Re: How stable is SiD ?
"Karsten M. Self" <kmself@ix.netcom.com> writes:
> on Sat, Sep 06, 2003 at 01:23:43PM -0400, Johan Kullstam (kullstj-ml@comcast.net) wrote:
> > Paul Johnson <baloo@ursine.ca> writes:
> >
> > > On Fri, Sep 05, 2003 at 03:27:47PM +0200, Joris Lambrecht wrote:
> > > > Can anyone advise on starting to use SiD as resource for my Debian
> > > > Workstation ? Doesn't it have to many issues left open, broken
> > > > dependencies etc.
> > >
> > > If you have to ask, sid is not stable enough for you.
> >
> > Perhaps his video card isn't supported with the woody shipped xfree86?
> > That's why I went straight for sid last August. (ATI Radeon 8500
> > needs 4.2.1.) At work I just got some crap corporate box with i845g
> > graphics. I need xfree86 4.3. Please advise.
>
> 1. Report this problem to your video card vendor. I don't care if
> you're CTO of IBM or a one-week-temp and Spam-R-Us or Joe's Internet
> Taco Cafe. Vendors need to know that their customers are using
> their products with GNU/Linux, and demand XFree86 support. Do this
> regardless of any subsequent steps you take to resolve your
> problem.
XFree86 supports it since February this year. Unfortunately, debian
doesn't have 4.3.0 in its lineup. Woody uses 4.1 which came out in
June 2001. XFree86 seems to put out reasonably timely releases.
Where do I point these card manufacturers to a time machine vendor so
that I can get proper debian support?
> 2. Consider the option of using a supported card. It's not going to
> cost much -- last time I tried this trick, it consisted of walking
> into CompUSA and asking for the lowest-end video card they had
> (suitable for most office work; who are you kidding about your
> Gnumeric FPS score), cost was on the order of $25. Submit the
> receipt for reimbursement, and pass along _this_ request to your
> internal purchase manager or whitebox vendor. Cannibalizing an
> existing dead box might get you the card for free.
This is what I did. I ordered a cheap-ass card. This is my own money
for a machine doled out to me at work. No, I do not have control over
what they give me. I am very on the edge even daring to run linux.
> 3. Go ahead and use 'testing'.
I am using sid on my home machine to support the radeon 8500.
> You're insulated from most of the
> borkenness of unstable.
On the other hand using testing at work and sid at home i've noticed
that things break more often in sid, but when broken in testing, they
can stay busted in testing for a long time. I am not sure which is
better. While the theory gainsays it, I am leaning towards sid being,
in practice, more stable.
> Using 'pinning' and apt preferences, you
> can include unstable and testing sources, pull from testing by
> preference, but install selected unstable packages on an as-needed
> basis. This is what I do. It works most of the time. Note that
> using pinning to bridge the testing/unstable gap works pretty well
> (they're relatively close), but bridging stable to either testing or
> unstable is a real mess. In the latter case, you're trying to
> bridge 1-2 years of software development, often with large
> changes in basic foundations.
> Based on two minutes' hunting through Google and Google Groups, you're
> going to need XF86 v 4.3, which may not yet be in testing,
And neither is it in sid/unstable.
> but is
> available via unofficial debs from http://www.apt-get.org
>
> http://tinyurl.com/mhvh
>
>
> The general answer is this:
>
> GNU/Linux remains disadvantaged by hardware support policies,
If by this you mean that GNU/Linux stable cannot support hardware
younger than about 3 years old due to a very slow release cycle
policy, then yes, I would agree.
> many
> of which have been historically shown to be influenced improperly by
> Microsoft.
You can blame MS for a lot, but this one is squarely on debian's head.
Yes, I could perhaps lend a hand instead of bitching. However, my
complaint is about the attitude behind "If you have to ask, sid is not
stable enough for you." comment more than it is about the debian X
strike force.
> Free software is a powerful tool for circumventing this
> situation, but it requires a positive response on your part:
> telling the vendor that you want support, researching whether and
> how support is available, and taking necessary steps to configure
> your system. If you're not willing to do this, then you're better
> off sticking to known supported hardware, or accepting the rule of
> an illegal software monopoly over your computing platform.
> Resistance isn't painless. It has its rewards.
I settled on compiling the xfree86 4.3.0 source myself and installing
over top of what I had. This seemed to be the easiest way to keep my
"testing" library set. It works for now, but I can't get the video
modes I want. (I have also ordered a suitibly obsolete and cheap AGP
video card which is supported so this will be taken care of shortly.)
Look, when I buy my own machine, I do research it to see if it has
linux/xfree86 support. The i845g box just came to me at work. It's a
large corporation and they just issue these things. I would *never*
have chosen that thing myself.
> If you're going to pioneer Debian GNU/Linux at your workplace, you'll
> have to balance:
>
> - Accepting supplied HW and current levels of support.
> - Risking the slightly less stable world of testing or unstable
> releases of Debian.
> - Learning something about your tools.
> - Using alternative hardware where necessary.
You forgot perhaps
- Using an alternate linux distribution
I was using redhat up to last August, I can always go back. I have
been playing with gentoo on a spare ppro box at home at it looks
pretty nice too. Both of these have support for the i845g via an up
to date XFree86.
I think it's wonderful that debian can support so many platforms,
e.g., 68k, sparc, powerpc, &c. However, it does suck that debian has
so much trouble supporting the much more common scenario of x86
hardware created in the past couple years.
> Inflexibility on all four points on your part isn't a problem on
> ours.
Perhaps. But rather than the flames the original poster got about
feasibility of testing or sid, some amount of accomodation of
*legitimate reasons* to run testing/sid should be allowed.
> Peace.
Likewise.
--
Johan KULLSTAM
Reply to: