I would imagine that a branch which is ready to be released would be
quiessed and so there would be no need for an svn rev # or am I
missing something?
Regards,
Alan
On Dec 21, 2006, at 3:19 PM, [EMAIL PROTECTED] wrote:
IMO using a svn rev # for a release is a good idea, that with a tag
ensures that code for that exact release can always be found at a
later time.
--jason
-----Original Message-----
From: Matt Hogstrom <[EMAIL PROTECTED]>
Date: Thu, 21 Dec 2006 18:10:46
To:dev@geronimo.apache.org
Subject: Re: [vote] Release geronimo-jpa_3.0_spec-1.0
On Dec 21, 2006, at 4:06 PM, Guillaume Nodet wrote:
I think voting on svn source for small projects / jars is good,
because people can build them locally, check that everything
is ok (for legal reasons), and vote. This is much more difficult
for Geronimo server, of course, and may not be applied.
This works well, I think, if the release process is just
mvn release:prepare release:perform
which should be the case for all projects ideally.
The benefit is that the jars will be deployed to their final
destination
as part of the relase, without having to tweak / corrupting maven
repository metadata by copying from a staging repo.
For my part, I'd prefer to follow this approach going forward. I
agree with Guillaume that it may not totally work for Geronimo unless
we choose an SVN number as the release point so people can track
changes to a branch.
Having been through the release process a few times I think that
using Maven to generate the artifacts is so much simpler and
automagically updating the repo is far easier as well.
I'd like to propose (in a separate thread) that we adopt this process
going forward for specs. If the vote succeeds then I think David
could follow it for these specs as a test case.
Matt Hogstrom
[EMAIL PROTECTED]
