Maybe spurious warning about shadowed single-float/short-float in typecase

[rtoy]

trafficstars

Maxima has this function:

(defun extreme-float-values (x)
  (typecase x
    (short-float (values most-negative-short-float most-positive-short-float))
    (single-float (values most-negative-single-float most-positive-single-float))
    (double-float (values most-negative-double-float most-positive-double-float))
    (long-float (values most-negative-long-float most-positive-long-float))))

CCL64 (1.12.2 (v1.12.2-66-g7e9ea2a9) LinuxX8664) prints this warning when compiling this form:

;Compiler warnings for "home:src;sourceforge;maxima;src;float.lisp.newest" :
;   Clause (SINGLE-FLOAT (VALUES MOST-NEGATIVE-SINGLE-FLOAT MOST-POSITIVE-SINGLE-FLOAT)) ignored in TYPECASE form - shadowed by (SHORT-FLOAT (VALUES MOST-NEGATIVE-SHORT-FLOAT MOST-POSITIVE-SHORT-FLOAT)) .
;   Clause (LONG-FLOAT (VALUES MOST-NEGATIVE-LONG-FLOAT MOST-POSITIVE-LONG-FLOAT)) ignored in TYPECASE form - shadowed by (DOUBLE-FLOAT (VALUES MOST-NEGATIVE-DOUBLE-FLOAT MOST-POSITIVE-DOUBLE-FLOAT)) .

I can see it can be useful especially if the clauses for single-float and short-float have truly different results. In this case, the results aren't actually different since most-negative-single-float and most-negative-short-float are eq.

Other lisps (clisp, cmucl) don't warn about this.

I suppose a way to suppress this warning would be acceptable. (Need to read the manual to see if this is possible.)

[xrme]

The warning is a ccl::shadowed-typecase-clause, which is a subtype of style-warning.

https://trac.clozure.com/ccl/ticket/872

So, you could write something like the following to suppress it.

(handler-bind ((warning
                 (lambda (c)
                   (if (typep c 'ccl::shadowed-typecase-clause)
                     (muffle-warning c)))))
  (load "~/f.lisp"))

Maybe cl:style-warning instead of ccl::shadowed-typecase-clause would be preferable, to avoid the internal symbol (I don't know why it's internal, to be honest).

[rtoy]

Thanks for your suggestion. I'll look into how to use this to silence the warning.

Feel free to close this issue, if this is the intended behavior.

[xrme]

I'm going to close this as "working as intended".

The entry in the spec for typecase says

The compiler may choose to issue a warning of type style-warning if a clause will never be selected because it is completely shadowed by earlier clauses.

I hope that it is not too burdensome to muffle style-warnings in cases like this where the shadowing is known to be harmless.