[PATCH] Fix for Issue1260

[Matt Mackall]

On Fri, 2008-08-15 at 15:53 +0100, Frank Kingswood wrote:
> Adrian Buehlmann wrote:
> > # HG changeset patch
> > # User Adrian Buehlmann <adrian at cadifra.com>
> > # Date 1218803955 -7200
> > # Node ID fcb89cadd5530f8d680b8bca9235d5bc9f3f4fd5
> > # Parent  438e02b4be73ffd80d4a6cc45b7c9273d6b26604
> > Fix for Issue1260
> > 
> > diff --git a/mercurial/statichttprepo.py b/mercurial/statichttprepo.py
> > --- a/mercurial/statichttprepo.py
> > +++ b/mercurial/statichttprepo.py
> > @@ -24,7 +24,7 @@
> >  
> >  def opener(base):
> >      """return a function that opens files over http"""
> > -    p = base
> > +    p = base.replace('\\', '/')
> >      def o(path, mode="r"):
> >          f = "/".join((p, urllib.quote(path)))
> >          return rangereader(f)
> 
> This seems the wrong place to handle this. Shouldn't the callers be 
> fixed instead, so that statichttprepo gets valid URLs?

Let's back up and look at the big picture:

A repo object uses a store object to figure out where its files are. A
store object uses an opener object to open those files. The typical
opener object takes local paths.

Now a statichttprepo looks just like a localrepo but with one
difference: it's accessed via http. We use a special opener to open our
files. So at some point we do something like this in store:

os.path.join(path, "store")

Which is a perfectly reasonable thing to do, as openers generally take
local paths. But path here ends up being our http URL.

So we can either fix it here with a small hack or rethink the way
openers and paths work.

-- 
Mathematics is the supreme nostalgia of our time.