From: hfinney@shell.portal.com (Hal Finney)
One thing that frustrates me is the difficulty of easily providing implementations of cryptographic algorithms that would be useful on a wide range of machines. [...]
The point I'm getting to is that it would be nice if all this were done ONCE, and a library made and tested which would work on a wide range of machines.
Well, I do know of someone who is working on a cryptolib package that will attempt to include a ton of crypto methods into a single library that anyone can link to. From email exchanged with this person it seems that he is looking to get a unix version up and then let people port it around. The math lib stuff is the most recent sticking point he was having: there arenot many fast multi-precision math packages out there that are free (and gmp does not cut it, he wants to be able to let anyone use the code however they want...) So far he has checked fgmp and bignum, but if anyone knows of a fast package that has a berkelyish copyright let me know and I will pass it on...
Let me know if you would be interested in participating in this effort. Hopefully a lot of the pieces already exist and it will just be a matter of pulling them together.
Maybe I can set up a list for this if people are interested. The existence of such a beast, even in a rudimentary form, would be useful to quite a few people I would bet. I will push him a bit to see if he will dump what he has now into a package so that others can help out. jim