Re: Remove lossy-operator RECHECK flag?

Teodor Sigaev <teodor(at)sigaev(dot)ru> writes:
> Now we choose - save compatibility or not.

> We can save flag RECHECK and introduce optional needRecheck argument for
> consistent function and new opclass can use new interface, old ones will work
> with RECHECK. Or we remove RECHECK and force opclasses to use new interface.

Yeah, that's what it boils down to.

I'm leaning towards removing RECHECK because it'll allow simplification
of the core code, and I doubt there are enough outside opclasses that're
using lossy operators for the compatibility loss to be a big deal.
We've certainly forced bigger changes than that in the past.

I seem to recall that you had some plans for other incompatible changes
in the call conventions for GIST/GIN support functions, too. If
anything like that is going to happen for 8.4, then outside opclasses
are going to need updates anyway, and forcing this one on them too would
hardly be much of a burden.

regards, tom lane