"Ryan Bradetich" <rbradetich@gmail.com> writes:
> ... I did have the following
> concern looking through src/backend/utils/adt/int8.c: There is code that is
> optionally compiled based on the INT64_IS_BUSTED pre-processor define.
> Is this pre-processor define something I should worry about for portability
> with this plan?
I wouldn't worry, really ;-). Five or more years ago, it seemed
important for PG to work on machines without functional int64 support,
but there is little if any evidence that anyone is using current PG
releases on such platforms. I might well be the last active PG hacker
who gives a damn about that case at all, and even I long ago stopped
expecting anything beyond core functionality to work on such a machine.
Since your proposed unsigned types certainly aren't core functionality,
I see no reason that they should need to work on INT64_IS_BUSTED
platforms.
> After I get uint types implemented, for fun I might try some benchmarks
> to see if I can detect the int8 overhead on a 32-bit system.
Right, you need to check that before drinking the kool-aid ...
regards, tom lane