Andres Freund escribió:
> On 2013-08-07 10:36:52 -0400, Alvaro Herrera wrote:
> > Yeah, I guess in isolation this doesn't make that much sense. I am
> > hesitant to create a third copy in the minmax patch, but I will do that
> > for now and propose the refactoring later.
>
> Well, you didn't mention upthread that you want to do this because
> you're going to need another variant of the same code. Imo that's
> sufficient reasoning.
Well, I would need to tweak the new code again, moving it from
heaptuple.c to htup_details.h as a STATIC_IF_INLINE function; and
furthermore I have noticed that if I modify the code in minmax, I can
have it do one more thing that I would need to do outside of it anyway.
(These tuples have two null bitmaps; I would have to decode the second
one separately. Doing it inside the new "deform" variant is much
easier.)
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services