maven-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stephen Connolly <>
Subject Re: [VOTE] Apache 3.1.0-alpha-1
Date Wed, 29 May 2013 08:10:52 GMT
On 29 May 2013 06:49, jieryn <> wrote:

> Greetings,
> On Wed, May 29, 2013 at 1:36 AM, Hervé BOUTEMY <>
> wrote:
> > I'd like to work on Arnaud's idea of error message enhancement in case a
> > plugin fails because of unavailable Sonatype Aether: if you can let me
> 12 more
> > hours from now, I'll do it tonight
> Version numbers are cheap. Can't we just make an alpha-2?
> I'm just a user, but I'm getting pretty sick of staged alpha releases
> being dropped.
> This happened with 3.0.5 as well. Just release it already. They are
> alphas. Christ.

Well for the NOTICE.txt issue, we cannot actually make a release for legal
reasons without;a=commitdiff;h=b0a83f62
part of the release.

The question is whether it is better to have the first 3.1.0 alpha release
be called 3.1.0-alpha-4 or 3.1.0-alpha-1

When people come back and look at the git history, they will see the
commits with "[maven-release] Prepare 3.1.0-alpha-1" through to
"[maven-release] Prepare 3.1.0-alpha-4" but only see the tag for
maven-3.1.0-alpha-4 and they might incorrectly assume that somebody just
forgot to push the tag (similarly they could undelete the SVN tag, so this
is not a GIT thing) and then you could end up with a situation where the
ASF is sued for making a release of Maven without attributing the Eclipse
foundation correctly...

Yes, I know that specific example is unlikely, but the point is that there
is potential for that type of thing... and we have the mailing lists as a
record, etc.

Jenkins 1.453's borked partial release was enough to convince me that
dropping the staging repo and respining with the same version number is
probably the lesser evil.... though that might be because I had to go and
do some workarounds for some automated analysis and other fancy shit I was

I guess my point is that 3 months later I had to go digging and it took
quite some time to find out that Jenkins 1.453 was actually a failed
partial release and while there were artifacts for jenkins-core published,
there were none for jenkins-war.

Prior to 1.453 I would have been agreeing with KK's assertion that the ASF
version reuse was just madness.

But having said all that, if we can find a good way to flag versions as not
released (e.g. a release history page or something) I am not against
skipping version numbers. Might confuse people though if that meant that
the first release of Maven 3.1.0 was 3.1.4 (i.e. if we had not been doing

> -Jesse
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
> For additional commands, e-mail:

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message