>>>>> "Amir" == Amir Karger <[EMAIL PROTECTED]> writes:
Amir> On Mon, Oct 04, 1999 at 05:33:49PM +0200, Jean-Marc Lasgouttes
Amir> wrote:
>> - 1.x.0 version will not happen very often, so people will be
>> encouraged to upgrade
>>
>> - all versions are (supposed to be) stable versions, so adventurous
>> users can try them out.
>>
>> Seriously, this scheme will not be worse than what happens
>> currently with preversions between stable versions. We just dropped
>> the notion of a development branch, which needs much more core
>> developpers than what we have currently.
Amir> Ah. You know, I've been wondering how this was any different
Amir> than our current scheme. In fact, if you think about it, this
Amir> sort of happened with 0.10. 10.7 was around for over a year
Amir> while the 0.11's were being worked on, and I think many many
Amir> users were using the late 0.11's, since they had so many better
Amir> features. So we are kind of using this scheme already.
Yes, but we should go much faster this time. A new stable release is
not supposed to take a year to happen (meaning we should not feel
pressed to add a lot of feature to justify the fact that this is a
major release). It will rather be a one-or-two-features-at-a-time
development cycle (if we can refrain ourselves from adding a lot of
things).
Amir> OTOH, I'm not sure I'd get rid of prereleases. They let people
Amir> know that a new version is coming soon, which encourages them to
Amir> check back (as long as we're honest and really only start
Amir> prereleasing when we really are close).
Yes, we can see how to do that.
Amir> The only problem I foresee with this is that even courageous
Amir> users might be afraid to download the latest 1.x.y version.
Amir> Maybe we should have a list of stable y's within a given 1.x (on
Amir> the new and improved web site, of course). That way, a user who
Amir> wants new features but doesn't have time for bugs could download
Amir> 1.x.y for the last stable y.
What I see is:
- everything is supposed to be stable
- x.y.0 are extra-stable.
JMarc