|
I am OK with the name change, but just a nit: In the test BlobClob4BlobTest
you have for example this line: } catch (StringIndexOutOfBoundsException se) { assertTrue("FAIL - This exception should only happen with DerbyNet", usingDerbyNet()); If you change the name of the method, the explanatory message should probably be changes as well: assertTrue("FAIL - This exception should only happen with DB2 client", usingDB2Client()); There are several other instances of this. A data point, the "Net=db2 client", "Client=Derby client" convention is still used in the old (non-JUnit tests), but I agree it is better to remove this source of confusion and be explicit as you propose. Thanks for the comment Dag.
I'll wait a while and see if anyone else has strong feelings about this. If I go ahead with the current proposal, I'll change the comments/strings as you pointed out. Since the current convention comes from the old test harness, I think it is acceptable to change it to something better in the new one. Others might think differently, if so please speak up. BTW; If it had been "DerbyNet" and "DerbyClient" it would have been kind of okay, but it's "DerbyNet" and "DerbyNetClient" and that I find confusing. Patch 1b changes some additional comments as suggested by Dag and includes an additional change in JDBCClient.
TestConfiguration is still using the string "DerbyNet" when parsing properties from the old harness. suites.All passed, running derbyall. Plan to commit tomorrow. My test runs passed with patch 1b applied.
Committed 'derby-3763-1b-usingDB2Client.diff' to trunk with revision 677848. Backported to 10.4 with revision 677849. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
There might be better names, I'm open for suggestions!
Patch ready for review.