>>>>> "David" == David Billinghurst <Billinghurst> writes:
David> Ray,
David> A recent change to sin.lisp has changed some trig integrals. The change
David> log indicates that you didn't expect any user visible changes.
David> ----------------------------
David> revision 1.12
David> date: 2006/02/23 14:14:08; author: rtoy; state: Exp; lines: +200 -159
David> o POWERL, NOTSAME, YY, BL, YZ are only used in this file, so change
David> the name to use the *foo* convention for special variables, and
David> remove them from the list of globally special variables.
David> o Declare them special in the routines that need them.
David> o Reindent INTEGRATOR, RAT8, SUPERTRIG, and RATS to make them easier
David> to read.
David> ----------------------------
[snip]
David> (%i1) integrate(-a*sin(2*x)/(a*sin(x)^2+b),x);
David> 2 2 2 2 2
David> (%o1) - log((a sin (4 x) + (- 8 a b - 4 a ) sin(2 x) sin(4 x) + a cos (4 x)
David> 2 2
David> + ((- 8 a b - 4 a ) cos(2 x) + 2 a ) cos(4 x)
David> 2 2 2 2 2 2
David> + (16 b + 16 a b + 4 a ) sin (2 x) + (16 b + 16 a b + 4 a ) cos (2 x)
David> 2 2 2
David> + (- 8 a b - 4 a ) cos(2 x) + a )/a )/2
David> (%i2) integrate(sin(2*x)/sec(x),x);
David> cos(3 x) + 3 cos(x)
David> (%o2) - -------------------
David> 6
Ouch. So everything is ok with 1.11, but is different with 1.12?
Something else must have gone wrong. I did a grep for all of those
variables (and it's B1, not BL), and they didn't occur anywhere else.
Except, I see that yy is used in risch. Perhaps my reindentation was
also more than just reindentation.
Thanks for looking into this. We need to add those two integrals to
the testsuite. :-)
Ray