[Botan-devel] Future directions, and RFC on use of Python

Jack Lloyd lloyd at randombit.net
Sat Feb 4 22:08:23 EST 2006


As of 1.5.5, I've pretty much gotten all the performance out of the public key
algorithms that _I_ can find (aside from writing all of mp_asm.cpp in
assembly), so I'm moving on to some cleanups I've been wanting to do:

 - The testsuite. I've realized this code hasn't really been touched since
   pre-1.0, and it could be a lot better. Probably won't really be any user
   visible chances, except being a bit more functional.

 - The configure script is not very nice. I'm going to rewrite it. The major
   question in my mind is if it should be in Perl again, or in something else
   (Python being the most likely choice in my mind). This is rather user
   visible, in that people will have to have Python installed - it looks like
   Python runs on anything Botan might be used on (Win32, Unix, BeOS, and
   OpenVMS), so this doesn't seem terribly onerous.

 - I'm also considering using Python as a pre-processor, at least on the
   assembly files. This will mean that not using Python to build Botan (eg, by
   using a VC++ project file instead) will result in a reduction of
   functionality. This is a seperate issue from the Python rewrite, really, and
   something I was thinking about in the context of the current Perl build
   script. I've been trying to clean up the assembler code; with some effort
   with the C preprocessor I managed to get rid of most of the repeated code,
   but I found myself wishing I had something more powerful than cpp. And I
   don't want to bring M4 into this, so the obvious choice is whatever language
   the rest of the scripts are written in.

Comments?

-Jack



More information about the botan-devel mailing list