Comment 3 for bug 207558

> We should:
> * fix the bug in _SmartClient.remote_path_for_transport. I have a
> branch that does this, I'll associate it with this bug.

I've submitted this to the list for review.

> * land the fix to bug 124089

This has now landed.

> * probably explicitly disable the HTTP smart auto-probing in the
> client for a release, to give existing broken smart server deployments
> a chance to deploy a fixed bzr server. We can probably turn this
> feature on in 1.5.

At Robert's suggestion, I've submitted a patch that does something a bit different: the auto-probing by default stays, and a "nosmart+" transport decorator is added. So you could do "bzr branch nosmart+http://host/..." to work around a misconfigured HTTP smart server.