On 23.03.24 18:42, Paul Jungwirth wrote:
> Now this is a long chain of reasoning to say rangetypes are safe. I
> added a comment. Note it doesn't apply to arbitrary types, so if we
> support those eventually we should just require a recheck always, or
> alternately use equals, not containedby. (That would require storing
> equals somewhere. It could go in ffeqop, but that feels like a footgun
> since pfeqop and ppeqop need overlaps.)
Ok, this explanation is good enough for now. I have committed the
patches v33-0001-Add-temporal-FOREIGN-KEYs.patch and
v33-0002-Support-multiranges-in-temporal-FKs.patch (together).
