On Mon, 2007-01-22 at 10:49 -0500, Tom Lane wrote: > "Simon Riggs" <[EMAIL PROTECTED]> writes: > > ISTM that pg_dump should default to have ANALYZE statements for a table, > > when both table definition and data are dumped. If only data (-a) or > > only table definition (-s) is dumped the default should be to *not* add > > an ANALYZE statement. > > Having pg_dump emit ANALYZE was discussed and rejected years ago. > Have you read that discussion? Do you have any new arguments to make?
Well, its been suggested before, but I can't see any good arguments against. Specifically, there was one person who spoke in favour of it last time it was mentioned. http://archives.postgresql.org/pgsql-hackers/2003-02/msg01270.php If its a performance tip, we should be doing it automatically. If we genuinely believe that autovacuum will handle it, then we can simply alter the docs to say: if you aren't running autovacuum, then don't forget to ANALYZE. IMHO it is not sufficient to rely upon autovacuum to do all of this work for us. Rod mentions the discussion has been raised before but doesn't state what the arguments were: http://archives.postgresql.org/pgsql-hackers/2003-02/msg01264.php These other posts also seem to be in favour of the idea... http://archives.postgresql.org/pgsql-performance/2006-10/msg00142.php http://archives.postgresql.org/pgsql-hackers/2003-02/msg01273.php -- Simon Riggs EnterpriseDB http://www.enterprisedb.com ---------------------------(end of broadcast)--------------------------- TIP 7: You can help support the PostgreSQL project by donating at http://www.postgresql.org/about/donate
