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

Re: How to make Debian more attractive for users, was: Re: The number of popcon.debian.org-submissions is falling

"Jesús M. Navarro" <jesus.navarro@undominio.net> writes:

> But once you forget your expectancies and put yourself under the skin of
> a newcomer, Sid breaks and sometimes breaks hard (no other thing should
> be expected -in fact, I feel sometimes that Sid breaks "too little"
> because due to the fact that so many people use it for practical
> purposes package upgrades tend to be not as much aggressive as it could
> be otherwise).  A bit to a lessen extent the same can be said about
> Testing.

I don't agree; I think it's very hard to say the same thing about testing.

Yes, sid sometimes breaks hard, although I think if you've been running
Linux for a few years the degree to which sid really breaks is somewhat
exaggerated.  I've never had something happen in sid that risked real data
loss, for instance; I know we've had cases, but I think they've been
really rare.  I've had an unbootable system where I needed to boot from a
rescue CD I think once, and a few cases where X didn't start until I
rolled back some package upgrades.  For breakage, that's not bad.

But on testing, it's been rock-solid for me for years.  It's not just
somewhat less breakage.  I think it's almost no breakage.  Occasionally
packages get stranded for a long time at back revs because of various
migration problems, and once or twice I've had to pin something (usually
because of non-free drivers like fglrx or nvidia that aren't really part
of Debian), but it's an experience that I can comfortably recommend.

I do think that it's hard to run Debian testing unless you have someone
around who knows Linux enough to figure out what's going on with a package
upgrade occasionally, but I think the same thing is true of running
Ubuntu, Fedora, etc.  That's more a problem with Linux on the desktop in
general, and I don't think Debian is any worse off than others.  (I'm also
dubious it's really worse off than Windows either; it's just that more
people know how to unwedge broken Windows systems enough to get them to
limp along than know how to fix Linux systems.)

> If anything Sid/Testing could be compared to a "rolling release"
> distribution ala Gentoo or Arch but not to any "fast releasing" like
> Fedora or Ubuntu.

No, having run both, I honestly think Debian testing is a superior
experience to Ubuntu.  It gets packages roughly as fast, with some
advantages both directions, but it's more reliable.  Packages in Ubuntu
universe break all the time, and worse, they release broken, and it can be
harder with Ubuntu to temporarily install just that package from a newer
release than it usually is with testing to temporarily install something
from sid.

> And even then their goals are different and as such the expectancies to
> be created: Ubuntu, Fedora or Arch are *products* by themselves while
> Sid/Testing are *tools* aimed to produce a product, which is Stable.

Eh, sort of.  I think you'll find that many package maintainers, such as
myself, are pursuing both of those goals at the same time.

> Forget that and you'll fastly find yourself in nasty places (i.e.: start
> "selling" Sid as a "Fedora/Ubuntu, only better" and be ready to put on
> your asbestos suit because users will start to yell each time it breaks
> something -as it happens almost daily, and asking yourself "well, since
> we can't risk Sid to be heavily broken sometimes, where do we develop
> integration for Stable?").

*boggle*.  Something breaking almost daily is *completely* alien to my
experience even with running Debian unstable.

Russ Allbery (rra@debian.org)               <http://www.eyrie.org/~eagle/>

Reply to: