Maybe possibly remove DEFAULT definition from the intarray initialization
SQL and eliminate in the documentation: "if you want to use GIN with _int4,
you have to specify the operator class explicitly and manually"? This at
least does not break the standard pg_dump behaviour. We checked, if we
remove DEFAULT keyword, a dump is restored correctly.
On 2/23/07, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> "Dmitry Koterov" <d@koterov.ru> writes:
> > [ pg_restore fails with ]
> > ERROR: could not make operator class "gin__int_ops" be default for type
> > pg_catalog.int4[]
> > DETAIL: Operator class "_int4_ops" already is the default.
>
> Yeah. I'd say that intarray's attempt to override the default status of
> the built-in gin opclass is simply a bad idea and should be removed.
> It's not even documented that it does that (in fact I see no mention of
> GIN at all in README.intarray :-(, so we have a documentation lack
> here too).
>
> Comments?
>
> regards, tom lane
>