From: Paul K. <pv...@pv...> - 2013-04-13 15:59:11
|
Stephan Frank wrote: > today I had a chance to browse the SOC2013 proposals where I stumbled > upon the bignums via GMP project. > > Are you aware of Waldek Hebisch's work for FriCAS: > http://www.math.uni.wroc.pl/~hebisch/prog/ Thank you! I was trying to remember where I had seen that. I'm adding that information to the project's description > It is already usable as a contrib though could probably be extended to > also make use of the rational functions of GMP whereas so far only the > interger bignum function are also used for rational arithmetic just as > within the current bignum implementation of SBCL. There's that, and avoiding a C shim would be nice as well. Also, I expect that using GMP for tiny multiplications isn't worth the trouble, for instance. It might also be interesting to use explicit malloc/free instead of consing temporary bignums. Plenty of interesting coding left (: Paul Khuong |