#784 dot-operator precedence 110 ?

[Volker van Nek]

-------------------------------------------------------------

Maxima version: 5.9.1
Maxima build date: 7:34 9/24/2004
host type: i686-pc-mingw32
lisp-implementation-type: Kyoto Common Lisp
lisp-implementation-version: GCL 2.6.5

-------------------------------------------------------------
(C54) a*[v1,v2] . b*[w1,w2];
(D54) [a*w1, a*w2]*([v1, v2] . b)
(C55) :lisp $C54
((MTIMES) $a ((MNCTIMES) ((MLIST) $v1 $v2) $b) ((MLIST)
$w1 $w2))
(C55) :lisp(describe 'mnctimes)
MNCTIMES - internal symbol in MAXIMA package
...
property RBP:
109
property LBP:
110
...
(C55) :lisp(describe 'mtimes)
MTIMES - internal symbol in MAXIMA package
...
property RBP:
120
property LBP:
120
...

I regard this information as the operator precedence,
or am I wrong?
If this is the precedence, D54 should be the result of
the following D55.

(C55) (a*[v1,v2]) . (b*[w1,w2]);
(D55) a*b*v2*w2 + a*b*v1*w1

Can someone explain, if the rbp and lbp values are the
operator precedence values and what the correct value
for dot is.

Volker van Nek

[Robert Dodier]

Logged In: YES
user_id=501686

LBP and RBP are indeed operator precedence values (see
"Syntax" under "Expressions",
http://maxima.sf.net/docs/manual/en/maxima_toc.html\).
However it is the LBP and RBP of the symbols $. and $* which
are relevant to parsing input expressions; the parser sees
only the atoms in the input stream, not MNCTIMES or MTIMES.

LBP and RBP for input symbols are defined in
src/nparse.lisp, which establishes () that "." has higher
precedence than "*". However, for the purposes of display
(essentially to determine which subexpressions need
parentheses), src/grind.lisp establishes precedence for
operators MNCTIMES, MTIMES, etc. The list in src/grind.lisp
is out-of-date with respect to src/nparse.lisp -- in
particular, "." has lower precedence than "*" there.

The list in src/grind.lisp should be changed to have the
same precedence values as in src/nparse.lisp.

Whether "." should have higher or lower or the same
precedence as "*", I don't know, and it is not immediately
relevant, so I'm going to leave that alone.

[Robert Dodier]

Logged In: YES
user_id=501686

For the record -- pasting this here so that it doesn't get
lost --

(%i1) :lisp (do-symbols (s "MAXIMA") (if (get s 'mheader)
(format t "~S ~S ~S ~S ~S ~S~%" (get
(car (get s 'mheader)) 'lbp) (get (car (get s 'mheader))
'rbp) (get s 'mheader) s (get s 'rbp) (get s 'lbp))))

LBP RBP MHEADER OP RBP LBP

160 NIL (MFACTORIAL) $! NIL 160
80 80 (MNOTEQUAL) $# 80 80
NIL NIL (NODISPLAYINPUT) $$ NIL -1
NIL 201 (MQUOTE) |$'| NIL NIL
NIL NIL (MPROGN) |$(| NIL 200
120 120 (MTIMES) $* NIL 120
100 100 (MPLUS) $+ 100 100
NIL NIL ($EV) |$,| NIL 10
100 100 (MMINUS) $- 134 100
110 109 (MNCTIMES) $. 129 130
120 121 (MQUOTIENT) $/ 120 120
NIL 20 (MSETQ) |$:| 20 180
NIL NIL (DISPLAYINPUT) |$;| NIL -1
80 80 (MLESSP) $< 80 80
80 80 (MEQUAL) $= 80 80
80 80 (MGREATERP) $> 80 80
NIL NIL (MLIST) $[ NIL 200
140 139 (MEXPT) $^ 139 140
180 20 (MDEFMACRO) |$::=| 20 180
60 60 (MAND) $AND NIL 65
NIL NIL ($GENFACT) $!! NIL 160
140 139 (MEXPT) $** 139 140
NIL 70 (MNOT) $NOT 70 NIL
180 20 (MSET) |$::| 20 180
180 20 (MDEFINE) |$:=| 20 180
80 80 (MLEQP) $<= 80 80
80 80 (MGEQP) $>= 80 80
30 30 (MDO) $DO 25 25
25 25 (MCOND) $IF 45 NIL
50 50 (MOR) $OR NIL 60
135 134 (MNCEXPT) $^^ 139 140
NIL

[Robert Dodier]

Logged In: YES
user_id=501686

I've committed changes to src/grind.lisp (r1.15) which
modify the binding powers for MNCTIMES and other operators
to make them the same as for the corresponding lexical
tokens, so that the parser and the display code have the
same ideas about operator precedence.

The binding powers of lexical tokens was not changed, so
expressions are still parsed in the same way as before; only
display properties were changed.

Closing this bug report as fixed.