db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bryan Pendleton (JIRA)" <derby-...@db.apache.org>
Subject [jira] Commented: (DERBY-1143) derbynet/prepStmt.java fails with JCC2.6
Date Fri, 24 Mar 2006 22:27:35 GMT
    [ http://issues.apache.org/jira/browse/DERBY-1143?page=comments#action_12371815 ] 

Bryan Pendleton commented on DERBY-1143:

The failure occurs as part of the new test added for DERBY-428. That regression test is attempting
to create a JDBC batch containing as many batch elements as it can. I determined the number
of batch elements to put into the test by experimenting with the Derby Network Client and
with JCC 2.4, but apparently JCC 2.6 supports a smaller number of batch elements than JCC
2.4 did.

Deepa is researching what the JCC 2.6 limit on number of elements in a batch is. Once we know
that, we can change the DERBY-428 regression test routine in prepStmt.java so that it runs
min(C, C24, C26) number of elements, where:
 - C is the number of elements supported by the Derby Network Client, which is 65534
 - C24 is the number of elements supported by the JCC 2.4 driver, which is 65532
 - C26 is the number of elements supported by the JCC 2.6 driver (which Deepa is researching).

0x7FF is 2047 in decimal. It would be somewhat surprising if JCC 2.4 supported almost 64K
batch elements in a single batch and JCC 2.6 reduced that to 2047, and it would certainly
weaken the test to reduce the counter from 65532 down to 2047, but that may be the simplest
solution. Since the original DERBY-428 bug only occurred when at least 9000 elements were
in a single batch, setting the test limit down to 2047 would no longer test the original bug.
If JCC 2.6 really supports only 2047 elements in a single batch, then we may want to enhance
TestUtil so that we can tell whether we are running with JCC 2.4 or JCC 2.6, and then have
the test use a different value depending on the driver in use, so that when we are running
the test with the Derby Network Client we can still test that we support a large number of
batch elements.

> derbynet/prepStmt.java fails with JCC2.6
> ----------------------------------------
>          Key: DERBY-1143
>          URL: http://issues.apache.org/jira/browse/DERBY-1143
>      Project: Derby
>         Type: Test
>   Components: Regression Test Failure
>     Versions:
>  Environment: JCC 2.6
>     Reporter: Deepa Remesh

> derbynet/prepStmt.java fails on Windows and Linux with Sun JDK 1.5 and JCC 2.6. 
> *** Start: prepStmt jdk1.5.0_02 DerbyNet derbynetmats:derbynetmats 2006-03-22 19:46:37
> 82 del
> < prepStmt Test Ends
> 82 add
> > com.ibm.db2.jcc.a.DisconnectException: More than 0x7FF chained requests not allowed.
 This error may be due to a batch that is greater than 32k.
> Test Failed.
> *** End:   prepStmt jdk1.5.0_02 DerbyNet derbynetmats:derbynetmats 2006-03-22 19:46:59
> The test passed for me with Sun JDK1.5, JCC 2.4 on Windows XP. The tinderbox runs do
not show this failure. It looks like this failure is occuring only with JCC2.6. 

This message is automatically generated by JIRA.
If you think it was sent incorrectly contact one of the administrators:
For more information on JIRA, see:

View raw message