documentation needs help

We read:

"Equivalent to write_term/2 using the options ignore_ops, quoted and numbervars after numbervars/4 using the singletons option."

This should be reformulated to:

"Equivalent to transforming Term with numbervars/4 using the singletons option, followed by write_term/2 using the options ignore_ops, quoted and numbervars.

As in:

?- Term = f(X,Y,X),
   write_canonical(Term).
f(A,_,A)
Term = f($VAR(X),$VAR(Y),$VAR(X)).

Same as (note that variables in Term are unified with $VAR/N terms):

?- Term = f(X,Y,X),
   numbervars(Term, 0, End, [singletons(true)]),
   write_term(Term, [ignore_ops(true),quoted(true),numbervars(true)]).
f(A,_,A)
Term = f($VAR(0),$VAR(_),$VAR(0)),
X = $VAR(0),
Y = $VAR(_),
End = 1.

See also numbervars/3 for more comment.

Slight error:

"This used to be the case anyhow, as garbage collection between multiple calls to one of the write predicates can change the _G<NNN> identity of the variables."

But variables are now just printed as `_$INTEGER`.

Quite an interesting predicate to see what syntax tree the parser has created from the stuff inside the parentheses of write_canonical():

?- write_canonical( f(p(X),q(Y)) ).
f(p(_),q(_))

?- write_canonical( f((p(X),q(Y))) ).
f(','(p(_),q(_)))

?- write_canonical( (a,b,c) ).
','(a,','(b,c))

?- write_canonical(\+(p(X),q(Y))).
\+(p(_),q(_))

?- write_canonical(\+((p(X),q(Y)))).
\+(','(p(_),q(_)))

Interesting:

?- write_canonical(\+ ((p(X),q(Y)))).
\+(','(p(_),q(_)))

List are written out in the special "way of the list":

?- write_canonical( [x,y,z] ).
[x,y,z]

?- write_canonical( [x,y|Z] ).
[x,y|_]

?- write_canonical( '[|]'(x,y) ).
[x|y]

?- write_canonical( '[|]'(x,Z) ).
[x|_]

?- write_canonical( '[|]'(x,'[|]'(y,K)) ).
[x,y|_]

?- write_canonical( '[|]'(x,'[|]'(y,[])) ).
[x,y]

This is not a list:

?- write_canonical( '[|]'(x,'[|]'(y,k)) ).
[x,y|k]