> I've had a closer look at the code now. It looks fairly reasonable, though
> there aren't a lot of Transport-level tests (is that done by
> bzrlib.tests.per_transport ?).
Yes, most of the testing comes from that, only webdav-specific parts are in the plugin.
>
> It's a bit surprising to see _raise_curl_http_error despite the backend being
> based on the urllib HTTP implementation. Is that a bug?
Well, more probably dead code.
If I'm not mistaken, these calls can only be triggered if an unexpected error code is returned which is *already* handled via accepted_errors=[...] in the requests definitions.
I've started looking at the coverage and will use this work to validate the above. Putting to wip until then.
> I've had a closer look at the code now. It looks fairly reasonable, though tests.per_ transport ?).
> there aren't a lot of Transport-level tests (is that done by
> bzrlib.
Yes, most of the testing comes from that, only webdav-specific parts are in the plugin.
> curl_http_ error despite the backend being
> It's a bit surprising to see _raise_
> based on the urllib HTTP implementation. Is that a bug?
Well, more probably dead code.
If I'm not mistaken, these calls can only be triggered if an unexpected error code is returned which is *already* handled via accepted_ errors= [...] in the requests definitions.
I've started looking at the coverage and will use this work to validate the above. Putting to wip until then.