[Discuss] Notes on kubuntu dist-upgrade to dapper from breezy
Daniel M. German
dmgerman at uvic.ca
Fri Jun 30 08:31:37 PDT 2006
Alan> In sum, the substantial number of dist-upgrade issues I reported and the
Alan> further two initial use issues reported above leave a quite disappointing
Alan> impression of Dapper compared to Breezy and especially for Dapper compared
Alan> to Debian stable. Perhaps Ubuntu does not have the resources to provide
Alan> quality assurance for their short release cycle, and actually the
Alan> historically long release cycle for Debian stable (which everyone including
Alan> me bemoans) is really required to achieve its high overall quality. I have
Alan> also seen it argued that Dapper is in worse shape than Breezy because the
Alan> Debian unstable snapshot that Dapper started with was in much worse shape
Alan> than the snapshot used to start Breezy.
I am going to wear my "FOSS researcher" hat first.
based on my observations (and a grad student who is looking into
code-reviews in FOSS) I think it is a difference in the process used
by each.
Some FOSS projects force "pre-commit" reviews, thus they make sure
everything works before it is committed/integrated. Others do
"post-commit" reviews: the include changes, and expect the users and
contributors to find the bugs (what my student calls "lazy review").
In a way it is a policy issue. Ubunto seems to be less worried with
bullet-proofness in favor of speed of change, while Debian seems to be
the other way around. And, as Alan says, it might also be a resources
issue. We have observed the two types of review in many pieces of
software, but mission critical ones (such as Apache and openbsd) tend
to do pre-commit reviews.
What this means, in my opinion, is that ubuntu users have to be more
worried/concerned/attentive about incremental updates of their
distribution, while Debian users can expect a more solid system after
installation.
In a way it is not that different from the emphasis in openBSD to
review every line of code, compared to Linux where code is expected to
be tested and reviewed by the submitter. One is more secure, the other
more dynamic. The choice is of the user. That is the advantage (one
can say) of FOSS.
Alan> Finally, despite the griping about the quality of Dapper, I do intend to
Alan> stick with Ubuntu for a while. The reason is that Debian stable is seriously
Alan> out of date for a Linux desktop because of that long release cycle, and the
Alan> Ubuntu security team seems to be much more active (especially with kernel
Alan> problems) than the Debian security team(s). Thus, I have decided to support
Alan> Ubuntu with some bug reporting activity which I hope they will use as part
Alan> of an effort to substantially improve their product.
(I am going to wear my "FOSS maintainer" hat now)
Which is what every project expects: that its community help in any
way they can. Every help (for small that it might seem) helps to
create better FOSS. If you really want better software then let the
maintainers know how it can be improved (even say thank you once in a
while, to make them feel their work is appreciated and used). If you
find bugs, report them. If there are things it can do better, say
so. If there are features you might want to see, say so. Users have
knowledge that is important too and that sometimes the contributors
lack.
--
Daniel M. German "If builders built buildings
the way computer programmers
write programs, the first woodpecker
that came along
would have destroyed
Weinberg's law -> all civilization. "
http://turingmachine.org/
http://silvernegative.com/
dmg (at) uvic (dot) ca
replace (at) with @ and (dot) with .
More information about the Discuss
mailing list