[0.14] bzrlib.errors.TransportError: Transport error: Server refuses to fullfil the request

[Vincent Ladeuil]

>>>>> "john" == John Arbash Meinel <john at arbash-meinel.com> writes:

    john> Jari Aalto wrote:
    >> I'm having a problem pulling from my own HTTP server. Deteils below.
    >> 
    >> - Could the error message be a little more verbose. Like printing 
    >> possible reasons for and suggest actions to check.
    >> 
    >> In the server size all files and firs are drwxr-xr-x, I've also
    >> restarted apache2 but this doesn't affect the error. Perhaps this is
    >> an error in my side, just can't read it from the error.
    >> 
    >> Jari
    >> 

    john> I agree it should describe what is going on. You are getting a 403 error.

    john> Try the attached patch. I think I also found a bug in the 404 handler
    john> (it was directly using HTTPError, but I couldn't find that imported
    john> anywhere. Which surprises me a lot, because we do quite a few NoSuchFile
    john> tests.)

You found a bug, but it's dead code, see below.

    john> John
    john> =:->

    john> === modified file 'bzrlib/transport/http/_urllib2_wrappers.py'
    john> --- bzrlib/transport/http/_urllib2_wrappers.py	2006-12-13 15:08:39 +0000
    john> +++ bzrlib/transport/http/_urllib2_wrappers.py	2007-02-11 13:43:48 +0000
    john> @@ -761,16 +761,19 @@
    john>      def http_error_default(self, req, fp, code, msg, hdrs):
    john>          if code == 404:
    john>              raise errors.NoSuchFile(req.get_selector(),
    john> -                                    extra=HTTPError(req.get_full_url(),
    john> -                                                    code, msg,
    john> -                                                    hdrs, fp))
    john> +                                    extra=urllib2.HTTPError(req.get_full_url(),
    john> +                                                            code, msg,
    john> +                                                            hdrs, fp))

This is dead code, the 404 code is filtered just above in
HTTPErrorProcessor.http_response (it's a bit hard to read because
it uses the call chains from urllib2: the self.parent.error will
end calling HTTPDefaultErrorHandler.http_error_default.

404 is not considered an error for urllib because we want to be
able to translate it into NoSuchFile, but not here.

So it can be safely deleted.

    john>          elif code == 403:
    john> -            raise errors.TransportError('Server refuses to fullfil the request')
    john> +            raise errors.TransportError(
    john> +                'Server refuses to fullfil the request for url %s'
    john> +                ' (code: %s, msg: %s, headers: %s)'
    john> +                % (req.get_full_url(), code, msg, hdrs))
    john>          elif code == 416:
    john>              # We don't know which, but one of the ranges we
    john>              # specified was wrong. So we raise with 0 for a lack
    john>              # of a better magic value.
    john> -            raise errors.InvalidRange(req.get_full_url(),0)
    john> +            raise errors.InvalidRange(req.get_full_url(), 0)
    john>          else:
    john>              # TODO: A test is needed to exercise that code path
    john>              raise errors.InvalidHttpResponse(req.get_full_url(),

+1 to merge that.

   Vincent