db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rick Hillegas <rick.hille...@oracle.com>
Subject outstanding bugs in interrupt handling
Date Wed, 09 Mar 2011 16:50:31 GMT
I see the following open bugs related to the improved handling of 
interrupts in Derby. Are there others?

First, there are a couple cleanup bugs which Dag logged. These have to 
do with handling error conditions more gracefully. Neither of these 
corrupts data or returns wrong results. Neither of these looks like a 
regression to me:

o DERBY-4871: This is an enhancement request for better error logging in 
an edge case.

o DERBY-4994: This amounts to converting an NPE into a "Connection 
closed" exception. This appears to me to be a pre-existing issue which 
simply surfaced during the testing of the interrupt handling.

In addition, a couple bugs have surfaced during test runs. I do not see 
any indication of a regression here. I do not see any evidence of data 
corruption or wrong results. At worst, Derby may not be handling an 
interrupt as gracefully as we would like. We are seeing these errors 
because we are bothering to test interrupts now. If we ran these tests 
on previous versions of Derby, the tests would fail more spectacularly: 
instead of terminating the connection, they would terminate the engine.

o DERBY-5081: There appear to be two issues in here. Neither appears to 
be a regression to me.

o DERBY-5109: This appears to be a duplicate of an already fixed issue.


View raw message