Maxima -- GPL CAS based on DOE-MACSYMA

Yes, your are right. On my system (CLISP and Linux) the example of this bug
report returns after about 1 hour without an error:

(%i1) plot2d(integrate(signum(x),x,-2,y),[y,-1,2]);
plot2d: expression evaluates to non-numeric value somewhere in plotting
range.

I think we can close this bug report. Maxima returns a noun form for the
integral. That is not a bug. We have a package which does the integration.
Unfortunately, Maxima needs a long time to return from plot2d and perhaps
runs out of memory, but the user can not expect to get a plot for noun
form.

Setting the status to pending and the resolution to "Wont fix".

Dieter Kaiser

It's not hung; it's just slow. With the adaptive plotter, maxima is trying
very hard to refine the graph because every calculated point is not a
number which makes maxima think some kind of discontinuity exists. I think
this is a deficiency caused by gcl not quite supporting the needed feature
to make it stop early.

Perhaps the level to which the adaptive plotter divides intervals should
be made smaller so fewer subdivisions are tried?

First, I think it is not a problem of the integration of signum. Maxima
does not find the integral and returns a noun form (without abs_integrate).
That is O.K., because we have the abs_integrate package.

Here another example:

(%i28) integrate(sin(x)*expintegral_ei(x),x,-2,y);
(%o28) 'integrate(expintegral_ei(x)*sin(x),x,-2,y)
(%i29) plot2d(%,[y,-1,2]);

At this point Maxima hangs on my system (Linux and CLISP). I have tried
several noun forms of integrals of this type and I have got the problem
that plot2d does not return.

Dieter Kaiser

Workaround

(%i1) load(abs_integrate)$
(%i2) plot2d(integrate(signum(x),x,-2,y),[y,-1,2]);