Hi,
I am trying to understand the reasons behind
updatableResultset test failure when using DerbyNetClient. Following is
what I have found so far.
Satheesh made a checkin on May 25th (revision 178519) to the
master file DerbyNetClient/updatableResultSet.out which was as follows
+++
incubator/derby/code/trunk/java/testing/org/apache/derbyTesting/functionTests/master/DerbyNetClient/updatableResultSet.out
Wed May 25 12:39:51 2005
@@ -307,14 +307,14 @@
delete using first resultset
attempt to send deleteRow on the same row through a different
resultset should throw an exception
SQL State : XCL08
-Got expected exception Cursor 'SQL_CURLH000C51' is not on a row.
+Got expected exception Cursor 'SQL_CURLH000C52' is not on a row.
Move to next row in the 2nd resultset and then delete using the second
resultset
Positive Test11 - setting the fetch size to > 1 will be ignored by
updatable resultset. Same as updatable cursors
Notice the Fetch Size in run time statistics output.
1
-----
Statement Name:
- SQL_CURLH000C54
+ SQL_CURLH000C55
Statement Text:
SELECT * FROM t1 FOR UPDATE of c1
Parse Time: 0
On May 26th, Bernt reported a diff which is reverse of master update
done by Satheesh. Checkin from today (revision 179592) submitted by
David seems to bring the master back to the state prior to Satheesh's
checkin.
Also, Bernt, I looked at
http://www.multinet.no/~solberg/public/Apache/Derby/Limited/testSummary-178249.html
and the reason you didn't see the failure on Linux 2.4.19 and
jdk1.4.2_08 I think is because the test never got run on that machine
for some reason. In the list of tests that ran as part of
derbynetclientmats on Linux 2.4.19 and jdk1.4.2_08, I don't see
updatableResultset test in there. Let me know if I have missed
anything. But if I am right, then we don't need
updatableResultSet_sed.properties
since we should get same cursor name irrespective of different
platforms.
The only question I have is why did
Satheesh need to change the master? I am running with classes and maybe
there is something that shows up only with the jar files. Satheesh,
please let me know if there is an environment that I have not tested
which required the master update.
thanks,
Mamta
On 5/26/05, Bernt M. Johnsen <[EMAIL PROTECTED]> wrote:
When
running derbyall, DerbyNetClient/lang/updatableResultSet fails
the following way:
*** Start: updatableResultSet jdk1.4.2_02 DerbyNetClient 2005-05-26
23:12:55 ***
Initialize for framework: DerbyNetClient
java -ms16777216 -mx33554432
-Dderby.system.home=/export/home/tmp/Derby/test/Der
byNetClient/updatableResultSet -Djava.security.manager
-Djava.security.policy=/e xport/home/tmp/Derby/test/nwsvr.policy
-Dcsinfo.codebase="" trunk/jars/sane
-Dcsinfo.serverhost=localhost -Dcsinfo.trustedhost=localhost org
.apache.derby.drda.NetworkServerControl start
Attempt to shutdown framework: DerbyNetClient
310 del
< Got expected exception Cursor 'SQL_CURLH000C52' is not on a row.
310a310
> Got expected exception Cursor 'SQL_CURLH000C51' is not on a row.
317 del
< SQL_CURLH000C55
317a317
> SQL_CURLH000C54
Test Failed.
*** End: updatableResultSet jdk1.4.2_02 DerbyNetClient 2005-05-26
23:13:22 ***
(Same error with 1.5 and 1.3)
I'm running with Linux 2.6.11. What I find strange, is that when I
inspect the test failures in
http://www.multinet.no/~solberg/public/Apache/Derby/Limited/testSummary-178249.html
I see that the same test fails in the same way on all platforms,
with the exception of the test run on a Linux 2.4.19 and jdk1.4.2_08
Comment anynone?
--
Bernt Marius Johnsen, Database Technology Group,
Sun Microsystems, Trondheim, Norway
