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

Debian Critical Mass

(Running the risk of sounding philosophical...)

I have been noticing some interesting things about Debian for a while
now. Note: This message is not a complaint. On all issues mentioned
below, I believe that Debian has been doing an excellent job.

Debian is Reaching Critical Mass

Debian is really starting to getter wider and wider recognition for
being the quality distribution that it is.

1. There are a significant number of people/projects interested in
releasing software as Debian packages. (i.e. Debian Freshmeat
repository, etc)

2. There are a number of distributions, nowadays, that are being based
on Debian.

3. Projects like Helix Gnome have put significant effort into
"supporting" Debian.

Debian is facing a number of obstacles which it much overcome to
maintain the level of quality which it currently has.

1. The size of the Debian developer collection/package collection has
expanded (and is expanding) to the point that it is increasingly
difficult to manage.

2. Debian's release schedule is far different then what we would like it
to be. The number of Debian packages currently present cause a "test til
we know it's stable" time period to last a very long period of time. I
hope this is not just shrugged off as Just Another Whine about Debian
releases. I think something really could be done about this.


My suggestion. I feel that it would be within Debian's best interest to
research some possiblities for the future of our project.

1. I feel that we need to provide a way for releases to occur more
frequently. This would serve our users by allowing them to have more
up-to-date selection of Debian stable packages. People should not have
to use unstable in order to have up-to-date software. And no, we should
not lower the quality of what we provide by releasing packages which
aren't ready. We all agree this would be bad and just downright stupid
(urm, Microsoft, cough, cough).

2. We REALLY DO need to start being a little friendlier to our would-be
Debian fans. Right now, we spend for too much time telling people we
don't want them to package stuff because we don't think they will do a
good enough job. Who cares if they do I good enough job? I sure HOPE
they do...but most of them will probably produce reasonable packages.
No, they won't be Debian perfect, but hey, that's life.

About these issues, I believe the solutions lie hand in hand. What about
this as an option...

The "Radical" Suggestion

I am going to suggest that the main/contrib/non-free way of doing things
is JUST NOT REASONABLE ANYMORE. What do I mean by this. Well, try this

What if we were to chop Main up into a series of smaller sections. These
sections (3-8 of them) would be independantly released. Yes, it would be
much harder to coordinate them so that they work well together. However,
I think this is worth a thought.

Maybe, as a possiblity, main could be broken up into categories based
upon rate of development. Applications that are updated very frequently
could be in a Rapid-Development section. This section could be released
every 1-2 months, with a specific timeframe. And if a package is not
ready for the next release (still buggy) instead of allowing it to hold
back other software, just release the most recent stable version again.

Or maybe there is another way of splitting up Main that would allow
sections to be released at a more reasonable rate.

Probably, the best solution is a combination of a couple different

What about these sections:

Debian-Core: the group of packages which make up Debian itself. Could
probably be release every 3-6 months, depending on the work going on at
the time.

Debian-RapidDevel: Rapid Development packages, like I mentioned above.
Released every 1-2 months.

Debian-Net: Group of Debian packages relating to anything network
related. This could include all networking applications, from Web to
Telnet to anything else network related. This could probably be released
every 3-6 months.

Debian-Static: Group of Debian packages containing software that is not
being actively upgraded, etc. These packages could be release on a
longer term basis - every 6-9 months, maybe.

Debian-Development: This section could contain all packages relating to
software development. This could include compilers, interpreters, ides,
etc. Released every 3-6 months.

Alongside this could (and I believe, should) fit a plan to allow other
groups to maintain collections of Debian packages. There could be a
Debian-Helixcode section which contained (yep) the Debian Helixcode
packages. These would not be officially "sanctioned" by Debian. However,
we would attempt to cooperate with these groups as a matter of providing
the best distribution for our users that is possible.

These collections would NOT start to include groups of packages which
would put us in any legal liablity or any such thing. This collections
would be started very infrequently and only after much consideration.
So, no, there would not be 15+ of these stinking little groups to deal
with. Maybe to start out with, 2-3. That could easily grow to 5-8 after
a couple of years.

Anyways, just my thoughts, for what they are worth. I believe that some
of these changes would truly improve what we provide to our users.

Lend me your thoughts.

Paul J Thompson

Reply to: