I wrote:
>
> Attached is my rework of David Fetter's array of composites patch. It
> has all the agreed modifications and checks, except altering the name
> mangling.
>
>
Here is the latest with the name mangling piece done - it's got quite
large with the catalog rewrite :-)
It could do with some extra eyeballs - especially the error codes.
I'm not sure if we need to adjust the test in pg_dump.c - maybe we
should look at using the same test as the backend (typlen == -1) to
disambiguate real arrays from subscriptable base types.
cheers
andrew