On Wednesday, October 30, 2013 20:41:55 Steve Holme wrote:
> Cheer Kamil,
>
> On Wed, 30 Oct 2013, Kamil Dudka wrote:
> > > Oops, my previous patch did not handle the CR character, so NTLM was
> > > still broken. The attached patch should fix it!
> >
> > One more follow-up attached.
>
> Here I am debugging my conversation with my SMTP Server and I for the life
> of me can't reproduce the problem :(
>
> Obviously I've got the proper padding when taking in real data ;-)
>
> Ooops - thanks for that and pushed.
>
> I'm pretty sure I copied the NTLM data from one of the other tests (not sure
> which) - does that mean they are also missing the padding

The length of base64-encoded strings have to be a dividable by 4 as required
by the checks you added to the code of Curl_base64_decode(). The original
strings in server's response were 214 bytes long, so I appended 2 bytes of
padding to get them working.

> or do we still need to strip of the CRLF in HTTP as well?

My other patch was not about stripping CR/LF, but about not counting CR/LF
into length for the assumptions to become satisfied.

Kamil
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html
Received on 2013-10-30