Rick Hillegas (JIRA) wrote:
I think that derby-dev should continue to discuss the issue which Olav has
analyzed in DERBY-1399. I will continue this discussion shortly.
If the behaviour change for DERBY-930 is intentional a special release
note is needed for this issue in the section for
"Changes that might affect existing applications":
It might read something like this ...
PROBLEM: Autoloading of driver causes changes to behavior of derby
properties.
SYMPTOM:
Derby does not recognize derby.system.home or other properties set by
the application and creates database in the wrong location or fails to
recognize properties.
CAUSE:
Automatic loading of the driver with JDBC 4.0 causes derby to be
booted before the derby.system.home property has been set. Therefore
the property is ignored.
SOLUTION:
There is no product solution to this issue.
WORKAROUND:
The only workaround is to set derby.system.home before any connection is
made to any database. For example derby.system.home can be specified
on the command line with -Dderby.system.home=<location>
Now that I put it like that I can imagine all the users in the world
that this is going to impact and I become more convinced that this is a
bug and a serious regression. I liked the suggestion to defer booting
derby embedded until the first embedded connection is made, but
apparently there are other implications for
derby.drda.startNetworkServer that I don't quite understand yet. IMHO
they all should be fixed if possible so we don't break existing
applications.
Kathey
