Gunnar Wolf escreveu isso aí: > [ I am adding a Cc: to the debian-perl list, as the PET wizards live > there and might have a say on this. Also, because the pkg-perl group > has some packages which could be moved into a packages-wip > directory, cleaning up our main stats? ] > > This thread starts at [1]. > > Antonio Terceiro dijo [Sat, Sep 13, 2008 at 01:15:50PM -0300]: > > Hi all, > > > > I was just wondering: since our QA report tool [1] supports listing > > packages not in the repository yet as "work in progress" [2], why don't we > > remove the packages-wip/ directory from our repository? > > > > This would have IMO the following advantages: > > > > * the WIP packages would appear in the QA report and we would be > > more aware of their status > > * less clutter in our repository > > I agree with your motivation, it's good for them to be more > visible. However, PET's logic is a bit different - And I think we > could import some changes in our logic from the pkg-perl group :)
Yes, personally I liked very much pkg-perl's changelog-based workflow. :) > pkg-ruby extras uses packages-wip for things that have started moving, > but are still not good enough for upload. Those are... well, somehow > out of PET's scope, IMHO (although telling us about new versions could > be good). > > What pkg-ruby-extras regards as 'work in progress' is packages that, > in their version in SVN, have an UNRELEASED distribution. You mean pkg-perl here, right? > It is customary in the pkg-perl group to always tag with > "svn-buildpackage --svn-only-tag --svn-noautodch". This means, we > never have changelog entries such as (taken randomly): > > liblog4r-ruby (1.0.5-8) UNRELEASED; urgency=low > > * NOT RELEASED YET > > IMHO, this entry says nothing, and probably will only be bothering us > later. If the next action we take on liblog4r-ruby is to import a new > upstream version, we will have to manually remove the entry for > 1.0.5-8 (which will never exist). What is done in pkg-perl is to > reserve the UNRELEASED tag for changes made in the package which are > not ready to be uploaded - Please refer to the changelogs in the > repository (the smaller number in parenthesis) at the "Work in > progress" section of the pkg-perl page [2]. I completely agree that such changelog entries are useless and should be reserved for packages in which there is actually any work going on. > We almost always have there an invalid changelog entry, starting with > a TODO: detailing what is missing in that package. > > I am willing to organize that, what do you think? > > I do think packages-wip is a good thing to have. Maybe it would be > better to "export" the idea to other groups, and add support for it in > PET, just as an extra section? What do you think of Martín's solution of listing never-uploaded packages in a different section? IMHO it solves your concerns of differentiating what we meant with packages-wip from what PET meant by "WIP". (this way we could move packages from packages-wip to packages and still have them differentiated from packages that were already uploaded and do affect Debian's QA). -- Antonio Terceiro <[EMAIL PROTECTED]> http://people.softwarelivre.org/~terceiro/ GnuPG ID: 0F9CB28F _______________________________________________ Pkg-ruby-extras-maintainers mailing list [email protected] http://lists.alioth.debian.org/mailman/listinfo/pkg-ruby-extras-maintainers
