Small bug in pg_dump - Mailing list pgsql-hackers

From Tom Lane
Subject Small bug in pg_dump
Date
Msg-id 7680.984440501@sss.pgh.pa.us
Whole thread Raw
Responses Re: Small bug in pg_dump  (Philip Warner <pjw@rhyme.com.au>)
List pgsql-hackers
Hi Philip,

I have not updated from CVS in a few days, but I suspect you haven't
noticed this yet: given a mixed-case table name and a scenario that
requires emitting UPDATE pg_class commands, pg_dump puts out
things like

UPDATE "pg_class" SET "reltriggers" = 0 WHERE "relname" ~* '"Table"';

BEGIN TRANSACTION;
CREATE TEMP TABLE "tr" ("tmp_relname" name, "tmp_reltriggers" smallint);
INSERT INTO "tr" SELECT C."relname", count(T."oid") FROM "pg_class" C, "pg_trigger" T WHERE C."oid" = T."tgrelid" AND
C."relname"~* '"Table"'  GROUP BY 1;
 
UPDATE "pg_class" SET "reltriggers" = TMP."tmp_reltriggers" FROM "tr" TMP WHERE
"pg_class"."relname" = TMP."tmp_relname";
DROP TABLE "tr";
COMMIT TRANSACTION;

Of course those ~* '"Table"' clauses aren't going to work too well; the
identifier should NOT be double-quoted inside the pattern.

Actually, this should not be using ~* in the first place --- why isn't
it just using WHERE relname = 'Table' ???  Seems like it's not cool to
gratuitously reset the trigger counts on other tables that contain Table
as a substring of their names.

And while we're at it, the temp table hasn't been necessary for a
release or three.  That whole transaction should be replaced by

UPDATE pg_class SET reltriggers =(SELECT count(*) FROM pg_trigger where pg_class.oid = tgrelid)
WHERE relname = 'Table';
        regards, tom lane


pgsql-hackers by date:

Previous
From: "Michal Maru¹ka"
Date:
Subject: Re: psql missing feature
Next
From: ncm@zembu.com (Nathan Myers)
Date:
Subject: Re: Uh, this is *not* a 64-bit CRC ...