Re: Some proposals
On Thu, 27 Feb 2003, Jaldhar H. Vyas wrote:
> On Thu, 27 Feb 2003, Fabio Massimo Di Nitto wrote:
>
> > I think this will just lead to a bloated archive. It will be much easier
> > to add a pkg that containts informations from apt-get.org
>
> Ah but much easier for who? Think like a user not a developer.
As user i would not find to difficult to tell debconf: yes please add
external archives to my sources.list to get certain extra features
And as a developer (not official yet) i would rather prefer to maintain a
relativly simple pkg to keep sources.list updated.
>
> > or able to build
> > an unofficial CD out of it.
> >
>
> unofficial or official doesn't matter too much as long as it is available.
I agree with you that it is nice to have but i would rather keep it
unofficial. What i personaly see in having it official is an increase in
the work developers will have to do to handle different behaviours of the
same piece of software compiled for woody and for sid (always considering
the worst situations of course) due to libraries, compiler, whatever...
>
> > I believe this will only create a lot of confusion. From the most simple
> > package that get 2 different fixes from 2 different developers that did
> > not coordinate the job to whatever can be. Things will really break badly
> > if for any reason one of the thousand DD will not coordinate correctly
> > with all the others.
> >
>
> Any project with more than 1 developer can face such problems. There are
> ways to deal with it. And any developer who doesn't play nicely with
> others will swiftly find himself getting kicked out.
Yes this is in the final resolution and we agree. to make my point more
clear what i meant is on a daily base. If it happens also for a mistake or
an email lost on the way there might be a "broken" sid for that day and as
far as i understand from your words sid should be the only distro.
Something might not work in this way. do you see my point?
> Sometimes things that need to be done don't get done quickly not because
> they are hard but because of a lack of time. Can we leverage our ~1000
> developers to improve this?
With might end up in a real philosifical discussion here but aren't DD
all voluntiers that dedicate some of their time to Debian? if for some
reason i end up without time i should definitely orphan my job/s. Well I
might be too "fresh" from reading the policies again or i am too close to
these rules but that how it should work. As weel there are policies to
handle with MIA DD :-)
> > probably this might work for a core team, because composed of relativly
> > few developers. What about teams that have to deal with tons of pkgs?
> > communication might become an issue and personnaly i would not like to see
> > my pkg NMU'ed just because i wasn't able to be in front of my computer for
> > half day. Of course this is an "exagerated" situation but it is a
> > situation that can happen. How would you handle that?
> >
>
> People would have to use common sense. If it is just a dependency error
> or due to a typo (like a bug that was reported on webmin-postgresql
> yesterday) why would you mind if someone did an NMU?
I wouldn't mine to let people giving me the minimum time to deal with "my"
bugs. I might have it fixed and not being able to upload in 20 minutes,
and find an NMU done in 19 minutes will not please me too much because I
will have to redo somework to upload a clean pkg that will close the NMU
in a clean way.
> > > Other companies/interested
> > > groups of people would be responsible for putting out finished products on
> > > whatever schedule suited them.
> >
> > Based on what?
>
> Whatever they are interested in.
>
> > sid gets updates on a daily base, saturday and sunday
> > included. Do you expect companies to track the archive of 12000 pkgs every
> > day and be able to take out a stable snapshot?
>
> Yes.
>
Basically this will never happen. Not to be negative but you are asking
even a small company made of 100 people (that has an average of 3/4
supporters) to build their own distro and track the same amount of work
done by ~1000 ???
> We already have forks of Debian: xandros, libranet, any person who ever
> recompiled the KDE packages to provide MP3 support, etc.
>
> But I'm thinking of forks along another axis. Some people never install
> Debian scientific packages, some people never install textmode games etc.
This remember me a discussion we had long time ago about reorganizing the
archive with better usage of priorities and sections.
>
> > I personally think that it will only
> > produce more and more confusion between end users.
> >
>
> There already is. Look at the post that started the "Debian is doomed"
> thread.
I didn't read that thread but i will look at it.
> > This is partialy true for "normal users". I would never do that on a
> > server installation. I love to be sure that my "real woody" will not
> > break because done in a hurry due to a dead line forced by someone that
> > probably have no idea of quality means.
> >
>
> So you and other people who have the same needs would get together and
> make a server version. Right now a release critical bug in frozenbubble
> can halt the release of your server OS. Does that make sense?
My server requirements might be different from others. This solution
sounds to me like having Debian-webserver Debian-sqlserver and so on...
and not integrated together when there might be a requirement for that.
Regards
Fabio
--
drac (1.11-7) unstable; urgency=low
* added IPv6 patch from the great IPv6 Team
-- Noel Koethe <noel@debian.org> Sun, 9 Feb 2003 19:33:00 +0100
Reply to: