Re: More ideas for speeding up sorting

Peter Geoghegan <pg(at)heroku(dot)com> writes:
> I know that in practice, even virtual address space is significantly
> less than 64-bits on x86_64, so I think that there should be several
> bits there for the taking, even if the addresses are not aligned
> (although, they are). However, I have no idea if there is any
> precedent for this general idea at all, and would feel better about it
> if there was. I guess we don't have to make it any more complicated
> than your argument about alignment.

There's a very long history of code stealing bits out of pointers, and
all of it is bad and unportable in the long run :-(. Let's not go there.
I thought the idea of getting rid of isnull by physically segregating the
null items sounded good though.

regards, tom lane