Re: What to do with draft packages?
On Wed, Aug 04, 2010 at 12:31:44PM +0200, Steffen Möller wrote:
> Andreas once took the effort to move my collection of draft
> packages from a past home page of mine over to to Debian Med,
> and funnily enough this has indeed helped me 5 years later :)
:-)
> > In principle we could drop such a file
> >
> > echo "Did not even find time to write a DRAFT file" > DRAFT
> >
> Could do, but I think we have "UNRELEASED" in the
> changelog for this and the README.Debian that could
> summarise the packaging status accordingly.
I don't think so. Perhaps I did not made my idea clear enough. While
"UNRELEASED" in the changelog is just a binary information (released or
not) I would like to know the reasons *why* the package is not (yet??)
released and whether the creator of the changelog ever intended to
release this package. So in principle we could also use the
debian/changelog file as the source of information but if you want
to seek for such drafted packages it would be easier to fire up find
for a certain file name instead of parsing changelogs. The draft
file could contain things like:
- started packaging but found a better alternative which is
now packaged and the package name is ...
- this version of software is not ready to be released but
we should look again after time X
- Upstream does not support my attempt to package and I gave
up
- There are so many dependencies to package first that I've
lost interest
- Just no time to finish this
- ...
Such dedicated information might be helpful for other people
who might be interested into cntinuing the work.
Hmmm, thinking even more structured: We might use
debian/upstream-metadata.yaml
for this as well. I really start to like this concept, because on one
hand it solves the "seek for draft packages" issue because the
information is there in a structured (parseable) way and we even will
find means to propagate this information to our web sentinel (once the
code is written, but in principle it is trivial). The information is
not necessarily information about upstream and so it might be some
"misuse" of this file, but about 50% of the reasons might be connected.
So we could use a field named "UnreleasedReason" (or something like
this) and handle this information appropriately.
> For some other bits I once asked for the pkg-escience
> project, just knowing that none of the packages would
> be of a quality sufficient for an upload in any reasonable
> time. Biojava has since moved from there. Other bits
> still remain. I personally am extremely happy to have
> this rogue repository as a fits it all backup, Andreas is
> less so.
Yep, we discussed pkg-escience. IMHO it just creates confusion to
others, but that's another topipc.
> Please upload your drafts, I am certain someone will benefit.
Yes ... and declare it in a proper way.
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: