Daniel Stenberg wrote:
>> This is a pretty ordinary HTTP fetch using the MULTI interface and
>> basic authentication. All libcurl compiles include C-ARES support,
>> and I --disable all protocols except vanilla HTTP support.
>
> Have you tried doing this without c-ares? It seems like it could be a
> source of error here. (Not saying it is, just that it would be good to
> rule it out.)

I'll roll a non-c-ares build for testers and report back within a
couple of days I hope. From a cursory inspection of libcurl,
the non-proxy case of CURLE_COULDNT_CONNECT can't happen until
resolution has already succeeded (right?), which is why I haven't
been quick to suspect c-ares.

>> One of the other ideas I had was that perhaps Win2K accidentally
>> always sets one of the "standard" proxy environment variables for
>> completely unrelated purposes, but I haven't been able to verify that
>> yet. Seems a long shot.
>
> I doubt that is the case.

Inspecting a tester's env vars, it doesn't seem to be the case.

I'll try that non-c-ares test.

Thanks,
--Adam
Received on 2005-03-03