cvs and ssh

[Dean Gaudet <dg...@hotwired.com>]

ssh has another feature that is useful -- it allows you to specify a command
to execute when someone connects via rsh.  This prevents the rsh from
doing anything but the one command you want.  The example they give is
for accessing remote tape drives.  In our case the command could be
'cvs server'.  That way the developers do not need full shell accounts.

To use cvs for what you folks want you'll have to agree on a standard for
what needs to be 'tagged' and what the tags will look like.  This is so
that you can unroll things that don't work right (or belong on another
branch).  I don't have any great suggestions... mostly because I've
never seen a group of people who debate so much over every patch :)

A working copy of a cvs project has the great property that it is portable.
For example, if you're in the middle of nowhere (.co.uk for example)
and your link to the repository is down, and you've just finished a patch.
You want to work on something else, but don't want to roll all your
changes into one commit.  You can just copy your source tree, and then
keep working.  Later you go back and commit your source trees in the
order you created them.  You can make a copy any time without contacting
the repository.

Laters
Dean