#264 GCL / polymod bug, was: mod(-2,4) => -2 not 2

[Stavros Macrakis]

mod(-2,4) => -2

It should be 2 to be consistent with mod(2,4), mod(6,4), etc. Otherwise there are FIVE distinct values for mod(x,4) instead of four.

Maxima 5.9.0 GCL 2.5.0 mingw Windows 2000

[Wolfgang Jenkner]

Logged In: YES
user_id=581700

For both Clisp and SBCL I get

(C1) makelist(mod(i,4),i,-4,-1);
(D1) [0, 1, 2, - 1]

Wolfgang

[Wolfgang Jenkner]

Logged In: YES
user_id=581700

GCL imports some symbols for modular arithmetic from the SYSTEM
package (see maxima-package.lisp). In particular,

>(lisp-implementation-version)
"GCL-2-5.2000000000000002"

>(let ((si:modulus 4)) (si:cmod 2))
2

>(let ((si:modulus 4)) (si:cmod -2))
-2

The other lisps define this in rat3a.lisp

* (let ((modulus 4)) (cmod 2))
2
* (let ((modulus 4)) (cmod -2))
2

[Robert Dodier]

Logged In: YES
user_id=501686

Post-5.9.2, mod was renamed to polymod and nummod was
renamed to mod. Relabeling this report accordingly.

[Robert Dodier]

Logged In: YES
user_id=501686

Per comment below this is a GCL-specific bug. Relabeling
accordingly.

Proposed solution: For GCL, instead of importing SI:CMOD,
implement Maxima CMOD in rat3a in terms of SI:CMOD. That way
we can get the presumed speed-up as well as consistency with
other Lisp implementations.

Incidentally Maxima 5.9.3cvs / gcl 2.6.7 yields:

(%i3) makelist(polymod(i,4),i,-8,8);
(%o3) [0, 1, - 2, - 1, 0, 1, - 2, - 1, 0, 1, 2, - 1, 0, 1,
2, - 1, 0]

So there are indeed just 4 distinct values, but a different
4 than sbcl or clisp. E.g. sbcl =>

(%i1) makelist(polymod(i,4),i,-8,8);
(%o1) [0, 1, 2, - 1, 0, 1, 2, - 1, 0, 1, 2, - 1, 0, 1, 2, -
1, 0]

[Dan Gildea]

Seems to be fixed now.

Maxima branch_5_38_base_351_ge2a6f50 http://maxima.sourceforge.net
using Lisp GNU Common Lisp (GCL) GCL 2.6.12
(%i5)  makelist(polymod(i,4),i,-8,8);
(%o5) [0,1,2,-1,0,1,2,-1,0,1,2,-1,0,1,2,-1,0]

[Robert Dodier]

Marking this "maybe fixed", I think I'd like to review this and verify the status of it.