postgresql/contrib/spi/README.timetravel
Bruce Momjian 524cfad23f Here is 4 file in tgz:
the new timetravel.c,
new timetravel.README (cut from spi/README and modified),
modified timetravel.sql.in
and modified timetravel.example.

Features:
- optionally 3 parameter for insert/update/delete user name

- work with CREATE UNIQUE INDEX ixxx on table xxx
(unique_field,time_off);
    (the  original version was work with unique index on 6.5.0-6.5.3,
and not work on 7.3.2,7.3.3)
     (before 6.5.0 and between 6.5.3 and 7.3.2 I dont know)

- get_timetravel(tablename) function for check timetravel-status.

- timetravel trigger not change  oid of the active record. (it is not a
good feature, because the  old version is automatice prevent the paralel
update with "where oid=nnn")

B?jthe Zolt?n
2003-07-27 04:51:45 +00:00

117 lines
4.4 KiB
Plaintext

2. timetravel.c - functions for implementing time travel feature.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
I rewritten this, because:
on original version of postgresql 7.3.2-7.3.3:
the UPDATE not work on timetravel.example if I added
>create unique index tttest_idx on tttest (price_id,price_off);
>update tttest set price_val = 30 where price_id = 3;
ERROR: Cannot insert a duplicate key into unique index tttest_idx
And UPDATE not work on table tttest after
>alter table tttest add column q1 text;
>alter table tttest add column q2 int;
>alter table tttest drop column q1;
>update tttest set price_val = 30 where price_id = 3;
ERROR: Parameter '$5' is out of range
And I add a new optional feature: my new timetravel have +3 optional parameters:
inserter_user, updater_user, deleter_user.
And I add a new function: get_timetravel for get timetravel status
without change it.
A big difference:
the old version on UPDATE changed oid on active ('infinity') record,
the new version UPDATE keep oid, and the overdued record have a new oid.
I sign with '!!!' my comment in this file.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Old internally supported time-travel (TT) used insert/delete
transaction commit times. To get the same feature using triggers
you are to add to a table two columns of abstime type to store
date when a tuple was inserted (start_date) and changed/deleted
(stop_date):
CREATE TABLE XXX (
... ...
date_on abstime default currabstime(),
date_off abstime default 'infinity'
... ...
/* !!! and (if have) */
ins_user text /* user, who insert this record */
upd_user text /* user, who updated this record */
del_user text /* user, who deleted this record */
... ...
);
!!! on INSERT my new version:
... and optionally set ins_user to current user, upd_user and del_user to null.
- so, tuples being inserted with NULLs in date_on/date_off will get
_current_date_ in date_on (name of start_date column in XXX) and INFINITY in
date_off (name of stop_date column in XXX).
Tuples with stop_date equal INFINITY are "valid now": when trigger will
be fired for UPDATE/DELETE of a tuple with stop_date NOT equal INFINITY then
this tuple will not be changed/deleted!
If stop_date equal INFINITY then on
UPDATE:
original version was:
only stop_date in tuple being updated will be changed to current
date and new tuple with new data (coming from SET ... in UPDATE) will be
inserted. Start_date in this new tuple will be setted to current date and
stop_date - to INFINITY.
On my new version:
insert a new tuple with old values, but stop_date changed to current date;
and update original tuple with new data, and update start_date to current date
and optionally set upd_user to current user and clear ins_user,del_user.
DELETE: new tuple will be inserted with stop_date setted to current date
(and with the same data in other columns as in tuple being deleted).
On my new version:
... and optionally set del_user to current user.
NOTE:
1. To get tuples "valid now" you are to add _stop_date_ = 'infinity'
to WHERE. Internally supported TT allowed to avoid this...
Fixed rewriting RULEs could help here...
As work arround you may use VIEWs...
2. You can't change start/stop date columns with UPDATE!
Use set_timetravel (below) if you need in this.
FUNCTIONs:
timetravel() is general trigger function.
You are to create trigger BEFORE UPDATE OR DELETE using this
function on a time-traveled table. You are to specify two arguments: name of
start_date column and name of stop_date column in triggered table.
Or add +3 arguments:
name of insert_user column, name of update_user column, name of delete_user column
currabstime() may be used in DEFAULT for start_date column to get
current date.
!!! I deleted this function, because I newer used this.
set_timetravel() allows you turn time-travel ON/OFF for a table:
set_timetravel('XXX', 1) will turn TT ON for table XXX (and report
old status).
set_timetravel('XXX', 0) will turn TT OFF for table XXX (-"-).
Turning TT OFF allows you do with a table ALL what you want.
get_timetravel() reports time-travel status ON(1)/OFF(0) for a table.
get_timetravel() and set_timetravel() not checking existing of table and
existing of timetravel trigger on specified table.
There is example in timetravel.example.
To CREATE FUNCTIONs use timetravel.sql (will be made by gmake from
timetravel.source).