Basically just had to add "/" to the path, as we already do for clisp,
sbcl, ecl, and openmcl.
Tested this by building maxima and installing it in a temp directory
and running "maxima -l gcl --list-avail". This lists all of the lisps
I configured for running maxima.
1982c57...
by
Robert Dodier <email address hidden>
Disable GCL readline symbol completion by setting SI::*READLINE-PREFIX* to a nonexistent package.
Fixes SF bug #3972: "gcl only: Autocompletion inverts command case"
This is a suboptimal bug fix. As it says in the comments,
;; This is kind of terrible. I don't see a flag to only disable completion,
;; or a way to set the symbol list to Maxima symbols and disable case inversion,
;; so set the completion prefix to a nonexistent package.
;; If ever package BLURFLE is actually defined, and contains external symbols,
;; those symbols will be completed. I can live with that.
This makes, for example, zeta(-7/2) return a result in terms of
zeta(9/2), and when you numerically evaluate it, you get the same
answer as you would get for zeta(-3.5).
This is nicer than just returning zeta(-7/2).
Add a couple of tests to rtest_zeta to verify the result.
elliptic_e(5*%pi/4,1) was returning 2-1/sqrt(2), but
elliptic_e(5*%pi/4,1),numer returned 2.707... or basically
2+1/sqrt(2).
This happens because we were simplifying elliptic_e(5*%pi/4,1) to
sin(5*%pi/4) = -1/sqrt(2). But we need to reduce the arg 5*%pi/4 by
%pi so we get sin(%pi/4) = 1/sqrt(2).
Add a test for this and that the symbolic and numerical values are
consistent.
Also needed to update the test for elliptic_e(x,1) to include the
reduction of the arg for the sin part.
I tweaked it slightly in the case of computing the eigenvectors so
that we print out the eigenvalues and the eigenvectors separately so
that we can see the column vectors more clearly.
Just add some docstrings to make it easier to understand what the
function does without having to read the code. Just to lower to
burden of understanding code.