hypermail 2b28 : Tue Sep 19 2000 - 00:24:24 EEST