Being more selective with NNWP ? [Was: Debian Project News 2010/05 freeze delayed]
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Le 12/06/2010 08:53, Alexander Reichle-Schmehl a écrit :
> Hi!
Hi,
>> By the way, I noticed that lots of NNWP have been deleted, but there are
>> still gazillions of them left, could we try and find some criteria in
>> order to select a bit fewer of them: it's a PITA for translators, and
>> I'm not sure that so much packages will be "noteworthy" (understand that
>> if the list is too big, no one would actually read it).
>
> Yes, I removed most of them as you proposed and only left one Eucalyptus
> package, as this cloud things are quite important buzzwords I would like
> to mention.
Thanks a lot (I had a hard time to figure out between -cc, -nc, -sc and
-cloud which one would be the best one for eucalyptus-* stuff, thanks
for taking care of this choice).
> The current praxis is, that I remove:
> * Debian installer packages, as they are not run by the user directly
> * Library packages, as they are not run by the user directly (at least
> if I notice them)
> * -common, -data, -dbg, -doc packages, as they don't add new
> functionality for the user
> * The same with packages, which where split in several packages; then I
> try to only keep the "main" package
>
> Beside that, I don't know any good criterias. Was it only a PITA for
> this issue, or also for the previous ones?
Actually, the last issue was quite OK, but the first three were quite
annoying from our point of view (with my fellow French translators).
There are still 55 NNWP for this issue (there were "only" 35 on the 3rd
DPN), and Thomas who took care of there translation for this issue
thought it was still much.
If possible (I understand that this issue is a bit special with the
"open the gate" work of the ftp-team), I think it would be best to stay
bellow 40 (maybe 30?) packages: again, even if it is a not so easy work
for translators, I'm afraid that a too large list won't be read by anyone.
For the record, I think that the packages I proposed for removal fitted
your praxis, and I added some more selective criteria in order to lower
their number:
* specific development toolkit (except Debian ones);
* specific administration tools;
* specific hardware drivers or tools;
* plug-ins (or features) to already existing tools;
* software which inclusion was doubtful (a controversial thread on
debian-devel happened, e.g. clamz);
I even thought to propose for removal packages which description failed
to be conform with Developer's Reference's guidelines [1] or the Debian
Policy [2] (we could so easily go down to a list of 40 packages top ;-).
[1]http://www.debian.org/doc/developers-reference/best-pkging-practices.html#bpp-pkg-synopsis
[2]http://www.debian.org/doc/debian-policy/ch-binary.html#s-descriptions
> If it wasn't for the
> previous ones, maybe we can leave it that way (I'll try to remeber that
> python-* and similar is most likely a library, too), and hope that the
> ftp-team doesn't do a "week of constant NEW processing" again?
Sure, If the "normal" criteria allow us to go under 40 packages, I think
it would be good; if not, maybe we could try to be more selective.
Anyway, I'm glad we talked about it, with a better understanding of the
current praxis, I hope we can help to remove the not fitted ones, and
maybe agree on some more objective criteria if needed (we'll see on the
next issue how much package are left in the list after some normal
clean-up).
Cheers
David
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
iEYEARECAAYFAkwUDNsACgkQ18/WetbTC/qEMQCfZwFo4MG1myOh6MhByay+O4iv
60cAnRYH/DtYpRLRWL0nhsuCWECUgn65
=djip
-----END PGP SIGNATURE-----
Reply to: