Re: moving toward a 3.0 release

[David Bateman]

John W. Eaton wrote:
> On 27-Sep-2006, David Bateman wrote:
> 
> | I see this as a pity as I feel there are two things needed for wide
> | acceptance of octave; these being a good windows installer (ie an exe
> | and one that doesn't suffer from the cygwin sjlj exception issue), and
> | graphic handles. I'd like to eventually see classes but think that can wait.
> 
> I would also like to have improved graphics, but I should also say
> that I'm not expecting 3.0 to have everything or be perfect.  But I
> think the current 2.9.x is much better than 2.1.73, so it would be
> good to have a real release relatively soon.

Agreed

> 
> Or, we could just declare 2.9.x as a "testing" version and continue on
> with snapshots until we have integrated all the features we want.
> 
> If others think that my timing for this release is bad, then I'd be
> willing to consider having someone else take over the job of "release
> manager" for Octave.  That person would then have the job of keeping
> things more or less on schedule, deciding what projects are merged for
> which releases, merging bug fixes into released versions while
> development continues on the main branch, when to have feature freezes
> and make release branches in the CVS archive, etc.  We could follow
> the model of the GCC project (or some other, as long as it works).
> 

God please don't point the finger at me.... What I'd like is some
feedback from Bill and Shai on what they think is needed, set a feature
freeze date for octave 3.0 but not at 2.9.9, call 2.9.9 a testing
release and the first release after the feature freeze call it 3.0rc1.
If graphics handles are ready at the feature freeze then accept them,
otherwise go for a release without them.

The only question in the above is when is the feature freeze. Can we
just wait for feedback from Bill and Shai before fixing that date?

BTW does 3.0rc1 sort alphabetically after 3.0? if so then perhaps we
need to consider the name of the release careful so as not to fool the
package manager.

Cheers
David