Re: Beta-testing and the glibc 2.1 (Was: Missing ldd? Have libc6 on hold? Get ldso from slink...
On Wed, 17 Mar 1999, Seth M. Landsman wrote:
> Okay, let's not turn this into a flame war. My point is that
> breakages in unstable are *REALLY BAD THINGS*.
I have no intention of flaming anybody; from my standpoint, at least,
everything I say is sober and reasonable. Everyone else's opinions may,
and likely do, differ. And yes, breakages in unstable are bad things,
even really bad. But that's what's very likely to happen when you upgrade
the major libraries.
> If you manage to alienate the community of people
> who do real work, debian won't be tested on real work machines and won't
> be as stable as it should be. You haven't addressed this point.
Just like development copies of other public-beta OSes, Debian does get
tested on real work machines, in unstable, frozen, and stable forms. But
what it boils down to with me is prioritization: for *this* particular
machine which I administer, which is more important -- helping the Debian
folks test out the new distro, or keeping my backside out of the fire and
my blood pressure in the three-digit range? When I lean toward the latter
(which is most of the time), I don't stray from stable, and help out
elsewhere.
> Oh, and my 20 gig DDS is in the shop. Sorry, telling people "back
> up often" is getting less realistic. The fact is, my important stuff is
> backed up in five places, but reinstalling is still a pita.
At the time I did my backup, I had no working tape drive -- for me, at
least, a lack of tape backup drives or media doesn't cut it. I mounted a
Windows partition and backed up into a compressed tar file there. And
there was no actual reinstall pita for me; again, I booted off a rescue
floppy, untarred my backup, and that was it. As freqently as I have
Windows crashes, I've gotten used to backing stuff off to other drives in
the event of massive failures.
> Most people don't have that extra machine sitting around.
Possibly, but I wouldn't assume that most people who are:
(a) testing potato, and
(b) have really important data
have only one machine.
--
Mike Renfro / Instructor, Basic Engineering Program
931 372-3601 / renfro@tntech.edu
Reply to: