Hi Jesper On 2019-Mar-01, Jesper Pedersen wrote:
> I'm getting a failure in the pg_upgrade test: > > -- > +-- Name: pk5 pk5_pkey; Type: CONSTRAINT; Schema: regress_fk; Owner: > jpedersen > +-- > + > +ALTER TABLE ONLY regress_fk.pk5 > + ADD CONSTRAINT pk5_pkey PRIMARY KEY (a); > + > + > +-- > > when running check-world. So I tested this case just now (after fixing a couple of bugs) and see a slightly different problem now: those excess lines are actually just that pg_dump chose to print the constraints in different order, as there are identical lines with "-" in the diff I get. The databases actually *are* identical as far as I can tell. I haven't yet figured out how to fix this; of course, the easiest solution is to just drop the regress_fk schema at the end of the foreign_key.sql regression test, but I would prefer a different solution. I'm going to post the new version of the patch separately. -- Álvaro Herrera https://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
--- /pgsql/build/part5/src/bin/pg_upgrade/tmp_check/dump1.sql 2019-03-18 17:28:00.201071102 -0300 +++ /pgsql/build/part5/src/bin/pg_upgrade/tmp_check/dump2.sql 2019-03-18 17:28:16.915627217 -0300 @@ -367111,6 +367111,14 @@ -- +-- Name: pk5 pk5_pkey; Type: CONSTRAINT; Schema: regress_fk; Owner: alvherre +-- + +ALTER TABLE ONLY regress_fk.pk5 + ADD CONSTRAINT pk5_pkey PRIMARY KEY (a); + + +-- -- Name: pk51 pk51_pkey; Type: CONSTRAINT; Schema: regress_fk; Owner: alvherre -- @@ -367127,14 +367135,6 @@ -- --- Name: pk5 pk5_pkey; Type: CONSTRAINT; Schema: regress_fk; Owner: alvherre --- - -ALTER TABLE ONLY regress_fk.pk5 - ADD CONSTRAINT pk5_pkey PRIMARY KEY (a); - - --- -- Name: at_partitioned_a_idx; Type: INDEX; Schema: public; Owner: alvherre --