Re: so, 3.x VCS

On ons, 2008-01-16 at 09:46 -0700, Alex Rousskov wrote:

> IMHO, we should wait until bzr 1.x is widely spread and, hence, better
> tested and documented. With folks pushing for Squid 3.1 release soon and
> a few large branches not integrated yet, I would rather not spend time
> on learning a new VCS and struggling with relatively immature software.

The branch integration is pretty much irrelevant to the discussion.
That's done using a separate CVS repository and exchanging patches
today, and can continue doing the same after switching the main
repository to bzr.

The big thing is that moving to bzr makes it a lot easier to work with
future branches, both locally and globally.

For the normal work of committing stuff to Squid-3 HEAD there is just
some new commands to learn, and is all you need to know until you start
actually making use of the new branch features bzr brings to
development. You can still continue using the cvs branches at
devel.squid-cache.org if you like. It's not like we are doing an
all-in-one-forget-everything-old switch. It's a switch of the main
repository to a VCS which has natural support for branched development.

The needed bzr versions (0.92 or later) is in all major and most minor
free/open OS distributions.

So the only arguments not to switch is time. The time needed to learn
2-3 new commands for working with Squid-3 HEAD, and the time needed to
verify that the tree has been properly migrated from CVS. The release
manager for 3.0 (and 3.1 when branched) will need to know a little more
to merge stuff over from HEAD to 3.0, but not much.

When you have time you can learn the new wonderful features a more
modern vcs brings which can ease workflow considerably, but you don't
need to in order to start using it.

Regards
Henrik
Received on Thu Jan 17 2008 - 02:59:54 MST