Re: [HACKERS] WIP: Faster Expression Processing v4

Andres Freund <andres@anarazel.de> writes:
> On 2017-03-20 16:06:27 -0400, Tom Lane wrote:
>> ... is there a reason why resultnum for EEOP_ASSIGN_* steps is declared
>> size_t and not just int?  Since it's an array index, and one that
>> certainly can't be bigger than AttrNumber, that seems rather confusing.

> Not that I can see, no.  I guess I might have "overcompensated" when
> changing it from AttrNumber - AttrNumber isn't a good idea because that
> needs an extra move-zero-extend, because 16bit indexing isn't that well
> supported on x86.  But that doesn't mean it should be a 64bit number -
> to the contrary actually.

OK, will fix in the edits I'm working on.
        regards, tom lane