Yang Tse wrote:

> But as a first impression I would have expected different results.

Why? As I see it this change only affects the code path where SSPI is
involved, so --disable-sspi behavior shouldn't change, as libcurl
handles the NTLM hashing itself in that case, no?

> I would prefer to have sorted out first the NTLM API actually needed.
> But at this point I still don't see the need for the patch (yours, nor
> mine).

The need for the patch is simply that Curl fails to build at all with
--enable-sspi on Cygwin at the moment. I didn't really intend to start
debugging NTLM (something I know nothing of), I just wanted to update
the Cygwin curl packages to include SSPI support. Of course, I
recognise that there may be an underlying problem and that simply
including windows.h may not be the right fix. So it looks like we need
somebody with a machine on a NT domain that can test this, which
unfortunately I don't have.

Brian
Received on 2007-04-09