Carsten Ziegeler wrote:
Nicola Ken Barozzi wrote:...
...We should start of taking the habit of voting the release of the actual tarballs.
Now, how do we want to do this practically. Imho this would require that we build the tarballs and give people time to test them (some days) so they can really vote on them.
A typical scenario is that the release manager uploads the tarballs along with the checksums on his private web space at Apache, and that a vote is done on those after - let's say - 72 hours. If the vote is ok then the release manager can simply move the files in the dist section.
My naiv understanding was that we all test the tarballs during our
code freeze, so actually I thought we vote on a "virtual tarball" so to speak.
Going 'real' instead of 'virtual' could help us not repeat the same mistake we have done now. It makes it possible to have the actual tarball be reviewed by all that want to (and that should if they vote IMHO).
--
Nicola Ken Barozzi [EMAIL PROTECTED]
- verba volant, scripta manent -
(discussions get forgotten, just code remains)
---------------------------------------------------------------------