RCBW - week #6

RC bugs squashed this week by yours truly:

This week I've cheated for at least two reasons:

  1. the first 4 entries in the list above are bugs of "mine", since within the Debian OCaml Maintainers team we migrated to a new package toolchain called dh_ocaml (with a lot of advantages I've anticipated before my summer break). The migration to that triggered (thanks, Lucas! :-) ) more than 30 FTBFSs which had to be dealt with. So, even if not cheating, it is at least breaking the NMU tradition of RCBW, but they were still RC bugs looking for a fix! All in all, I'm impressed by the fact that in about a week we dealt with most of them.

  2. the lxml fix above is severity "important" and as such, not RC. I attacked that because I intended to help out with the Python 2.6 transition which is still, at the time of writing, 41 bugs away from hitting unstable. I confess I did not find the task particularly fun (which is a requirement for me when I work on packages which are not mine) yet, we'll see next week ...

The feeling of the week I've to share is that I'm more and more impressed to not have received any complaint about my NMU activity, and I'm approaching the 50 NMUs. Actually I've received several "thank you" replies (you rock guys and you know who you are).

On one hand this is probably a good indication that I haven't (yet) done any serious mistake; I do love peer-review! On the other hand, and more importantly, this is dispelling inside me the folklore / cultural block we have in Debian that "NMU is bad", while NMUing should become a way to help each other routinely. Really.

ubuntu-like?
in not very uninformed opinon, I recall in Ubuntu, no package is 'owned' in the way they are in Debian. Ever upload is an NMU because their is no 'maintainer'. So if the Debian maintainers dont maintain their packages, as opposed to changing their status to "O" as in orphan, then its like what is already happening in Ubuntu--an NMU-fest.
Comment by Kevin Mark Sun 11 Oct 2009 08:36:26 AM CEST
not (only) ubuntu-like, but rather Debian "the old way"

Hi Kevin, actually, as Bdale Garbee frequently recalls in his insightful talks about the evolution of the Debian project, Debian used not to have the "Maintainer" field in the beginning. That field got introduced a bit later in project history (I don't remember the year exactly, check Bdale's slide at this year FOSDEM). The presence of that field has grown quite radically in Debian culture.

It has grown positively (people feeling proud about responsibility, peer recognition, etc.), but also negatively: people are ashamed of NMU-ing. To the latter feeling has contributed also the fact that people sometime complains very loudly and very aggressively about received NMU (sometime that is justified by mistakes done by the NMUer, but hardly the loudness of the complaint is). The most detrimental effect of this is that, when packages get de facto orphaned (maintainer has lost interest but has failed in publicly admitting that), there is a huge inertia in acting on the package, even for the most simple and surely harmful fixes.

That is to say that, yes, I completely agree that Debian has to benefit a lot from diminishing the sense of package ownership, but it is very difficult to get there, once a community has internalized in his culture the principle of package ownership.

Comment by zack Sun 11 Oct 2009 10:30:55 AM CEST
not very long ;)
according to Bdale's 2009 fosdem odp, here at http://www.gag.com/~bdale/talks/2009/fosdem/, one early release was in 09/1993. On 03/95, release 0.93r5, the "concept of explicit package maintainers" was added. The project had about 60 maintainers. So for the first 1.5 yrs, there were no package maintainers in the current sense. That is not a very long time in a 16 yrs project.
Comment by Kevin Mark Sun 11 Oct 2009 12:09:36 PM CEST
re: not very long

I'm not sure against which part of my reply you are arguing.

While I did not use the word "very long" my self, I surely feel that we have had the field for "very long" now, not that we haven't had it for "very long" :-)

Good to Ubuntu that they spotted this and did it differently, but note that for them is somewhat easier, given that the core part of the distro is managed in a "cathedral" way under the control of Canonical. It is fully understandable of course, but with pure-volunteers setting you need to find ways to reward people and the Maintainer seemed like a good idea back then.

Comment by zack Mon 12 Oct 2009 09:48:48 AM CEST