tddb bug-fix.
0 views
Skip to first unread message
Dhruv Matani
Apr 4, 2006, 7:38:14 AM4/4/06
to The Distibuted DataBase, jayan...@yahoo.co.in
There was a bug when you perfromed the following operations:
[Client-1] CREATE TABLE tab1(roll int, name varchar(200));
[Client-2] SELECT * FROM tab1;[table doesn't exist --> CORRECT].
[Client-1] COMMIT;
[Client-1] INSERT INTO tab1 VALUES(1, 'dj');
[Client-2] SELECT * FROM tab1; [I get values from a very old table which was there b4 formatting the DB].
This was happening, because I wasn't invalidating the buffer cache after allocating a page, and thus the other TRX was seeing the old values in the page, left over from a possibly older table.
Fixed some other race-conditions too.
Fixed code for CLOB handling. There is currently no way to test this because the parse is not type-aware.
--
-Dhruv Matani.
http://www.geocities.com/dhruvbird/
"The biggest room is the room for improvement."
-- Navjot Singh Siddhu.
[Client-1] CREATE TABLE tab1(roll int, name varchar(200));
[Client-2] SELECT * FROM tab1;[table doesn't exist --> CORRECT].
[Client-1] COMMIT;
[Client-1] INSERT INTO tab1 VALUES(1, 'dj');
[Client-2] SELECT * FROM tab1; [I get values from a very old table which was there b4 formatting the DB].
This was happening, because I wasn't invalidating the buffer cache after allocating a page, and thus the other TRX was seeing the old values in the page, left over from a possibly older table.
Fixed some other race-conditions too.
Fixed code for CLOB handling. There is currently no way to test this because the parse is not type-aware.
--
-Dhruv Matani.
http://www.geocities.com/dhruvbird/
"The biggest room is the room for improvement."
-- Navjot Singh Siddhu.
Reply all
Reply to author
Forward
0 new messages