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

Re: Skipping version numbers in adopted packages

On Wed, 22 Aug 2007 13:11:52 -0500
Luis Rodrigo Gallardo Cruz <rodrigo@nul-unu.com> wrote:

> So far, all discussion I've seen about whether to collapse changes
> made during sponsorship review into a single debian revision for
> upload have focused in the case of an initial package upload. 

Really? My perception was that upgrades and updates are also covered in
the same way as the ITP. It makes little sense to change the method
after initial upload unless the maintainer is also changing sponsor.

> Does anyone have any special arguments for doing it one way or another
> in the case of an upgrade?

? My preference is mainly regarding upgrades.


I have no problems with "missing numbers" when the update is 1.2.3-4
etc. Others disagree. I don't want to use ~foo (be that ~rfs or any
other suffix) because I find that ugly. Others want to use tilde
suffixes of various kinds. None of these preferences are "invalid" or
necessarily better than any other, it's just preference. Unfortunately,
that means that changing sponsor can mean changing the method but that
is inherent in the current system.

So 0.0.1-1 gets an RFS but it is 0.0.1-4 that I upload to close the ITP
with appropriate -v and -sa options to debuild and pdebuild.

0.0.2 is released, 0.0.2-1 has a minor problem, so 0.0.2-2 gets uploaded.

The PTS records 0.0.1-4 uploaded to unstable and the changelog lists
0.0.1-1, -2, -3 and -4. The next PTS record is 0.0.2-2 and clicking on
the changelog link shows 0.0.2-1 and 0.0.2-2 (on top of the previous
changelog entries).

By 0.0.3 or 0.0.4, the maintainer is in a better position to prepare
updated packages and -1 is likely to be good enough to upload. It is at
this point that I'd be considering activating the proposals from the
GR, once the infrastructure for that is in place.


Neil Williams

Attachment: pgpbbYCH1iZER.pgp
Description: PGP signature

Reply to: