🚀 go-pugleaf

I was surprised to learn that constant pointer expressions are
not allowed in case labels and that the expression in a switch statement
must be an integral type.  (The reason for the exclusion of floating point
is obvious to me).  I am interested in the technical reasons for this.

Consider:

T a [] = { ... }, *b = a + sizeof a / sizeof *a, *c;

switch (c) {
	case a: ...
	case a + 1: ...
        case a + 3: ...
	default: ...
}

where T is some type.
The assignment to b has worked on all compilers I have ever tried it on
(even and especially K&R I).  I take this as a strong indication of
conformance.  The switch construct above is explicitly disallowed, although
there does not seem to be much difference in the expressions themselves.

I am of course aware that

switch (c - a) {
	case 0: ...
	case 1: ...
	case 3: ...
	default: ...
}

is equivalent code, but it suffers from the division needed to evaluate
c - a.

Konrad Schwarz