On Fri, 2005-04-01 at 18:39 +0100, Philip Martin wrote:
> Ben Collins-Sussman <sussman@collab.net> writes:
>
> > How is my hand-test different than yours?
>
> You need to create a file that replaces a deleted directory.
>
> > The other thing I don't understand about your repro recipe: in your
> > bash script, I see no '-r' argument in your final 'svn cp' command,
> > yet you show that in the failure. So what exactly *is* the repro
> > recipe here?
>
> The copy without a -r creates a file that replaces the deleted
> directory. The copy that fails was not part of the script, I put it
> explicitly in the email:
>
> $ svn cp -r3 http://localhost:8888/obj/repo/foo/z.c http://localhost:8888/obj/repo/x1
>
> or in script terms
>
> $SVN mkdir $REPO/foo -m "r1 a directory"
> $SVN import Makefile $REPO/z.c -m "r2 a file"
> $SVN import Makefile $REPO/foo/z.c -m "r3 another file"
> $SVN rm $REPO/foo -m "r4 delete directory"
> $SVN cp $REPO/z.c $REPO/foo -m "r5 replace with file"
> $SVN cp -r3 $REPO/foo/z.c $REPO/x1 -m "r6 fails over dav"
>
> The problem is that $REPO/foo is now a file in HEAD. I have no idea
> why svn_ra_check_path for r3 of $REPO/foo/z.c should look at HEAD,
> even if there is a path that matches the request there is no guarantee
> that the HEAD path as any relation to that in r3. The only thing that
> can be guaranteed to match is the repository root.
I just created a new copy_test (r13834) that triggers this bug over DAV.
Thanks Philip.
-Fitz
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Apr 1 20:21:30 2005