[plt-scheme] inferred names of procedures must be symbols

I finally fixed something that had been bothering me for a while in my
code, and I thought I'd post it here in case anyone else has had a
similar problem.  I don't know if this should be regarded as a bug in
MzScheme or a documentation bug, so I'm not going to submit a bug
report for now.

You can set the inferred name of a procedure expression to a symbol,
so that when you print the procedure you get a name instead of a
syntax location.  This is useful in syntax transformers that create
procedure expressions:

  (define-values (struct:ap make-annotated-proc annotated-proc? ap-ref ap-set!) 
    (make-struct-type 'annotated-proc #f 2 0 #f null #f 0)) 
  (define (proc-annotation p) (ap-ref p 1))

  (define-syntax (annotated-lambda stx)
    (syntax-case stx ()
      ((_ formals annotation body1 body2 ...)
       #`(make-annotated-proc
	  #,(syntax-property #'(lambda formals body1 body2 ...)
			     'inferred-name (syntax-local-name))
	  annotation))))

The bug here is that `syntax-local-name' returns a syntax object, not
a symbol, and if you set the inferred name of a procedure to anything
but a symbol, it silently ignores it (and constructs a name from the
syntax location instead):

  > (define plus1 (annotated-lambda (x) "adds 1 to its argument" (+ x 1)))
  > plus1
  #<procedure:...nnotated-proc.ss:13:11>

The fix was to replace `(syntax-local-name)' with
`(syntax-e (syntax-local-name))':

  > plus1
  #<procedure:plus1>

I think the documentation ought to point out that the inferred name of
a procedure expression must be a symbol, or else it should just be
smart about converting an identifier to a symbol too.

--dougo at place.org