[Protux-devel] Regarding Release Coordination

[Remon Sijrier]

Hi all,

I had some little discussion about this with Reinhard and unfortunately this 
conversation didn't go via protux-devel and I deleted those mails per 
accident. So I have to do it with my memory ;-)

Most important point was this:

"Do not release protux until everything is finished/tested/updated etc.."

I understand this puts a high limit on development when nearing to a release, 
but it has been a month ago since 0.20.0 is out, and the website and docs are 
still _very_ out of date....
Of course, protux is meant to in development mode til 0.50.0, but if you rely 
to much on this, it doesn't make sense to release "major" releases at all


So I suggest to do it different in the future:

Split CVS into "stable" and "head" This must be somehow possible. When nearing 
the next major release, fork CVS into "stable" and "HEAD", freeze the 
"stable" branch and continue development on "head". Only bugfixes will be 
excepted for the "stable" branch. (which makes it for me much easier to add 
bugfixes to a release. It's very hard for me to do it now. Also for other 
people to download the bugfixes to test it)
Then go for testing "stable" until there are no more important bugs, and 
_after_ the docs and website has been updated, announce the new release.

One of the things which happend just before the 0.20.0 release was that there 
were a bunch of features added which introduces new bugs. That's of course 
annoying and IMHO should be avoided. With the release structure mentioned 
above this will no longer happen, and also prevents the developer to do such 
things.

I also prefer to have something like a "release coordinator". Although it may 
seem a bit of overstructuring the whole process, it puts of the load from the 
developers, and a release only comes when it is time....


Thanks and RFC please.

Greetings,

Remon