[BUG?] Invalid URL join request pulling from John's branch

John Arbash Meinel john at arbash-meinel.com
Wed Jun 21 22:21:41 BST 2006

Wouter van Heyst wrote:
> On Wed, Jun 21, 2006 at 01:02:19PM +1000, Michael Ellerman wrote:
>> Hey John,
>> I can't pull from your http branch:
>> concordia ~/src/bzr/mirrors$ ./trunk/bzr branch http+urllib://bzr.arbash-meinel.com/branches/bzr/http/ jam-http
>> bzr: ERROR: Invalid URL join request: ['http+urllib://bzr.arbash-meinel.com/branches/bzr/http', '../../../../../../home/jameinel/bzr/mirrors/bzr/bzr.dev/']: Cannot go above root
>> It works with an older branch I have lying around (revno 1414), smells
>> like a URL encoding bug?
> urlutils.join in branch.get_parent() fails. 
> The parent branch should in this case be reachable as
> http://bzr.arbash-meinel.com/mirrors/bzr/bzr.dev, so I'm not sure why
> the relative path is so different. 

I'm using a symlink in my home directory which isn't getting normalized.
So I did 'bzr branch ~/bzr/...' rather than doing 'bzr branch

The problem is that we aren't normalizing paths across symlinks. And
that is because we don't want to normalize anything underneath the
working tree, or we may end up pointing into the wrong branch.

> So in the case this is a valid path, which would result in a
> publically non-reachable parent, what should we do? I'm inclined to just
> return other when the join would go above the root.
> Wouter van Heyst

return 'other' or return None.

What also confuses me is why it needs to contact 'parent' at all. If I
am getting your branch, I shouldn't care who your parent is.

In the mean time, I've fixed my parent to point to something that should
be accessible.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 254 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060621/ff4fc347/attachment.pgp 

More information about the bazaar mailing list