On 2/8/11 13:04, Richard S. Hall wrote:
On 2/8/11 12:29, Peter Kriens wrote:
New version number for any reason makes a lot of sense. A bsn +
version must be as unique as a SHA-1 hashcode of the bundle.
I think everyone agrees that a new version number should always be
used for every release. The issue here is about a release that was
never released because it was canceled for some reason during the
voting period.
After re-reading your message, I read "reason" as "release"...
However, the tricky part is as I described...at Apache there is no
release without a proper vote, so the question is, do we need the
version number to reflect these failed attempts?
Give the ongoing vote on this, it doesn't look like we have any clear
consensus. As a result, I'm happy to let the person doing the release
decide...
-> richard
-> richard
Kind regards,
Peter Kriens
On 4 feb 2011, at 09:39, Carsten Ziegeler wrote:
Guillaume Nodet wrote
On Fri, Feb 4, 2011 at 02:52, Richard S.
Hall<he...@ungoverned.org> wrote:
If enough people respond maybe we can reach some sort of
consensus...or else
we could call a vote on it.
Can other felix members speak here ?
I guess we don't find consensus by just discussing :) Some of us prefer
it this way, others the other way. I prefer increasing the version
number for each retry, it requires no additional work (except changing
the version in jira) and makes it easier to track if people are using
failed releases. Sure, comparing the hash of the binary artifact would
solve this as well, but just looking at the version number is soo
easy :)
If - as a project - we agree to use the same version number for retries
I'm ok with as well.
But I agree we should do this consistently and just write it down
somewhere. So let's call a vote and the majority wins :)
Carsten
--
Carsten Ziegeler
cziege...@apache.org
