[dev] Release Engineering: Versions

Eric Rostetter eric.rostetter at physics.utexas.edu
Fri Dec 12 11:34:11 PST 2003


Quoting Jan Schneider <jan at horde.org>:

> With 30+ modules eventually being released, we need to think about our
> branch naming. We would have stable branch names from RELENG_1 up to
> RELENG_4 and a pretty complex matrix of which versions goes with which
> other modules.

While the matrix being complex is bad, I don't know if the naming scheme is.

> Already now a lot of people are confused about which versions work with each
> other, and even I have to think a few seconds before I can tell. For some
> modules I rarely use I even don't know at all off my head.

Mostly they mix RELENG and HEAD.  That is the biggest problem.  Given
that, I'm not sure how important naming within those categories are.

> Additionally we have two modules (Ingo and Klutz) that already have a
> RELENG_1 branch but no releases, so that the first releases of these
> modules have to be 2.0 or we would release the 1.0 versions from RELENG_2
> branches. Still following? :-)

The branch can stay RELENG_1 when they are released as 1.x versions.

> As the possible combination of module versions mainly depends on the Horde
> version, I suggest a new brach naming theme that is based on the major
> Horde version.

I think I even suggested this before.  But we've been on the other
system so long I'd hate to have to change now.

> Something along MAJOR_3 OR FRAMEWORK_3 for all stable
> branches of to be released modules and *_4 when the next Horde version
> comes out.

I see no reason for this.  In fact, it would mean we would have to release
a new version of all modules each time a new Horde came out, even if there
was no need to do so otherwise.  So if, for example, the "current" version
of "accounts" module works with both version X and Y of horde, I shouldn't
have to release a new version of "accounts" just to make the name match...
Will that situation ever arrive?  Maybe not. But it might...

Just my $0.00002 worth...

> Jan.

--
Eric Rostetter
The Department of Physics
The University of Texas at Austin

Why get even? Get odd!


More information about the dev mailing list