> Firstly, I can't get that change into Maven trunk/branch until wagon would be released with it's changes.
Agreed, but I do not have any idea, how to solve the problem without any API changes. The current API depends on the assumptions that a) there is exactly one protocol being used by the wagon provider and b) the Wagon user is responsible to know which protocol that is. Both assumptions are, IMO, plainly wrong, as the webdav provider obviously demonstrates.
> However, the upshot of the patch is that the webdav wagon ignores the proxy that is passed in.
I can't follow you here. The old proxy configuration is still in place and the patch should take care that the old proxy configuration is used, if no ProxyInfo is present.
> I think it would be better to take the earlier solution provided by Jochen where:
> a) we fix the wagon manager to pass the correct proxy in (presumably it's giving dav instead of http - surely we can figure that out), or otherwise
You refer to the solution proposed by Joakim, not Jochen? It is up to you to determine whether such an ugly workaround should be installed. My strategy would clearly be to push out a wagon release ASAP (perhaps even a version 1.0-beta-2.1, which could be identical to 1.0-beta-2, apart from this patch) and use that in the Maven core, if 1.0-beta-3-SNAPSHOT is undesirable.
> b) pass in all the proxies and let the wagon decide which to use.
That is, IMO, exactly, what the ProxyInfo does.