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

Deficiencies in Debian



This mail is some major rant.  I've already ranted in private and was
asked several times to move this to debian-devel.  I've added some
more problems I have detected.  Most of the following was discussed
either with a bunch of developers and keen people at the Linux
Kongress.

We currently lack:

 . Release plans

   There are only fuzzy plans on what will be released as next Debian
   release.  There are some people who are trying to give slink an
   update while the majority believes that potat will be 2.2.

   It is still not clear what will be 2.2 and what will happen if
   there should be an updated slink called 2.2.  Our project leader is
   affected, though.

   Some ideas against an updated slink in favour of potato being 2.2:

    - Confusion as something different than stable, frozen, unstable
      will be relelased as stable

    - No defined procedure about what will go in and what not

    - If an updated slink would be released as 2.2 we must not release
      potato within the next four months since a stable release needs
      to live for a while (also to justify the name, stable).

    - An updated slink will automatically postpone potato and decrease
      the importance to work on problems with potato.

    - General change in the way how debian works, without letting the
      developers vote about it.

    - Ignore potato in favour of something else will keep potato
      unreleaseable.

 . Release goals

   Officially "we" have decided that we don't want to have release
   goals anymore.  Now there were rumors that all packages will have
   to be converted to be FHS complient for potato.  There still was no
   public announcement.

   As a maintainer for several packages, I'm still confused what to do.

 . Policy changes

   It is unclear how policy changes affect packages.  The changelog
   within debian-policy*.deb is (imho) not detailed enough to tell Joe
   Clueless Maintainer what he needs to change in his package other
   than s/2\.4\.0\.0/3.0.0.0/ in debian/control to create a package
   complient with the current policy.

   It's even unclear which policy packages have to support for potato.

   It's also unclear what will happen to packages that are not policy
   complient.

 . Boot-Floppies

   It's not clear if/who is working on boot-floppies and into which
   direction we're going.  Obviously there is nobody working on it
   speaking of a general direction, and nobody's about fixing all
   those darn bugs.  Even worse the current boot-floppies are not even
   compilable or will work on potato if they would be.

 . Release cycles

   "We" decided that there will be two releases within a year.  (One in
   spring and one in autumn.)  Well, speaking of potato this goal
   can't be met anymore.  Although it's been said that there will be a
   freeze on November 1st several developers believe that this date is
   rediculous and impossible to be met.

   At the current state we won't even have proper boot-floppies.  I
   don't have to say how much that sucks.

 . Stable subreleases

   It seems that only very few people care about our stable release.
   Out of the security team only one person spent brain on it, the
   ftpmasters only worked on unstable, the new Stable Release Manager
   worked on something between stable and unstable by forgetting about
   the stable release.  Even though there were common opinions that
   only security updates are going into stable, there were tons of
   packages uploaded for it.

 . Configuration Management

   We have ben acknowledged that we need to reduce pre/postinst
   interactions and some proposals have been made that are known as
   "Configuration Management".  pre/postinst questions will interact
   with a database that is able to contain preconfiguration so cluster
   installations are easier.

   We have to face the truth, doing a cluster installation, debian is
   one of the most difficult distributions.  Maintaining a cluster,
   though, it would benefit of Debian.

 . Buggy packages

   Even though there are quite a lot registrated developers (~550 by
   counting logins) there are over 4000 packages with outstanding
   bugs.  Even though many packages don't seem to be maintained
   anymore it is still quite difficult to take over them or to rip
   them off of the distribution.

   There are also only very few (four) people working on general
   bugfixing.  It's still the case that (too) many people are working
   on their tiny little packages but miss general tasks while even
   "old" people are not managing the distribution - as one would have
   expected.

 . New maintainers

   There are a lot of people wanting to become maintainers of some new
   (often little) packages, often also without a clue.  Packages are
   buggy, partially not well maintained, also for packages taken
   over. Just adding them to the list of people who are allowed to
   upload into the main archive will just increase the distribution in
   size, not in quality.  [No new-maintainer bashing please!]

I have to acknowledge that Debian has reached the point where it has
grown too much and cannot continue as before.  At the moment we
already have chaos all over with no proper leadership.  The next
release is months away, boot-floppies are not working, several goals
are only slowly getting passed, still it is the bazaar of little
cathedrals.

Most developers are only working on their tiny five packages or are
even entirely inactive nowadays.  Only very few people are taking care
of general management tasks.  Remember this is an association of >500
people.  There is still no proper management.  Guess what would have
happened if it were a company...

I believe that this "strategy" will lead Debian into death if we
continue as before.  Therefore things have to change.  Since I'm not a
manager either I can't come up proper ideas all the time, and since my
time is limited I cannot force people to do the right thing.
Processing mail already takes much more time than working on software.

Regards,

	Joey

-- 
The MS-DOS filesystem is nice for removable media.  -- H. Peter Anvin

Please always Cc to me when replying to me on the lists.

Attachment: pgplK8DUCrk8n.pgp
Description: PGP signature


Reply to: