In message <f470f68e0909151327l67b16625s24a5e78097edf...@mail.gmail.com>, Robert Burrell Donkin writes: > > On Tue, Sep 15, 2009 at 4:59 PM, Stuart Monteith <stuk...@stoo.me.uk> > wrote: > > <snip> > > > I was wondering what people (esp. mentors) thought about that. The > > code itself would be like an alpha release. This would hopefully > > encourage discussion and early adoption, and possible participation. > > > > > release often, release early :-) > > > > > > i don't know the code but IMHO it's worthwhile thinking about whether > > > 'alpha' is really the best description > > > > > > for example, IMHO > > > > > > * kato-1.0-alpha implies that 1.0 is pretty much done but the code > > > hasn't had the real life testing required to be sure that the quality > > > is there. once an alpha is shipped, i would expect the code to be > > > stable, with just bug fixes until a 1.0 > > > * kato-0.1 implies that kato is short of features and is not mature as > > > an API but that the code quality is reasonable as far as it goes > > > * kato-M1 implies that kato is immature and under very active > > > development but that this milestone offers an island of stability > > > > That last option is the most promising, I'll follow up with what I think > > we should do. > > IIRC apache harmony uses milestones. might be worth taking a look at > how they do it...
Yes, we use milestones. We use an svn revision number rather than a version number and I think this helps to set expectations appropriately. We've recently attempted to produced a java 6.0 milestone. We failed and decided to delay the release for another iteration. If we'd delayed trying at all, I think we'd still have failed next time. Just going through the motions of doing this release was hugely productive and I'm pretty confident about getting the next one out the door. So I'd certainly encourage you to think about producing a release even if you think it might be a little too early to do so. Regards, Mark.
