Thread: [7.4beta3] pg_dump -t xxx won't output sequences
Hi, Not sure if this is a bug, but still. :) Compare: <cut> pgsql7.3.4$ pg_dump -a -t seq_wtmp db1 -- -- PostgreSQL database dump -- \connect - kala SET search_path = public, pg_catalog; -- -- TOC entry 1 (OID 800427) -- Name: seq_wtmp; Type: SEQUENCE SET; Schema: public; Owner: kala -- SELECT pg_catalog.setval ('seq_wtmp', 3928, true); <cut> pgsql7.4beta3$ pg_dump -a -t seq_wtmp db1 pg_dump: specified table "seq_wtmp" does not exist <cut> seq_wtmp is a perfectly normal sequence. Bye, -- Tomas Szepe <szepe@pinerecords.com>
Tomas Szepe <szepe@pinerecords.com> writes: > Not sure if this is a bug, but still. :) > pgsql7.3.4$ pg_dump -a -t seq_wtmp db1 > [ works ] > pgsql7.4beta3$ pg_dump -a -t seq_wtmp db1 > pg_dump: specified table "seq_wtmp" does not exist Looking at the CVS history, I see that Neil added code to deliberately reject this case, but I don't understand why. /* * If the user is attempting to dump a specific table, check to ensure * that the specified table actually exists (and is a table or a view, * not a sequence). */ if (selectTableName) { for (i = 0; i < ntups; i++) if (strcmp(tblinfo[i].relname, selectTableName) == 0 && tblinfo[i].relkind != 'S') break; Neil, what's the reason not to allow dumping just a sequence? regards, tom lane
On Sat, 2003-09-27 at 14:23, Tom Lane wrote: > /* > * If the user is attempting to dump a specific table, check to ensure > * that the specified table actually exists (and is a table or a view, > * not a sequence). > */ > if (selectTableName) > { > for (i = 0; i < ntups; i++) > if (strcmp(tblinfo[i].relname, selectTableName) == 0 && > tblinfo[i].relkind != 'S') > break; > > Neil, what's the reason not to allow dumping just a sequence? Good question. I can't see a good reason not to allow it, and I can't recall my original reason for including that code. Please go ahead and revert it (or let me know and I'll submit a patch myself). Sorry about that. -Neil
Neil Conway <neilc@samurai.com> writes: > On Sat, 2003-09-27 at 14:23, Tom Lane wrote: >> Neil, what's the reason not to allow dumping just a sequence? > Good question. I can't see a good reason not to allow it, and I can't > recall my original reason for including that code. Please go ahead and > revert it (or let me know and I'll submit a patch myself). Okay, I removed the check on relkind. regards, tom lane