Hello Tom,
> I've got doubts about this too: it introduces fundamental inconsistency
> in pgbench itself, which may come back to bite us later. Who's to say
> that '=' will never be a meaningful operator in pgbench expressions?
Hmmm... it would not be an issue as long as '=' is not a unary operator?
If you add "expr '=' expr" in the syntax, there should be no conflict with
the result target "'=' expr" anyway?
I've just tried that and there is no issue raised by bison.
So ISTM that it should not bite later on that ground.
--
Fabien.