Re: URLs again

[Karl Fogel <kf...@galois.collab.net>]

Jim Blandy <ji...@zwingli.cygnus.com> writes:
> To be clear: Karl agrees that there should be some form of URL people
> can use to refer to individual files or subtrees in a repository.
> When someone wants to check out your project, you can just hand them a
> URL.
> 
> The question is whether we should require the URL to have a form that
> allows one to syntactically determine where the Subversion filesystem
> path starts --- in other words, to find the URL for the filesystem's
> root.

+1

You should be able to hand someone a single Thang, which when fetched,
gets them the versioned resource.  But that thang has two behaviorally
distinct components: the repository, and the versioned path within the
repository.  By merely looking at a thang, a person can tell whether
or not what they're fetching is part of a larger versioned context.

I've often found this property handy with CVS repositories...

Re: URLs again

[cm...@newton.ch.collab.net]

Karl Fogel <kf...@galois.ch.collab.net> writes:

> You should be able to hand someone a single Thang, which when fetched,
> gets them the versioned resource.  But that thang has two behaviorally
> distinct components: the repository, and the versioned path within the
> repository.  By merely looking at a thang, a person can tell whether
> or not what they're fetching is part of a larger versioned context.

+1, perhaps because I just like the notion of the CVS module, even
though my interpretation of such is that modules are just immediate
subdirectories of the repository, and I think that concept disappears
in the URL-only schema.