In message <[EMAIL PROTECTED]> on Mon, 17 May 2004 00:44:53 +0200, Andy Polyakov 
<[EMAIL PROTECTED]> said:

appro> > appro> > appro>   Log:
appro> > appro> > appro>     size_t-fication of message digest APIs. We should 
size_t-fy more
appro> > appro> > appro>     APIs...
appro> > 
appro> > Well, I think I've worked in a different way than you.
appro> 
appro> I simply had different "excuse," introduction of fresh code,
appro> SHA-512 and co:-)

OBTW, you should add license texts in all the files you committed.

appro> > I started with
appro> > the memory allocation functions and let the effects spread.  I've also
appro> > taken a look at some header files and adjusted what needed to be
appro> > adjusted, then modified all C files where the compiler complained.
appro> > This work is still going on...
appro>                 ^^^^^ That's the problem with this approach. It
appro> gets so big, you start feeling insecure, don't you?

Not at all.  Think about it a little bit, if I change one cipher, I
have to change them all.  Oh, and I have to change the macros
BLOCK_CIPHER_func_ofb, BLOCK_CIPHER_func_cbc and
BLOCK_CIPHER_func_cfb, because they're casting inl to long (I removed
the cast), meaning you need to change the EVP layer functions, and why
should one stop there and introduce some temporary casts?  As soon as
the EVP layer is affected, it ramifies to the rest of OpenSSL.  Note
that we *need* to make sure there are no unsigned vs. signed clashes,
or we will have the Windows horde on our backs more or less
immediately.

appro> Not that is's wrong approach, it just takes more guts to attack
appro> the problem...

AND time.  The only reason that I haven't done more (and as you could
see, there's a lot done) is that I've had other priorities.  The
size_t stuff is something I've done in between other things.

Also, there are a few cases when I'm not entirely sure that converting
to size_t is the right thing...

appro> Yet I'd prefer to see changes applied in "zones," say ciphers,
appro> bio, asn1, etc., not as monolith mumbo-jumbo.

You do have a point, and sure, I can commit things that I've finished,
no problems.  The ciphers with the EVP-level remifications could be a
good start, for example...

appro> It would be easier to grasp and audit. And I don't think
appro> everything has to be size_t-fied. In some situations it might
appro> be easier and in a sense more appropriate to leave API
appro> unchanged.
appro> BN and PKI stuff for example. In a cipher case there is a
appro> *remote* possibility that someone would like to take full
appro> advantage of size_t.  But in BN or PKI case on the other it
appro> would/should be considered as a DoS attack. A.

There's gonna be quite a lot of new casts with that approach...

-----
Please consider sponsoring my work on free software.
See http://www.free.lp.se/sponsoring.html for details.

-- 
Richard Levitte   \ Tunnlandsv�gen 52 \ [EMAIL PROTECTED]
[EMAIL PROTECTED]  \ S-168 36  BROMMA  \ T: +46-708-26 53 44
                    \      SWEDEN       \
Procurator Odiosus Ex Infernis                -- [EMAIL PROTECTED]
Member of the OpenSSL development team: http://www.openssl.org/

Unsolicited commercial email is subject to an archival fee of $400.
See <http://www.stacken.kth.se/~levitte/mail/> for more info.
______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       [EMAIL PROTECTED]
Automated List Manager                           [EMAIL PROTECTED]

Reply via email to