Using bzr with pycurl on Windows

[vila]

>>>>> Eli Zaretskii <eliz at gnu.org> writes:

    > Maybe adding some debugging printout somewhere will help?

That's what -Dhttp does. The only missing part (that I looked at
locally, you could too if you're able to modify _urllib2_wrappers.py and
set DEBUG = 2 there but since you're on windows I know it's a pain) only
reveals that we receive 'incomplete headers'[2] once, treat it as a
transient failure, issue the same request again and fail.

In both cases we receive the 'incomplete headers' line.

>From there the best course is probably to file a bug against bzr-git as
github being what it is, it would be nice to have it properly supported.

AFAIK jelmer is away this week or he would have answered here I presume.

>From the bzr http transport layer there is nothing I could do without
understanding what the server is complaining about and not knowing what
the bzr-git is trying to achieve here[1] and whether or not bzr-git or
the server is wrong... I'm out of ideas ;-/

       Vincent

[1]: It just asks for the existence of a file apparently. Why this is
     supported for some files and not for some others on github is a
     complete mystery to me...

[2]: Which is complete garbage as far as HTTP and python httplib module
     are concerned.