From: | JanWieck(at)t-online(dot)de (Jan Wieck) |
---|---|
To: | PostgreSQL HACKERS <pgsql-hackers(at)postgresql(dot)org> |
Subject: | update on TOAST status |
Date: | 2000-07-05 16:56:30 |
Message-ID: | 200007051656.SAA10934@hot.jw.home |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
FYI,
during the day I committed a couple of changes to TOAST.
- Secondary relations for the toaster (to move off values)
are now automatically created during CREATE TABLE, ALTER
TABLE ... ADD COLUMN and SELECT ... INTO, whenever the
first toastable attribute appears in the table schema.
- The TOAST tables are now of kind RELKIND_TOASTVALUE.
- TOAST tables cannot be vacuumed separately. They are
allways vacuumend if their master table is, while VACUUM
still holds the lock on the master table.
- VACUUM doesn't propagate ANALYZE to TOAST tables.
Statistics for them are needless because the toast access
is allways hardcoded indexed.
- TOAST tables are protected against manual INSERT, UPDATE
and DELETE operations. SELECT is still possible for
debugging purposes. The name of the TOAST table is
pg_toast_<oid-of-master>.
- The chunk_data attribute has been changed to type bytea.
For now, "lztext" is the only test candidate datatype to
invoke the toaster. It can hold up to multi-megabytes now.
But be warned, this datatype will disappear as soon as "text"
is toastable.
Next I'll make pg_dump TOAST-safe. Will only take a couple of
minutes I think.
Toast tables aren't automatically created for system
catalogs. Thus I'll add
ALTER TABLE pg_rewrite CREATE TOAST TABLE
to initdb. So we'll get unlimited view complexity for free.
As soon as arrays are toastable, we might want to add
pg_class because of relacl too.
Jan
--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck(at)Yahoo(dot)com #
From | Date | Subject | |
---|---|---|---|
Next Message | Jeffery Collins | 2000-07-05 16:59:12 | lztext and compression ratios... |
Previous Message | Philip Warner | 2000-07-05 16:54:27 | Re: Re: pg_dump and LOs (another proposal) |