(subtypep 'number '(or complex real)) ;=> NIL T

[phoe]

(subtypep 'number '(or complex real)) ;=> NIL T. This implies that CCL is sure that there is a subtype of number which is neither complex nor real. Either this type exists, and I'm completely puzzled by its existence, or CCL's subtypep has a bug (just like SBCL's).

[Symbolics]

That's a great question. A shame Gary Byers isn't around any longer...

[xrme]

The CLHS ssue REAL-NUMBER-TYPE http://clhs.lisp.se/Issues/iss290_w.htm seems to dance around this.

Also note:

? (subtypep 'real '(or rational float))
NIL
T

[informatimago]

On the other hand, this leaves the door open for extensions such as realib (real reals), or quaternion, octonions, etc.