possible bug with integration of unit_step



Here is a bug I found using Maxima CVS + SBCL 1.0.22 + XP. I know that
SBCL + XP is experimental--is this problem unique to SBCL + Windows?

$ ./maxima-local

This is experimental prerelease support for the Windows platform: use
at your own risk.  "Your Kitten of Death awaits!"
Maxima 5.16post http://maxima.sourceforge.net
Using Lisp SBCL 1.0.22

(%i1) display2d : false$

Not OK:

 (%i2) integrate(x * cos(x^2) + unit_step(x),x);
 STYLE-WARNING: redefining SIMP-UNIT-STEP in DEFUN
 STYLE-WARNING: redefining SIMP-POCHHAMMER in DEFUN

 (%o2) sin(x^2)/2
 (%i3) integrate(unit_step(x),x);

OK:

 (%o3) 'integrate(unit_step(x),x)
 (%o3) 'integrate(unit_step(x),x)

Some of you know that I have a bit of code for integrands of the form
p(x) * |q(x)|, p(x) * unit_step(q(x)), ... where q(x) is a product of
linear factors. The bug in (%o2) isn't related to my (what will be
share/contrib) code, (I didn't load my code). Using 5.17.0 + GLC + XP, I
get a
nounform for (%o2).

Barton