Re: contrib/intarray vs empty arrays

>>>>> "Tom" == Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> writes:

[empty arrays and containment ops]

Tom> From what I understand of GIN's internal workings, this is
Tom> unfixable because there is nothing to make an index entry on
Tom> when looking at an empty array. Unless you know of a trick to
Tom> get around that, we've got a problem here.

Umm. In theory, could the extract function return some distinct
value when applied to an empty array, and the extract_query function
include the same value when extracting a query?

It's not very clean, since it means that the element type for an
index of sometype[] is no longer sometype, but it would allow the
semantics to be kept consistent.

[whether it is worth the pain of doing this is a separate question]

--
Andrew (irc:RhodiumToad)