http://www.debian.org/News/2011/20110617
Απλώς μάζεψα λίγο τις συζητήσεις,αφαίρεσα ενδιάμεσα μηνύματα και έδωσα όσο μπορούσα μορφή ερώτησης-απάντησης προς τον Debian Project Leader,Stefano Zacchiroli.
(Η συζήτηση έγινε στο irc.debian.org #debian-meeting την Τετάρτη 22 Ιουνίου ώρα 19:00 UTC.)
Q: < eliad> Is Debian going to become something more than "just a GNU/* distribution" (if it is not already)? Defining certain standards/guidelines, for example. At the time Debian was established it was a revolution. but is Debian like that now? e.g. Debian could define some guidelines for development of new applications, preferred langs,.... This could potentially make bugs easier to find and repair. The number of packages are increasing and Debian simply would not be able to cope and stay "stable
A: <zack> thanks for the question eliad
think Debian aims at remaining a *distribution* in the proper sense of the word
that is, at remaining an intermediary between upstream authors and final users
trying to integrate everything properly together
but without changing too much, as long as it's possible
in that sense, I don't think it's up to something like Debian to establish standards to develop apps or the like but of course, the day a group of upstreams will do that, we (as in Debian) will do our best to support it properly in our distro (and in all our derivatives, fwiw)
Q:< Black_Prince> I heard something about Debian Rolling. Are those only rumors or that could happen?
A:<zack> gotcha, I saw that coming Black_Prince ;)
surely, it *could* happen
but as everything in Debian, it won't happen automagically, someone needs to work on it
there are some people which have made already a lot of work on that
first discussing the idea in a lot of depth (see the monster thread on -devel)
that discussion, imho, has been very useful, and has highlight the risks we should be aware of in doing something like "rolling"
then, other great people have proposed a way to go forward technically and are working on it to implement
so, I think it could happen, but --- as an old mantra in Debian --- it will happen faster if people will step in to help
Q: < raido>How would the DPL define the health of the dev community currently?
A:<zack> uhm, very interesting
I confess I don't have a good "metric" to propose, if that is what you are asking
but some indicators are for sure the turnover of developers
i.e. the amount of new developers joining and of old developers leaving
the latter might not be relevant in "young" projects, but it's something which is quite normal in a project like Debian, with its 18 years of history)
so, if I'll have to judge from that
I'd say that we are doing pretty good, the amount of new devs is quite steady
(sorry, I don't have the numbers at hand)
it's also interesting that quite a bit of the new devs are coming from derivatives
Q: < __Alex_> When do you thing Debian Hurd will have Linux 2.6 drivers?
A:<zack> ah, urgh, I confess I'm not much of an expert when it comes to hurd
love the fact that Debian has a hurd port being worked on
(and I've been told is sort of the best way to actually use gnu/hurd)
but I fear I can't help with technical questions about that
I'm sure people on debian-hurd can help with that, though!
Q:< liw> zack, do you think Debian's discussion culture has become more healthy in recent years? Have you noticed that trend reversing again? What could or should we do about it?
A:<zack>liw, once more, it's hard to have metrics for stuff like that
otoh, my gut feeling is that debian's discussion culture has got a lot better over the past years
unfortunately, my memory is tainted by big flames of several years ago, so I'm not sure I can appreciate evolution in smaller time periods
what could we do about it? take part in discussion with a positive
attitude
sometimes, I've the impression that people that can improve the quality of a discussion simply don't take part into it
leaving the ground to people who want to flame (which in turn risks giving the impression that everyone in debian just likes flaming)
Q:< devil> my question to zack is: what is the current state of debian-rolling and is there an agenda what needs to be done in the near future to make it become reality?
A:<zack> devil, my impression is that we need essentially two things: a team and some infrastructure
the infrastructure should be something like the existing software to migrate package from unstable to testing (aka "britney")
t could be another instance of that very same software or it could be something completely new
(people has started working on both possibilities, afaik)
the team is, well, a team of people with competencies similar to those of the existing release team
now, were to propose you're willing to help in that respect it's a related question
there is the "cut-team" discussion list on alioth for that, but most recent rolling discussions have actually happened on -devel directly
Q:< liw> zack, have you considered training people who might want to run for DPL next year? what kinds of skills or talents would be most useful for a DPL?
A:<zack> liw: yes, absolutely, I've thought about that and it's a great idea! (not by me, it's been advanced during this year dpl campaigning, iirc)
as step 0, I've asked for volunteers to do some dpl-ish tasks in recent "bits from dpl" mails
if there are volunteers for that, I'll be glad to "outsource" some tasks to give an idea of what a dpl routinely do
as step 1, I've submitted a "dpl wannabe" event at the forthcoming debconf11 conference
ideally, at that event both past DPLs and wannabe DPLs should take part, and share experiences
as an ideal step, I'd also like to write some "tips & tricks" as stable documentation
but I've never found the time to actually do that
about skills, dunno, I believe anyone should propose her own way of doing the job
but for sure, communication and "people" skills never hurt
Q:< Black_Prince> What do you think Debian Rolling could do to Debian's popularity? Do you think people would more use it when they get newer software more often in Debian?
A:<zack> Black_Prince: so, I'm sure we have several different views on that
my own take is that something like "rolling" can be very interesting for two potential kind of users:
1) "advanced" desktop users, which want recent software and have skills to cope with upgrade annoyances (in the _upstream_ software, not related to packaging)
2) developers, who need up to date libraries, but still can't risk using unstable and having their development environment destroyed by an RC buggy upgrade
I don't think that those kinds of users are against using "stable"
they are just different targets
Q:< valhalla>what are the areas of debian that you think are more in need of some "love"? (more people, more work, etc.)
A:<zack> valhalla: oh, ouch
if I'll name only one, everyone else will jump at me :-)
<Ganneff> ftpteam can always take people, especially coding ones!
<zack> in short, many teams are in need of more peoplepower
you probably won't find any single team which is *not* in need of more people
so, one way to spot the one which are most in need of "love"
just go and find "single point of failures"
it's not nice being in that position
not for the team/person, not for the project
when you find one, step in and volunteer to help
in general, even the most delicate tasks, could use at least some kind of help
Q: < algernon> question: Going a bit back at the developer flow: do you perhaps have a vague idea how many old developers come back after spending some time off Debian?
A:<zack> algernon: ah, if you want a number, I'm sorry but I don't have it either (although maybe zobel-little-helper will kick in)
<zack> but I've in my memory various people who has recently returned
<zack> (or maybe DAM, actually)
<enrico> gut feeling, one every 6 months or so, I'd say
<Ganneff> hard to say a definite number
<zobel> some 3-5 per year
<zack> it's something I didn't remember in the past
<algernon> zack: something between "a few" and "more than my fingers" is about what I'm looking for :)
<Ganneff> nut all properly quit / come back, some just go mia for a time
<zack> if you look back a few years, i'd bet on ~ 10 then
<zack> algernon: it's good, imho, and once more it's something that I don't feel any other project out there has faced (yet)
Q:< RoboTux>Stats about emails the DPL receive (how many? what are the main topic?)
A:<zack> oh, I've the numbers for that, I've computer them a while ago for an interview :)
just a sec
stats taken in february 2011, from the beginning of my (first) term up to then
average of 15 mails/day (9 received/day, 6 sent/day)
the topics vary quite a bit
from "urgencies" of some kind, to conflict solving, ... to crazy people asking nonsense :)
the most, err, "interesting" I've ever got it's just after the start of my first term
someone proposed that debian should drop the linux kernel and ship only a mach kernel
note the *drop*, not just *add* a new port
Q:< Black_Prince> What is average age of Debian Developers/Maintainers? Are there any Debian Developers or Maintainers that are 18 or younger?
<zack> Black_Prince: eh, I'm interested into that myself, but I don't think we have all the data to compute it (specifying your age is optional, iirc)
<Ganneff> we dont have much data on that - and we have VERY few very young people
<zack> my feeling is that we have an average age around 35, but with an important variance
we've had cases of very young DDs, but not that many
<Tolimar> zack: Four years ago I did a talk about Debian at my university. Back then we had one DD who said he was 16 if I remeber correctly.
aven't checked since then.
<zack> Tolimar: ah, thanks
Q:< lindi-_>speaking of stable, what version of gwibber are you using with identi.ca currently? Afaik squeeze's version does not support twitter (since it lacks some OAUTH thingy)
A:<zack> lindi-_: my laptop runs a "pure" unstable, so I'm using the gwibber which is there
I'm aware of the problem you mention
it's something that _might_ deserve a stable proposed upload, but only if it's a self contained change (which I doubt so)
(and of course the final decision is of SRM)
Q:< nadir> i don't fully understand the relation between the freedombox and debian. to me it looks like it was a pure Debian project, right? If so: Can one be confident about its future?
A:<zack> nadir: so, some technical activities around the freedombox started by a group of debian folks at debconf10, after a talk by eben moglen there
that's why the hosting of sw, discussion list, etc it's on alioth
then, the freedomboxfoundation has born to support the freedombox "idea"
and that is a separate entity from debian
nevertheless, we're in very amicable terms with them
and they respected, in good foss tradition, the already existing hosting infrastructure
no matter what, the ties among debian and the freedombox project are important
so, I can't tell if you should be confident about its future
but for sure the initial debian activites on the project has been very much respected by the fundation
Q:< raido> It is no secret that Debian derrivitaves enjoy a popularity from time to time (recent history especially) thath eclipses Debian. Do you find a level of discouragement from the community/devs that requires some encouragement or do these things work themselves out? Is it something you actively deal with?
A:<zack> raido: at a social level, sure, we've felt that in the past in the debian community
it's just human, and applies a lot to volunteering communities
still, I've the impression that we're now realizing that we should be actually _proud_ of that
a lot of distros out there exist thanks to us
still, we need to make clear what we want from others if they want to "play well" with us
and with foss in general
(giving credits + giving back, essentially)
which is very much like what debian has promised to do with our upstreams in the social contract
Q:< Eulenspiegel69> question to zack: Linux is known to be significantly fragmented; software written for one distribution often needs some modification in order to run on another. What can we as a Community do to ally with developers of other notable distributions in order to create common interfaces, etc.?
A:<zack> Eulenspiegel69: I fear we're back at the "platform" problem (see first question)
I think that if a distribution, like Debian, starts doing that
then it will appear as something that want to be "different" than the sum of its parts
that level of standardization, imho, should happen upstream
upstream authors need to spot the need of standardization
and establish the needed bodies
e.g. freedesktop
I don't think a distro is a place for doing that, no matter it's market share
Q:<eof> question: zack, how do you get enough time to read and respond to all letters sent to dpl@d.o, participate in discussions on the mailing list, to speak at various conferences with talks, ... and sure to do other things in your life (outside the Debian Project)? ;-)
A:<zack> eof: actually, you know I don't do *that* well, given I've an answer in the pipe to you since more than week :-P
seriously, I've no idea, but I'm sure I'm not the most occupied person in the world, there are people with way much more than me who still manage ;)
Q:< zobel> question: how much money has been spent for real life meetings in the last year (roughly)
A: <zack> zobel: roughly, we've spent on average 1000 EUR / sprint
and we've had about 10 sprints
(an easy way to actually have precise numbers, it's something the auditors are working on)
Q:< gnugr>We Greek debian community have a problem with some school packages from our Education administration wich they need a software or package named "prism" to run. They run on ubuntu as well, but we need it as Debian users too.
A:<daemonkeeper> gnugr: We were discussing this with you in #d-mentors some time ago. Note Ubuntu does not include your package either and may come from a third party repo you(r school) might be using.
<zack> daemonkeeper: thanks a lot :)
<daemonkeeper> welcome :)
<zack> so, if it's a 3rd party repo, chances are it'll work on both distros
<zack> (or at least that it can be made to work on both distros)
<zack> still, if you need it to work with debian, you should contact the maintainer of the repo
<zack> and ask them to support debian explicitly
Q:<babilen> question: If we look at software development and Debian's relations to upstreams like, say, the Ruby community it seems that a lot of developers do *not* use Debian packages, but rather third party packaging tools such as gem, pip (Python), leiningen (Clojure) etc. What direction do you think Debian/packaging needs to take to cater to the needs of developers?
<zack> babilen: very interesting (and tricky) question
so
I'm no ruby packager myself, but in several languages there is a problem
namely, the habit of embedding (specific versions of) libraries
instead of using them as external dependencies
that's very typical among java developers for instance
but I've the impression that gems/eggs/etc. are all abstractions that make that easy
<babilen> indeed
<zack> in Debian, we've a policy of de-duplication of code
because it's the only sane way to deliver, for instance, security upgrades
no one can guarantee security support for $x * 10'000 packages if code embedding is the rule
that considered, I'm not surprised we have grudges with communities where that practice is commonplace
I think we're technical right on that
and we should insist in convincing upstreams of our reasons
Q:< liw> QUESTION: zack, what's your foremost suggestion for making Debian development more fun for everyone?
A:<zack> liw: be *bold* !
spot something that needs improvement
<liw> not _italic_?
<zack> take responsibility for improving that
and do that
seeing a lot of people doing that will induce a lot of excitement everywhere in the project, I believe
Q:<MadameZou> should be a DPL term longer than one year ?
A:<zack> MadameZou: I think it should, yes
<Ganneff>should be a DPL term longer than one year ?
<Eulenspiegel69> for example 4 years, like the American president :)
<zack> 2 years is probably a good length, but it should be balanced with guarantees for the community that they can easily recall the DPL if goes MIA or the like
<zobel> zack: so we should have a GR on that?
<zack> we've had a nice discussion about that in the last campaign period, IIRC
<jhr> yes, let's make zack retroacively been voted for four years :)
<zack> zobel: that would be the only way, given the term length is written in stone^Wconstitution
<zobel> zack: so we should make a GR of that
Q:<MadameZou> what is the next myth you will destroy? (after "NMUs trigger aggressive responses" and "-devel is a constant flamefest")
<MadameZou> quite provocative, but interesting ;)
A:<zack> ah ah
fwiw, I don't think I've destroyed neither of them
<vorlon> -devel triggers aggressive NMUs
<zack> I've worked a bit on the former, and just _said_ the second (although I hope it's true)
a good target would be: "there is (no) cabal"
Q:<MadameZou> what about "local" debian groups? I know it was part of your platform: which is the state of the art on it?
A:<zack> it's something I believe it's very important and we've in fact all of the ingredients for it
we just need to streamline a bit the procedure and advertise the possibility
a little helper has stepped in to help me out on that (hello eof)
and we've basically concluded a review of the state of the art of what other projects are doing
if all goes well, you might expect some news about that in a month or so
Q: < Laney> QUESTION: There's a movement amongst a small number of (web browser) upstreams towards rapid release cycles with corresponding reduced security support for older releases. Do you think this can be made to fit in with Debian's release model? Do we have any options other than swallowing it?
A:<zack> Laney: I'm quite scared by that, yes
<zack> in particular, I'm scared by the fact they seem to ignore that also on desktops, there are users who need a long time frame of (security) support
I know of several realities where they have desktop that they want to be fully functional and minimally changers for 5+ years
clearly, fast browser releases do not fit well with debian stable processes
they might fit well with rolling, but that's not an answer
I think we should do some pressure to support, side by side with fast release, some sort of long term support (upstream) releases
and aim at including them in our stable releases
Q:<MadameZou> I have only one question for you zack and it is mine
<MadameZou> don't you miss the hacking work on debian?
A:<zack> MadameZou: totally :), but it's been fun to change as well
Q: <MadameZou> do you want to run for a third term?
A:<zack> ah ah, that's too early, i really have no idea
(I know you won't believe me, but that's just too bad :-P)
<zack> if I may, I've one concluding comment
we badly need irc meetings, both like this one and within teams
this kind of meetings, especially if they are periodic, are a way to drop communication barriers as well as a way to be more transparent about what happens in the project
so, please everybody, peruse this channel, peruse it's wiki page
and go propose to your team to have periodic meetings
either there or on the team channel
thanks everybody
and a big thank to MadameZou for hosting and organization!