15 Apr
2008
15 Apr
'08
12:36 p.m.
On Mon, Apr 14, 2008 at 1:43 PM, Vincent Massol <vincent(a)massol.net> wrote:
On Apr 14, 2008, at 1:35 PM, Jean-Vincent Drean wrote:
We haven't wait that long for 1.3.1 and got more commits AFAIR.
+0
About releases votes and votes in general it would be good to follow
the 72h rule from now.
It would reduce the risks of important changes at the last second and
force us to plan things a bit more.
My view is a little bit different. If we wait a long time then you'll
get more stuff committed at the last moment (check XE 1.3.2 for
example, there's been quite a few commits since the vote was issued
and thus those who voted should technically all vote again!). What we should do is stage releases, i.e. prepare a
release and only
ask for the vote when the version has been released in a staged repo.
In that manner, the vote will mean that users/committers should test
the release and vote on releasing it as is.
Having stage releases would be nice, even if I'm not sure commiters
would have the time to test them properly.
I realize the problem we discuss here is mostly related to xwiki-core
bugfix releases. Since 4+ products relies on them the release votes
are sometimes taken as a reminder thus product developers commit last
minute changes at this moment.
May be the best solution would be to keep our release/vote process as
it is and defuse the risk of last minute commits by sending a reminder
email 3 days before the vote (only for bugfix releases). Wdyt ?
Thanks,
--
Jean-Vincent Drean