Ben wrote, on the issue of separating repository from path in the WC:
> I've not yet heard people put out real arguments one way or another;
> and I'd like to hear them *now*. I'm tired of beating around the
> bush.
I think it's the same old question of mindset.
Greg S hails from a DeltaV mindset, where you have a bunch of more or
less independent versioned resources. From that point of view, the
name of the resource is the only important piece of information (well,
and the version of it you have checked out). This mindset is also
similar to the CVS mindset.
Most of the rest of us hail from a mindset where you have a
"repository" which is versioned as a whole, like in PRCS. We have
some functionality (like checkouts of subdirectories, and selective
updates) which allows users to play with sub-hierarchies of the
project, but that's just a convenience, a way of getting a narrow or
slightly inconsistent view of the larger project.
is a directory or file; it is a versioned resource; its name is
<blah>. Splitting its name into two different pieces which are only
meaningful to the back end is just extra hair.
elegant. What you have is not just a versioned resource but a
component of a versioned project with a defined root. It may be
interesting to know what this root is.
As for myself, I've been waffling a bit. I think I currently believe
in separation, because it means that ra_local doesn't have to
artificially decompose the URL into repository and path except perhaps
(depending on our user interface) for initial checkouts.
Received on Sat Oct 21 14:36:22 2006