A number of derbyall tests failed on a recent run on an ibm iseries machine. The test failures were as follows:
1. each time a session connects to a jar file for the first time, a message is printed (e.g. 'Attaching to /home/MYRNA/jars/derbyclient.jar'). This results in some diffs.
The solution would be to devise a pattern for each jar file & add that to the delete line in harness/Sed.java. A workaround would be to run a selection of troublesome tests (metadata.java, derbyrun.jar, metadata.java with DerbyNetClient) before attempting derbyall.
2. tools/ijConnName.sql fails because the line 'ERROR 08001 - No suitable driver' pops up in a different location. This is probably on a slow machine. A solution would be to remove the line from this test's master altogether by adding a ijConnName_sed.properties file.
3. derbynet/testconnection.java fails because the jvm generated message is a little different, instead of:
< Could not connect to Derby Network Server on host xxxFILTERED_HOSTNAMExxx - port 9393: Connection refused ...
we get more text:
> Could not connect to Derby Network Server on host xxxFILTERED_HOSTNAMExxx, port 9393: A remote host refused an attempted connect operation.
Solution would be to add more filtering to the testconnection_sed.properties file.
4. derbynet/NSinSameJVM fails because of different jvm generated strings for the main thread.
expected: main-NSinSameJVM: Derby drivers loaded
iseries: Thread-0-NSinSameJVM: Derby drivers loaded
Solution would be to add masking of this difference to the NSinSameJVM_sed.properties file.
If these tests could get converted to junit that would be preferable to fixing up masters.