Here's the latest.
@hahnjo I made the requested changes
@old wrote in guile/guile#183 (comment):
I might suggest that one should add some big integer tests to benchmark-suite/benchmarks/arithmetic.bm and run the benchmark with and without mini-gmp and make sure that mini-gmp isn't much slower.
The hard requirements are that
Yeah. In a reply I made to one of your other comments, I was wondering if there is value in trying to disambiguate between the three integer sizes: fixnum, mp_limb_t and uintptr_t. Or to just make...
Probably a good idea.
I've been working on a new version of that commit which I'll present as its own PR. I won't change that in the new PR.
I've been working on a new version of that commit which I'll present as its own PR. I won't change that in the new PR.
Yeah. It is extremely likely that SCM_I_FIXNUM_BIT will, for all future versions of Guile, be either 30 bits or 62 bits. But that does not mean that this condition is incorrect. I was torn between...