Rename sequence bug/feature
I noticed in a post recently that it was possible to rename objects other
than tables in pg_class using ALTER TABLE RENAME. I've now implemented this
in pgAdmin II for views, sequences and indexes.
Today I've had cause to dump my test database and found a minor problem:
dumping database "helpdesk"...
pg_dump: query to get data of sequence "cat" returned name "dog"
pg_dump failed on helpdesk, exiting
This can be recreated easily:
helpdesk=# create sequence dog;
CREATE
helpdesk=# alter table dog rename to cat;
ALTER
helpdesk=# select sequence_name from cat;
sequence_name
---------------
dog
(1 row)
Regards, Dave.
(now wondering how many responses will say it is a bug and how many will say
a sequence is not a table and I shouldn't be doing that anyway :) )
Dave Page <dpage@vale-housing.co.uk> writes:
I noticed in a post recently that it was possible to rename objects other
than tables in pg_class using ALTER TABLE RENAME. I've now implemented this
in pgAdmin II for views, sequences and indexes.
Today I've had cause to dump my test database and found a minor problem:
dumping database "helpdesk"...
pg_dump: query to get data of sequence "cat" returned name "dog"
Well, we could either add code to ALTER RENAME to hack the sequence
name stored in sequences, or we could remove that check from pg_dump.
I kinda lean to the latter myself; it seems pretty useless.
regards, tom lane
-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: 27 February 2002 05:20
To: Dave Page
Cc: pgsql-hackers@postgresql.org
Subject: Re: [HACKERS] Rename sequence bug/featureDave Page <dpage@vale-housing.co.uk> writes:
I noticed in a post recently that it was possible to rename objects
other than tables in pg_class using ALTER TABLE RENAME. I've now
implemented this in pgAdmin II for views, sequences and indexes.Today I've had cause to dump my test database and found a minor
problem:dumping database "helpdesk"...
pg_dump: query to get data of sequence "cat" returned name "dog"Well, we could either add code to ALTER RENAME to hack the
sequence name stored in sequences, or we could remove that
check from pg_dump. I kinda lean to the latter myself; it
seems pretty useless.
That could potentially break any user apps that (for whatever bizarre
reason) do a select sequence_name of course, though I can't imagine why
anyone would do that. pgAdmin certainly doesn't.
Either fix would be fine for me though...
Thanks, Dave.
Import Notes
Resolved by subject fallback