>Well, hell.  According to K&R's table of evaluation order, things get
>evaluated from right-to-left with assignment operators last in line; i.e.
>things happen to the variables, then the assignment ops go through.

Ahh, but that's not evaluation order.  That's *parsing* order.

This summarizes the whole issue.

Compilers may reorder operations, and, because we can never have to writes
to one object, or a read of an object *and* a write to one object, between
sequence points, they can do so quite a lot, and safely.

>[Hmm, I suppose that if the exact address is not in scope it would be
>considerably difficult.  Okay, I concede...]

An interpreter could do it.  My C implementation (to be written as soon
as anyone contributes the $80+k I'd need to live on for the time it would
take to write it) will detect it.

>Hmmm.  Perhaps the compilers should define the undefined behaviour to
>start up "/usr/games/rogue".

I assume you're aware of the historical precedent?

-s