Amit Kapila <amit.kapil...@gmail.com> writes: > To avoid these confusions, we have disallowed adding a table if its > schema is already part of the publication and vice-versa.
Really? Is there logic in ALTER TABLE SET SCHEMA that rejects the command dependent on the contents of the publication tables? If so, are there locks taken in both ALTER TABLE SET SCHEMA and the publication-modifying commands that are sufficient to prevent race conditions in such changes? This position sounds quite untenable from here, even if I found your arguments-in-support convincing, which I don't really. ISTM the rule should be along the lines of "table S.T should be published either if schema S is published or S.T itself is". There's no obvious need to interconnect the two conditions. regards, tom lane