Partial locking for a row

Question

Hello,
For example, I have two queries, the first one just accesses
the first entry of the row. And the second one just accesses the
second entry of the row.
Is there any locking method ? Such that I can lock the first
entry of the row when I use the first query and lock the second
entry of the row when I use the second query. So that I can
speed up the performance of my system in this way.
null

Answer 1

If you're trying to lock on a per-column basis, you won't be able
to, since locks are granular to a per-row basis, not below.
If you want to lock a specific row on a table, you may use a
2-phase commit method like so:
EXEC SQL
SELECT A.ROWID, B.ROWID, <rest of rowid's>, <rest of data>
INTO :blabla, :blabla,.....
FROM TEST_TABLE_1 A, TEST_TABLE_2 B, <rest of tables>
WHERE <where clause w/joins, etc...>
FOR UPDATE;
This will acquire row-locks on all tables referenced, for all
rows fetched by the SELECT.
This lock will be released when you close the current transaction
with a COMMIT or a ROLLBACK (or, if using from within a cursor,
when the cursor is closed if your configuration is set this way).
That means that you would:
1) Acquire the locks on all the rows
2) Do all your work using the ROWID's to locate the rows to query
or manipulate (while they're locked)
3) Commit or rollback
This will allow you to use row-level locks, which increases
concurrency and makes the SQL code a lot faster (because of the
use of rowid's).
I've found that the best method to do this is to use a cursor,
and then work on that cursor's fetches using the ROWID's (NOT the
WHERE CURRENT OF CLAUSE when there is more than one table
involved in the lock mechanism).
Best
Diego
Stephen Lee (guest) wrote:
: Hello,
: For example, I have two queries, the first one just
accesses
: the first entry of the row. And the second one just accesses
the
: second entry of the row.
: Is there any locking method ? Such that I can lock the
first
: entry of the row when I use the first query and lock the second
: entry of the row when I use the second query. So that I can
: speed up the performance of my system in this way.
null