OpenSSL digest

[Milan Tomic <mi...@setcce.org>]

Berin,

While examining OpenSSLCryptoHash.cpp, I have noticed that destructor is
empty and that call to EVP_MD_CTX_cleanup() is missing in it, as it is
shown in this example:

http://www.die.net/doc/linux/man/man3/evp_md_ctx_cleanup.3.html

Best regards,
Milan

RE: OpenSSL digest

[Milan Tomic <mi...@setcce.org>]

> Are you cool to add the code in the 
> destructor?

OK.

> I've been doing optimisation work 
> all over 
> the place

I'm very happy to hear this. :)

Re: OpenSSL digest

[Berin Lautenbach <be...@wingsofhermes.org>]

Hmm - you're right.  Not good.  Are you cool to add the code in the 
destructor?

I'd fix it - but my checked out copy is broken at the moment, so I 
wouldn't be able to test.  (I've been doing optimisation work all over 
the place - I found this site ( http://xmlbench.sourceforge.net/ ) that 
benchmarks various signature implementation.  I never thought the C++ 
library was quick, but I hadn't realised it was *that* slow.  So I've 
been doing all the performance improvements I've meant to do for a long 
time.

Most are done - with significant speed increases.  I'll document them in 
an e-mail soon.  I have one more that I'm working on, which is to 
canonicalise in UTF-16 and then transcode the results to UTF-8. 
Currently the code transcodes each separate part (e.g. node name, then 
node value) to UTF-8 separately, and then outputs.  That's putting a 
huge load on the transcoder as it gets called thousands of times in a 
big document for tiny bits of data, when we could do it once or twice 
with large blocks.

Cheers,
	Berin

Milan Tomic wrote:

> Berin,
> 
> While examining OpenSSLCryptoHash.cpp, I have noticed that destructor is
> empty and that call to EVP_MD_CTX_cleanup() is missing in it, as it is
> shown in this example:
> 
> http://www.die.net/doc/linux/man/man3/evp_md_ctx_cleanup.3.html
> 
> Best regards,
> Milan
> 
> 
>