On Mon, 1 Dec 2014, Patrick Monnerat wrote:

> Please find a big patch in attachment.
 
My... That is a big patch :-/

Are you able to split it up into more manageable / bisect'able chunks?

> It factorizes most of the SASL code in IMAP, POP3 and SMTP into its
> own state machine. This reduces the authentication-handling states for
> these protocols, and shrinks the code by more than 1800 lines.

As you know I wrote quite a bit of the SASL authentication logic but - which
has grown and grown over the time I have been responsible for it.

Even so this sounds like a great idea to try and reduce the code base and
any duplication where we can - in principal I'm all for it.

> On linux 64bit, the static memory footprint is also reduced by more than
2%.
>
> I submit this patch here before pushing it to call your comments: all
tests pass in
> my linux64+openssl environment, but I would like to be sure it doesn't
break
> M$W builds, etc.

It's unfortunately late so I've not had time tonight to apply the patch and
test it with either a) Windows base OpenSSL build or b) a SSPI based build.

> Thanks for your comments.

Additionally, I'm in the middle of refactoring the authentication code at
the moment which as you've probably seen I have started with the splitting
up of NTLM SSPI code from the native messaging code but it is unfortunately
too big to get done for 7.40.0 - I've already pushed 100+ commits in
November!!
 
Not only that but I am hoping to get GSS-API support for SASL GSSAPI pushed
over the next couple of days.

As such, is this something that can wait until 7.40.0 is out the door and I
have finished off my rework / refactor which should be done early January at
the latest - I'll probably finish it before then locally and push once
7.40.0 is out?

Kind Regards

Steve
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html
Received on 2014-12-02