[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Subversion 1.5 beta1, "malformed URL for repository"

From: Julian Foad <julianfoad_at_btopenworld.com>
Date: Tue, 09 Sep 2008 19:19:20 +0100

On Tue, 2008-09-09 at 09:08 +0400, Fyodor Sheremetyev wrote:
> On Tue, Sep 9, 2008 at 1:23 AM, Julian Foad <julianfoad_at_btopenworld.com> wrote:
> > 1) Can you (or someone) confirm that the test fails without the fix, and
> > passes with the fix? I can only test on ra-local and ra-svn at the
> > moment, and this new test passed both, even without the fix. I'm not
> > sure if that's expected. I guess it means those protocols don't check
> > for well-formed URLs.
> >
> > If the test and the fix are behaving, we should commit this.
> I can confirm that the test fails on http without the fix and succeeds
> with the fix.

Thanks. I have now committed your patch in r33003.

> > 2) Having become aware of this problem, if we look further, there are
> > lots more calls to svn_path_join() that look wrong. Especially, 'grep
> > "url.*svn_path_join"' in Subversion source code reveals a bunch that
> > mostly look wrong. It also reveals a few that use uri_encode() on the
> > second argument that could be replaced by svn_path_url_add_component().
> >
> > Do you fancy (a) fixing those and/or (b) figuring out tests for
> > exercising those?
> Yes, I'm going to take deeper and wider look at the encoding/decoding
> problem after this patch is reviewed and committed.


> > 3) Can we modify our test suite to include paths with spaces in, and
> > thereby catch many URI encoding errors automatically? We can add
> > specific tests for specific cases, but I'm wondering if we can modify
> > the Greek Tree to include some nasty names that will get exercised
> > through several different code paths.
> Having all such cases embeded in the whole test suite might make it
> more complex to understand and maintain. I think it's more of a broken
> design than a bug and can be fixed by explicitly setting boundaries
> which would care about URI encoding (and the rest of the code would be
> using decoded URIs).

- Julian

To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-09-09 20:19:39 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.