On Mon, 23 Feb 2009 08:45:27 +0100, Daniel Stenberg wrote:
>
> I think it should be. Feel free to poke on the source to make it so and
> send us the patch!

;-)

> We've fixed some Negotiate-related things since 7.18.2 so it's not
> unlikely that the bug is gone or at least different in 7.19.3.

Hrm. I've yet to build it to see how it fares but just looking at the
source, I'm still seeing the conn->data->->state.negotiate vs. conn->data-
>state.proxyneg discrepancy in 7.19.3. To re-iterate:

output_auth_headers() is testing data->state.negotiate.context for
validity before calling Curl_output_negotiate(conn, proxy) with proxy ==
TRUE, but in Curl_output_negotiate() if proxy == TRUE, neg_ctx is set to
conn->data->state.proxyneg, not conn->data->state.negotiate (which is
used for the FALSE case).

> Negotiate is a tricky procotol since we don't have anything in the test
> suite that verifies its functionality, and most of us main devs seem not
> to not use it.

That's fair enough. I'm certainly willing to debug it here with some
help from people more generally knowledgeable about the code.

Cheers,
b.

-------------------------------------------------------------------
List admin: http://cool.haxx.se/cgi-bin/mailman/listinfo/curl-users
FAQ: http://curl.haxx.se/docs/faq.html
Etiquette: http://curl.haxx.se/mail/etiquette.html
Received on 2009-02-23