[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Danger Will Robinson! Danger!



> On Mon, Mar 13, 2000 at 11:02:04AM -0500, Mark Mealman wrote:
> > I really don't like unstable either, but I've pretty much abandoned the stable tree as too behind the times back when slink was nearing freeze.
> 
> Here's a serious question for you: which parts are too old on slink
> to perform the functions you need? Seriously?

Well off the top of my head I need the 2.2 series kernel for Compaq Smartraid controller support.

But it's as much principle as anything else.

When apache 1.3.11 is on the streets, there's little excuse to be running 1.3.1 on a production server.

95% of the security notices I see are on versions of software far older than anything in the unstable tree.

> I only just upgraded two of my slink boxes to potato on the weekend,
> and it turns out that I didn't even need to. A friend of mine still
> has a hamm box; before that it was a rexx box. Works fine, no need to
> upgrade.
> 

It's great that your friend can run systems on software that's 2 years old.

Most of us are required to be state of the art on production systems.
 
I'm not alone in my displeasure of these long release cycles. I've been running unstable on production web servers for over a year and a half now.

And quite frankly stable is useless on workstations. xmms, gnome, kde, netscape, window managers are changing on a monthly, sometimes weekly basis. New software is hitting the net at rapid speed and being added into the Debian unstable archives in a timely manner, but because a user is pointed at stable he or she simply doesn't have access to it.

Debian unstable is on par, stability-wise, with most major Linux distrubutions. We could outright kill the need for a "stable" Debian if we had a way to ensure that critical packages didn't break and there was someway to "roll back" to an older package version should the new one be junk.

Or we could create a development branch of Debian that filters packages into unstable after the packages have been "cleared" of containing any box-stopping bugs.

I think if you did a poll you'd find most Debian users run unstable on workstations despite warnings of "This can trash your system at any time".

That tells me there is a need to go beyond a totally stable / totally unstable development mindset.

-Mark


Reply to: