Agree'd programmers should check in code. There are some reasons
to have a 'cvs administrator' that checks what is checked in
(conforms to company standards, will not break build, does not contain
obvious errors...). The administrator, if needed, should not
check in others code.
Hmm... now that I think of it, this is exactly what Linus T. does/did:
the comunity submitted patches and Linus vetoed/ok'd them and checked
them in. He used Bitkeeper though. They have recently moved to
another system altogether now (seem to remember a /. article or
something.).
Oh. Now I see. That particular mode of operation does not
fit cvs (or subversion). See this article 'Please stop trying to
get Linus to use Subversion'
http://subversion.tigris.org/subversion-linus.html and Linus'
'SCM saga..." from the kernel mailing list:
http://lwn.net/Articles/130681/.
If you have to work like this (because of policy or geography etc.)
maybe you need to re-think your SCM plan (i.e. CVS may not be the
solution for you.) If it is because of politics and/or you have
already made a significant investment in CVS (time/effort etc.) and you
want to stick with it, you'll probably have to re-think how you will
use CVS (as the other posters have stated).
Good luck. All the best.
--Russ
Damn it all to hell! I forgot my sigline. I thought of a great
one today in the shower. I knew I should have wrote it
down. Poop on a stick. It was really good too. Oh
well.