Marko Kreen <marko(at)l-t(dot)ee> writes:
> This is unary plus. One time somebody complained about it,
> then some time later thought about it and implemented it.
I'd vote against the gram.y part of this change (doUnaryPlus),
but the rest looks OK.
The reason we have doNegate is mainly so that MININT can be represented
without overflow (assuming that atoi gets it right, of course). There's
no corresponding issue with unary plus, and no good reason to hard-wire
assumptions about the operator's semantics into gram.y.
regards, tom lane