🚀 go-pugleaf

scjones@thor.sdrc.com (Larry Jones) writes:

> I suspect that what Paul meant to say was that the pairs cannot interleave.

>        VALID                   INVALID
>   -----------------        -----------------
>   va_start(a, foo);        va_start(a, foo);
>   va_start(b, foo);        va_start(b, foo);
>   va_end(b);               va_end(a);
>   va_end(a);               va_end(b);

Sorry, I was too terse there.  Here's what Gwyn's article says:

	In subclause 7.8.1, page 122, line 21, change:
		invoked
	to:
		invoked in corresponding pairs

	Append to subclause 7.8.1.1, page 122, line 29:
		`va_start' shall not be invoked again for the same `ap' without
		an intervening invocation of `va_end' for the same `ap'.

So assuming a != b, both columns quoted above are conforming.
But the following is not conforming:

	va_start(a, foo);
	va_start(a, foo);
	va_end(a);
	va_end(a);

and that is what I meant when I wrote ``the pairs cannot nest''.