[ psqlodbc-Bugs-1000498 ] rollback to savepoint in case of error

[ psqlodbc-Bugs-1000498 ] rollback to savepoint in case of error

am 24.12.2005 11:40:45 von noreply

Bugs item #1000498, was opened at 2005-12-21 15:19
You can respond by visiting:
http://pgfoundry.org/tracker/?func=detail&atid=538&aid=10004 98&group_id=1000125

Category: None
Group: None
Status: Open
Resolution: Accepted
Priority: 3
Submitted By: Tom Zschockelt (toz)
Assigned to: Ludek Finstrle (luf)
Summary: rollback to savepoint in case of error

Initial Comment:
the current snapshot driver only supports manual rollback to savepoints
if ( and only if ) all previous statements return with success.

e.g.

My env :

win2003 server
8.1.0 postgresql
psqlodbc 8.01.0105


CREATE TABLE testtable (
f1 int2 primary key,
f2 varchar(10) null);


the following snippet works

BEGIN
INSERT INTO testtable values ( 12, 'hello');
SAVEPOINT SP_INSERT;
INSERT INTO testtable values ( 13, 'hello1');
ROLLBACK TO SAVEPOINT SP_INSERT;
INSERT INTO testtable values ( 13, 'hello123');
COMMIT;

the following snippet don't work

BEGIN
INSERT INTO testtable values ( 12, 'hello');
SAVEPOINT SP_INSERT;
INSERT INTO testtable values ( 12, 'hello1'); <-- auto rollback occurs !!!
ROLLBACK TO SAVEPOINT SP_INSERT;
INSERT INTO testtable values ( 13, 'hello123');
COMMIT;

Why does an error breaks the savepoint environment ?

By now I used an application side workaround as I do an select before
each insert to check if a certain row already exists in the table.
If it exist I try an update on the row.
( but this is a real performance killer ... )

------------------------------------------------------------ ----------

>Comment By: Ludek Finstrle (luf)
Date: 2005-12-24 11:40

Message:
Please try new development snapshot 8.01.0106. The problem may be fixed there.

I attach patch. It have to be applied after psqlodbc-sqlstate.diff.

------------------------------------------------------------ ----------

Comment By: Ludek Finstrle (luf)
Date: 2005-12-23 10:35

Message:
I accept the bug in case autocommit = off => don't take a care on transaction state.

What driver version do you use? Do you use 8.01.0105 patched with SqlState patch?

Could you attach mylog output for second example (the failure one)?

Thanks

------------------------------------------------------------ ----------

Comment By: Tom Zschockelt (toz)
Date: 2005-12-22 20:17

Message:
I switched autocommit off with the
SQLSetConnectAttr() Function

See the first example which works well !!!



------------------------------------------------------------ ----------

Comment By: Ludek Finstrle (luf)
Date: 2005-12-22 19:57

Message:
What about autocommit StmtOption? Don't you have autocommit on?
Driver doesn't support savepoints in this case yet.

------------------------------------------------------------ ----------

You can respond by visiting:
http://pgfoundry.org/tracker/?func=detail&atid=538&aid=10004 98&group_id=1000125

---------------------------(end of broadcast)---------------------------
TIP 1: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to majordomo@postgresql.org so that your
message can get through to the mailing list cleanly