On Fri, Sep 12, 2008 at 5:29 AM, Pierre-Arnaud Marcelot <pa@marcelot.net> wrote:
Hi Emmanuel,

On Fri, Sep 12, 2008 at 11:13 AM, Emmanuel Lecharny <elecharny@gmail.com> wrote:
Last minutes changes are always bad, but this is something we have to deal with...

Yeah of course...
Can I suggest that before entering in release mode, we do a code freeze just before ? As we have a 72 hours period between the vote and the real release, the code freeze period would be used to avoid some last minute changes during the release.

Yeah, sounds great. Maybe we could create candidate branches, so we can continue to work on the trunk during those 72 hours. If something is wrong for the release, we can fix in these candidate branches.
Also, I would say that the release manager will always be blamed for what he breaks, but he is not responsible, as he is just the guy who paint the house, and notice the cracks in the wall...

Yeah, it's true and I was absolutely not blaming Alex. ;)

No no I FU here and should have left things as is.  I noticed the artifactId and module name discrepancy.  But we can retag and regenerate the installers and push them out again. 

I know very well, how hard it is to cut a release... :D
(Especially when maven plugins try to prevent you from releasing)

Not that bad I'm just a retard these days.

BTW, I checked out the code and fixed it, so the installers are working fine now.

Here's what's working and has been tested so far:
- Windows installer
- Mac OS X installer
- Solaris (x86 and sparc) installers
- Archive (zip and tar.gz) installers (on Unix and Windows)

Here's what's need to be tested:
- Binary installers
- Debian installers
- RPM installers

I'll check these out.  Did you deploy these to the dist area?  Remember we'd have to resign them.