PG Bug reporting form <noreply(at)postgresql(dot)org> writes:
> The following script:
> CREATE TYPE ctype AS (i int, j int);
> CREATE TABLE ctbl(a int, cf ctype);
> CREATE UNIQUE INDEX ctbl_idx ON ctbl(cf);
> INSERT INTO ctbl VALUES (1, '(1, 2)'::ctype), (2, '(1, 1)'::ctype);
> ALTER TYPE ctype DROP ATTRIBUTE j;
> Results in the UNIQUE constraint broken:
> SELECT ctid, * FROM ctbl;
> ctid | a | cf
> -------+---+-----
> (0,1) | 1 | (1)
> (0,2) | 2 | (1)
Meh. I'm happy to classify this as "so don't do that".
The cost of having ALTER TYPE detect such situations seems
vastly out of proportion to the value (and I think it'd
be impossible to prevent race conditions in the detection
anyway).
regards, tom lane