On 11/4/16 4:28 PM, Tom Lane wrote:
My proposal therefore is for SPI_fnumber to ignore (not match to) dropped columns, and to remove any caller-side attisdropped tests that thereby become redundant.
Yeah, SPI users certainly shouldn't need to worry about attisdropped, and exposing attnum doesn't seem like a good idea either.
It's possible that it'd make sense for pltcl_trigger_handler to ignore empty-string column names in the returned list, so that the behavior with stupid trigger functions would be a bit more forgiving; but that is more or less independent of this patch.
I'm a bit reluctant to do that since it'd be nice to be consistent with regular pltcl functions returning composites. The same kind of issue exists with the holes in $TG_relatts; we shouldn't be exposing the details of attnum that way. Any code that's expecting those holes is going to blow up after a dump/restore anyway.
-- Jim Nasby, Data Architect, Blue Treble Consulting, Austin TX Experts in Analytics, Data Architecture and PostgreSQL Data in Trouble? Get it in Treble! http://BlueTreble.com 855-TREBLE2 (855-873-2532) mobile: 512-569-9461 -- Sent via pgsql-hackers mailing list (email@example.com) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers