On Monday 28 of June 2010 22:52:23 Daniel Stenberg wrote:
> Looks good to me. I think we have enough test cases for NTLM to be able to
> rather quickly verify that it doesn't break anything and that the new code
> works good enough.

Thanks for review! As for the test cases, unfortunately none of them covers
the newly added code. If I comment out the MD4, MD5 and DES, all the tests
still pass fine. It's probably a consequence of the following change:

http://github.com/bagder/curl/commit/be285cd

So I tested it only with test67, test81 and test169 for exact match against
the output I got from the OpenSSL and GnuTLS alternatives. I have no idea
how to automate it easily, while keeping it still independent on the
hostname.

> The #ifdef set of http_ntlm.c could be a subject for a future cleanup, as
> that file is getting quite messy now...

I just tried to make the NSS-related changes as much non-intrusive as
possible. I am a bit new to the NTLM code and not sure if I understand most
of the defines there. I'd rather keep the more intrusive refactoring to
somebody skilled enough :-)

Kamil
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html
Received on 2010-06-29