Release Process

[Jasper Huijsmans]

On Wed, Sep 22, 2004 at 03:52:31PM +0530, Biju Chacko wrote:
> Jasper Huijsmans wrote:
> >Structure is good, three month is too short ;)
> >
> >Maybe we should try to keep to 6 month. We could (should?) start doing real
> >alpha releases after three month. With real alpha I mean, no feature 
> >freeze,
> >and no API stability guarantee.
> >
> >Trying to define goals is good too, but pretty hard. Especially since  
> >none of us can guarantee how much time they will be able to spend on it. 
> >Well, I
> >can't at least ;-)
> 
> Alternatively, we should follow the dual release philosophy followed by 
> many projects. Have regular development releases. I don't think that 
> would add much overhead to the system ...
> 

That's what I meant. Of course it depends largely on how many problems are
encountered with the stable branch, how much time we have to work on new
features.

What I called real alpha releases are simply tarballs generated from CVS HEAD 
with no guarantees. I suggest a period of three month before doing such 
releases to be able to flesh out some new features first.

> As it is we have a separate stable branch, all this would add is the 
> process of creating tarballs of CVS HEAD.
> 
> The point is that the life of a project is often judged by the 
> recentness of it's releases.
> 

We've done 8 releases in a year, that's not bad. Just no feature additions.

> And we really should reduce the entry barrier for running our latest 
> code. It is in our interest that as many people as possible test our 
> code. Limiting that number to the people who have the inclination to 
> pull from CVS is probably not a good idea.
> 

This gets more important as time between stable releases increases ;-)

	Jasper