> Imagine you want to develop a massive new feature for Postgres. You
just
> create a branch on your own machine, do all your changes, commits,
etc. and
> keep it current with the main branch. Then, you can merge it back
into the
> main tree... That way you can have a history of commits on your own
branch
> of the repo!
The same thing can be accomplished with CVS as well -- it's just not
as pretty. There is a reason that the FreeBSD group uses $FreeBSD$
and leaves $Id$ untouched.
Basically, check out of one, drop CVS directories, check into the
second, check out of the second, and when doing work with either
repository you specify which repo with the -D flag. Coupled with
the -j (merge) flag you can accomplish most tasks.
That said, if the work was thought through and beneficial you may be
able to obtain a branch in postgresql cvs to work with.