Re: Our SVN layout is not that smart OR should we change the SVN layout? (take 2)

[Sam Hocevar <sam@zoy.org>]

> Well, my understanding is that experimental is a thing which is a real
> branch since most of the things should happen in trunk which is
> unstable.
>
> From time to time one actually does something dedicated for
> experimental, starting from the trunk = unstable code in branches.

   Well, I have at least one counterexample in the games repository:
etw only exists for experimental, and unstable will be copied from
experimental in due time.

> >to choose between copying trunk/ into branches/experimental, which may
> >be nonsensical if the main development still happens in the experimetal
> >branch, which will be forked from time to time to unstable, or into
> >branches/unstable, which is inconsistent with all other packages.
>
> I personally don't find it that inconsistent, and probably most
> packages will not have that much branching anyway. Moreover, do we
> want that much consistency in the SVN repo ? ;-)

   Indeed. That's why I'd like to handle the things I care about my own
way so as not to lose time adapting to other layouts. :-)

> As I said, I don't feel that stong in any way, but I think the flow
> which was followed until now is something along the lines of:
> 
> - make changes in trunk, keep suite to UNRELEASED until the release
> and in between
> - tag at each release, upload the tagged version
> - branch things like backported packages or experimental things
> - generally development happens in trunk which is unstable or unstable
> unreleased.

   And my only major objection is that subversion force-feeds us
the "trunk" naming scheme without any real reason. Replace it with
"unstable" in your above lines and I mostly agree.

-- 
Sam.