Re: [linux-audio-dev] testing the waters

New Message Reply About this list Date view Thread view Subject view Author view Other groups

Subject: Re: [linux-audio-dev] testing the waters
From: Erik Steffl (esteffl_AT_pbi.net)
Date: Tue Jun 27 2000 - 23:53:41 EEST


Joe Miklojcik wrote:
...
> RedHat and probably others are organized around the idea of being either a
> networked shell/emacs/netscape/gnome workstation or a networked
> web/mail/file/print server. As such, many network services are installed and
> running in the default configuration, for example. Very different choices in

  that's a security issue, but not a performance issue.

> default configuration, and even included software, would be made in a "Linux

  there already are very different choices of configuration...

> Audio Distribution". My RedHat installation is *severely* hacked in the name of
> better rtcsound and MIDI performance. I cut boot time in half or better by
> rooting around in /etc/init.d getting all the default crap out of my way.

  most of the installation procedures (well, I can speak mostly for
slackware and debian, I think redhat too) let you decide what type of
workstation/server you want (debian: barebone, development, server,
etc...) or let you choose from packages (like slackware). so you only
need to add new type to this, if at all.

  for example in debian - it is very easy to install/remove packages and
customize the system. it would be a very good distro to build upon, or
to integrate with.

  the problem I see with new distribution is mainle maintenance - the
distro itself has to be maintained and the distro must be designed so
that the user can maintain h(er|is) machine - that's why I left
slackware and started using debian.

  providing few debian packages would be few orders of magnitude easier
(measured in time spent working on it)

  the services don't matter, you can easily turn them off, you can
supply package that will turn them off (part of your audio packages) or
you can simply ignore them. they do not slow down system, they take up a
small amount of memory but they are eventually swapped out... I mean if
you don't use nfs the nfs server daemon simply sits there, or even not
that, if it's called by inetd...

  just because the ps shows 100 processes does not mean that the
computer is under high load... (you probably know it, I just wanted to
state it explicitly).

  you might have cut the boot time to half but how much performance gain
did you get?

> Not to mention the special kernel hacks and device drivers you need.

  that's what we're trying to avoid by incorporating necessary stuff
into standard kernel (its basically low latency and alsa (once it's
ready), AFAIK).

  anyway, even if they are not in kernel, nothing prevents you to have
custom kernel, which user would install by simply typing apt-get install
audio-kernel (note that apt-get is temporary command line inteface, the
gui will be added later so it will be even easier).

  the actual implemetation of changes might be:

  CD - packaging system would be simply pointed to the CD (for speed)
  internet - packaging system would be pointed to the net (for bleeding
edge updates)

  it's easy to use (you gain basically nothing by having your own
distro).

> Dave's CD seems like a good idea, but if it were accompanied with something like
> Zipslack, then you'd really have something. The success of ZipSpeak comes to
> mind.

  demo CD might be cool idea though...

> RPMs really don't seem like enough, unless you're willing to repackage a lot of
> the RedHat (or SuSE, etc.) base packages, which then makes your suite of RPMs
> somewhat difficult to install unless you're starting with a virgin RedHat
> install.

  why would you want to repackage base RPMs?

        erik


New Message Reply About this list Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b28 : Wed Jun 28 2000 - 01:14:06 EEST