Am 30.01.2017 um 23:21 schrieb Brandon Williams:

On 01/30, René Scharfe wrote:

quoted

Am 30.01.2017 um 22:03 schrieb Johannes Schindelin:

quoted

It is curious, though, that an
expression like "sizeof(a++)" would not be rejected.

Clang normally warns about something like this ("warning: expression
with side effects has no effect in an unevaluated context
[-Wunevaluated-expression]"), but not if the code is part of a
macro.  I don't know if that's intended, but it sure is helpful in
the case of SWAP.

quoted

Further, what would SWAP(a++, b) do? Swap a and b, and *then* increment a?

That might be a valid expectation, but GCC says "error: lvalue
required as unary '&' operand" and clang puts it "error: cannot take
the address of an rvalue of type".

René

Perhaps we could disallow a side-effect operator in the macro.  By
disallow I mean place a comment at the definition to the macro and
hopefully catch something like that in code-review.  We have the same
issue with the `ALLOC_GROW()` macro.

SWAP(a++, ...) is caught by the compiler, SWAP(*a++, ...) works fine. 
Technically that should be enough. :)  A comment wouldn't hurt, of course.

René