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

Report from Debconf's QA BOF


Here is a list of what was discussed/agreed upon during Debconf's QA

Handling of orphaned packages
After lenny release, the handling of O, ITA and RFA bugs will be
- O:, ITA: and RFA: are reassigned to the package they are about.
- They are marked as affecting wnpp (new BTS feature) so they show up in
  the wnpp page
- Severities: O: serious
              RFA: not serious
	      ITA: keeps the severity of the previous state
- O: bugs are now RC, so packages are removed from testing using the
  release team's existing policies. Which means that some O: packages
  might stay in testing for a longer time because they are dependencies
  of other packages.

The idea of having a script inside devscripts that lists the RC-buggy
and orphaned packages that are locally installed was proposed. I filed a
bug about that. See #495152

Removing packages for unstable for QA reasons
- orphaned packages
- useless packages
- packages in bad shape

One must be *very* careful about this: diversity and universality are
amongst the fundamental values of Debian, and we should not turn into
"removal nazis". Many packages with very low popcon are very useful, for
example. Many packages which haven't been uploaded for a long time are
still used. There's not a single criteria that will allow to find all
the suspects. One need to combine different criterias.
BAPASE <http://qa.debian.org/~lucas/bapase/> should be used and extended
to get a list of packages to investigate (not remove them automatically,
of course!)
Ideas of improvements for BAPASE:
- MIA status of the maintainer (time of last activity)
- When a package was last in sync with upstream

Ganneff announced that snapshot.d.o will become an official service
(ETA: some months). This will allow to easily get back packages that
were removed.

UDD improvements
The idea of having historical data about the list of files of each
package was proposed. I'm not sure it's a good idea, because of the
amount of data it would represent. It sounds better to use snapshot.d.o
for that.

That's basically all I have, but it's possible that others have taken
more notes. Don't hesitate to post them.
| Lucas Nussbaum
| lucas@lucas-nussbaum.net   http://www.lucas-nussbaum.net/ |
| jabber: lucas@nussbaum.fr             GPG: 1024D/023B3F4F |

Reply to: