[
https://issues.apache.org/jira/browse/DERBY-2492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12487186
]
Myrna van Lunteren commented on DERBY-2492:
-------------------------------------------
I just committed the first version of the converted test for checkDataSource
and checkDataSource30: DataSourceTest.java with revision 526087.
Also temporarily added a method to SecurityCheck.java, which will be replacing
the similar inspect method when I remove checkDataSource*.
The only place where SecurityCheck is actually used, is in the checkDataSource
tests, as far as I could see. ResultSetMiscTest loads the class, but doesn't
actually call any methods, or so it seems to me).
There are some things left to do with this test, marked with '// TODO:'
- the fixtures testGlobalLocalInterleaf() and testReuseAcrossGlobalLocal()
leave the system
in a locked state - with embedded, shutting down the database and
reconnecting release the
locks, but with Network Server / DerbyNetClient that action results in a
DRDA_Protocol
error for testGlobalLocalInterleaf, and another, severe, protocol error for
testReuse...
This is not new - I've had trouble before running checkDataSource* with
remote server, because
of the bad state the test leaves the database in.
I think there might be a bug somewhere, because even though all statements,
preparedstatements,
connections, resources, have been closed, the locks are stell held. Or I
could just be missing
something...
Hopefully someone else can have a look and identify the problem.
I will log a separate bug for it, for I don't think fixing this need be part
of the conversion effort.
In the meantime, these two tests do not run with DerbyNetClient.
- there are some other differences between Network Server/ DerbyNetClient and
embedded that I need to find out/check if they are known bugs, or need to log
new bugs for and hang off the bug for differences (DERBY-310). The differences
were present in the original masters for DerbyNetClient, so there *should* be
bugs, or the behavior *should* be documented.
these differences are:
1. testAllDataSources(), through assertConnectionOK() and
testReuseAcrossGlobalLocal() through queryOnStatement(), returns a different
connection when comparing a connection obtained from a ConnectionPoolDataSource
or an XADataSource, and Statement.getConnection() on a statement created by
doing Connection.createStatement().
Note, that currently, testReuseAcrossGlobalLocal has been switched off
for NetworkServer, but I first added a if(usingEmbedded()) block to ignore this
difference in both places where I saw this diff pop up.
2. There is a difference in the ReadOnly value returned in
testGlobalLocalInterleaf() in the XATransaction after setting ReadOnly.
NetworkServer appears to not listen, Embedded has the changed value.
3. In testReuseAcrossGlobalLocal() when testing on a closed XAConnection,
xac2.getXAResource() does not return an error with NetworkServer, but does with
Embedded.
4. testXAHoldability(), when testing holdability of a statement created
with holdability true, outside the global transaction, the holdability inside
global transaction is different.
- Finally, the setUp() and tearDown() methods are a little inelegant and
probably could be improved.
> convert checkDataSource, checkDataSource30 and checkDriver.java to junit
> ------------------------------------------------------------------------
>
> Key: DERBY-2492
> URL: https://issues.apache.org/jira/browse/DERBY-2492
> Project: Derby
> Issue Type: Test
> Affects Versions: 10.3.0.0
> Reporter: Myrna van Lunteren
> Assigned To: Myrna van Lunteren
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
