Re: CVS front-ends

[Stephen Leake <stephen_leake@acm.org>]

Benjamin Ketcham <bketcham@drizzle.com> writes:

> Ludovic Brenta <ludovic.brenta@insalien.org> wrote:
> > Stephen Leake writes:
> >> If Meta-CVS is a purely client tool, then each developer can use it
> >> or not; it doesn't have to be a project-wide choice.
> > 
> > Yes, it has to be project-wide.  The way Meta-CVS works is this:
> 
> Oh goody, more layers.
> Shouldn't it really be called "Meta-Meta-RCS"?
> 
> > - every file in your source tree gets a unique ID
> > 
> > - every file is stored in the root directory of the module in the CVS
> >  server under that ID
> > 
> > - a separate text file establishes the mapping between the unique IDs
> >  and the path name of each file
> [...]
> 
> Seems eerily reminiscent of Win32 "long filenames"....

Yes.

> Don't get me wrong, I think CVS is a half-baked hack with some
> amazing oversights and omissions. 

Ok.

> (And is the documentation situation still as frustrating?) 

I find the documentation adequate.

> Still, it's the de facto standard half-baked hack, and by now all
> the build-managers have written their own personal versions of the
> scripts necessary to automate the common operations that CVS
> overlooks or makes needlessly tedious/counterintuitive. 

Yes. That is the CVS philosophy; provide basic features, let people
write higher level tools that fit their process.

> This "Meta-CVS" would seem much more appealing if it *were* purely
> client-side; has anyone attempted CVS improvements that way (I mean
> more than just GUI front ends)?

I've written some small extensions to Emacs pcl-cvs (aka pcvs). In
particular, I have one function called cvs-dtrt, for "do the right
thing" :). Given the pcl-cvs display, it invokes the appropriate
action for the file under the cursor, or prompts the user if there is
more than one.

A minor function, but quite a productivity boost.

See http://www.toadmail.com/~ada_wizard/

-- 
-- Stephe