Ben Barrett wrote:
> > very neat. CVS is really lacking for some things....
> Seth (others too?), what do you want cvs to do?
> I've used it for 3 projects, just the basic command-line
> tool w/ options, but notice thatadd-ons like
> sourceforge itself, and emacs frontends add a lot
> of usefulness...
>
> I've been noting-but-pleased with it, and wonder what else
> I'd want from it...
I'd better not tell you then, no reason to spoil your contentment. (-:
Take a look at the top level page on the subversion site for what
the subversives want to fix. http://subversion.tigris.org/
But the two main things are:
1. Commits are atomic at file level, not project level.
That means that if I change three files and check them in, the
repository goes through three states, only one of which is
consistent. Worse, if I'm checking out a big project across a
slow network and it takes an hour, I'll get versions of files
from anytime in that hour. In some projects I've worked on,
it's been basically impossible to get a consistent snapshot,
because files were coming in faster than I could check out and
test a tree. The IRIX kernel tree was that way. (SGI used a
homegrown alternative to CVS, but it was similar enough that
it had the same problem.)
2. CVS doesn't support renaming files.
When you rename a file, CVS thinks you deleted the old one and
started a new one. That means all the edit history of the old
file is not associated with the new file.
--
K<bob>
[EMAIL PROTECTED], http://www.jogger-egg.com/
