Philip Martin <philip@codematters.co.uk> writes:
> Karl Fogel <kfogel@newton.ch.collab.net> writes:
>
> > > When I started the access baton stuff, and started storing/comparing
> > > paths I immediately had problems like this. Most of the libsvn_wc
> > > code doesn't appear to care whether paths are canonical, so my guess
> > > is that non-canonical paths get in.
> >
> > So if "canonical" means "has a trailing slash" (which I'm not sure it
>
> The opposite. svn_path_canonicalize simply removes a trailing slash
> if there is one.
>
> > does, but that's a different question), then the real fix is to fix
> > the callers of libsvn_wc, not to start patching up libsvn_wc to accept
> > less-than-canonical patches...
> >
> > If you want to detect them and *error*, that's fine. But I think
> > silently papering over them is a mistake. We need to discipline the
> > callers.
>
> OK, so libsvn_wc can assume its paths are canonical, what about other
> layers? Where should the canonicalization occur, in the application?
> In libsvn_client? At the libsvn_wc interface?
I think that canonicalization should fall at the same line as
case-correctness and UTF-8 conversion and URI-encoding ... it is the
job of the client binary to supply such to libsvn_client (and
libsvn_whatever as the case may be).
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Aug 1 21:37:20 2002