>>>> "Andy" == Andy Piper
<andyp(a)parallax.co.uk> writes: 
Andy> 
Andy> At 01:20 PM 4/28/98 -0400, Greg Klanderman wrote:
> (setq vc-handle-cvs nil)
> 
> maybe this should be the default
> greg 
Andy> 
Andy> Well, in theory I still wanted cvs support - so I could do diffs and the
Andy> like, it was more the read-only file that I didn't want. IMHO its a bit
Andy> presumptuous for cvs to make the file read-only based on the fact that you
Andy> can't change the backend. I believe this is almost never what you want.
My (possibly incorrect) understanding is that vc.el uses the RCS/SCCS
paradigm, in which you must check out a file from version control in
order to edit it.  Only one person can check out a given file at a
time.  Hence, by default, a file is read-only - you must check it out
to edit.  CVS has a much different paradigm.  Any number of people can
be editing a file simultaneously, and the changes are merged upon
checkin.  So vc.el may not be ideal for working with CVS, hence the
variable vc-handle-cvs, and my suggestion for defaulting it to nil.
You might check out pcl-cvs or cvs.el.  Pcl-cvs operates on a project
basis, somewhat like dired.  It is packaged for XEmacs, and despite
needing some work, it is very usable and I like it.  I'll be acting as
maintainer for the XEmacs package.
From what I hear, cvs.el operates from the file buffer itself.  It is
not a package (yet).  I need to check it out at some point.
greg