On Thu, 2004-11-25 at 00:06, Justin Erenkrantz wrote:
> --On Wednesday, November 24, 2004 10:36 AM -0600 Ben Collins-Sussman
> <sussman@collab.net> wrote:
>
> > Sure there is. It adds complexity, and I maintain that it's a feature that
> > no linux user cares about.
>
> +1.
>
> BTW, if we now start to base64 encode the paths in .svn/entries, it means that
> normal human beings won't be able to read the paths. I think that would
> outweigh any benefit of this feature. Even if it's 'conditional' on the
> failure of the is_xml_safe() check, it's still goofy because we'd likely break
> DAV clients who would choke on these names (or even more hilariously: check
> them out as their base64 name??). -- justin
These are DAV clients who wouldn't be linking against svn libraries? If
so, yes I hadn't considered this specific situation. But it seems
similar to the problem of symlinks, what does a Windows svn client do
while checking out a symlink? Or even a filename with a trailing
underscore or a ':' or '\n'? All of these are currently passed by
is_xml_safe() and could cause problems for some DAV clients.
Peter Lundblad raised the point of having a separate function to verify
acceptable paths in a comment on the issue
http://subversion.tigris.org/issues/show_bug.cgi?id=1954
Since svn_xml_is_xml_safe() doesn't currently support non-ASCII
characters, using it as a check could prevent Subversion users from
using non-ASCII pathnames.
Sameer
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Nov 25 02:23:21 2004