I've been working to make our test suite less dependent on ratsimp
and evaluation of the expected output. I'm almost done
making the needed changes to the test suite. I've tried to be careful;
if I've made a mistake, let me know and I'll fix it. Next week, I
plan to replace the code (in mload) that does the checking for the
test suite. Along the way, I've found a few (not many) glitches such as:
(%i1) e : sum (lambda ([i], i^2), i, 1, n);
(%o1) n*lambda([i],i^2)
(%i2) ?print([e, expand(e,0,0)]);
((MLIST SIMP)
((MTIMES SIMP) $N ((LAMBDA SIMP) ((MLIST) $I) ((MEXPT) $I 2)))
((MTIMES SIMP) ((LAMBDA SIMP) ((MLIST) $I) ((MEXPT) $I 2)) $N))
(%o2) [n*lambda([i],i^2),lambda([i],i^2)*n]
The expressions e and expand(e,0,0) don't print the same.
Byt the way, we could modify fpprintprec so that all the digits of
a double float can be shown:
(%i1) x : cos(1.0)$
(%i2) fpprintprec : 942$
(%i3) ?print(x);
0.54030230586813977
(%o3) 0.54030230586814 <-- missing a few digits
Barton