For what it is worth:
In /share/contrib/altsimp there is a file altsimp.lisp that has some of my
ideas about an
infinity correct simplus function. There are some source code comments that
might be useful;
After loading altsimp, the testsuite gives bugs for
rtest15.mac, problem (190 212)
rtest16.mac, problem: (77 134 186 190 218 221 224)
rtestsum.mac, problem: (141 142 171 226)
rtest_taylor.mac, problem: (34 36 63)
rtest_gamma.mac, problem: (504 505 506 507 508 509)
rtest_expintegral.mac, problem: (148 149)
It runs the testsuite in about 1388.235 seconds, compared to about 488
seconds without
altsimp. That's humbling. There are sums for which altsimp is much faster,
I think.
Barton