Rick Hillegas wrote:
Kathey Marsden wrote:
Hello All,
Just wanted to give a quick recap of today's lunch. Attending were
Rick, Bryan, Lily, Myrna and myself.
It was great to see everyone and chat and we did touch on some Derby
topics that I wanted to recap.
Releases
- We talked a little bit about what might be the minimum
required for a lightweight Derby maintenance release, for example on
10.5 to provide a fix for DERBY-4677 there. Hopefully it would be
something much more lightweight to produce, without docs, release
note's, mirroring etc, maybe even without source if that is ok by
apache. I will file a Jira issue for this.
Kristian mentioned that the jars produced by Oracle for nightly tests
should be public. They are available for the trunk and the last branch
(10.6). Can you see them here:
http://dbtg.foundry.sun.com/derby/test/
That may be useful for your purposes and for getting quick user
feedback on bug fixes.
- We talked about 10.7 and Rick mentioned that it was his hope to
branch mid November and try to include a fix to make Derby less
sensitive to interrupts.
JDK 1.7 and JDBC 4.1
- Rick mentioned JDK 17 withJDBC 4.1 was in the works and that
javadoc for JDBC 4.1 is in the Open JDK repository, so available to
the public.
- I suggested that JDK 1.7 have some option to disable JDBC
driver autoloading to provide a quick solution in the event of
accidental Driver interaction, resource issues and potential upgrade
issues related to loading all drivers on the classpath. He asked me
to send him an email on the issue and I will do so and cc this list.
I sent mail about this to JDBC spec lead Lance Andersen. I think that
the door has closed for Java 7 changes (jdbc 4.1), but now is the time
to raise issues which you would like to see addressed in Java 8.
Lance responded quickly to the mail I sent him, which dealt with the
interaction of driver autoloading and the special
"jdbc:default:connection" url. His sense, with which I agree, is that
the problem here results from vagueness in the SQL Standard. The
"jdbc:default:connection" url doesn't contain enough information for
drivers to figure out whether the connection request is directed at
another database. There is not much language around this in the SQL
Standard. That url is glancingly referenced in part 13, section 3.1.3.2.
There is a more extensive but still brief discussion of the url in a
1999 paper on SQLJ written by Andrew Eisenberg and Jim Melton, which you
can google up using these keywords: sqlj andrew eisenberg jim melton. In
that paper, look for the section titled "Database Access from within a
Java Method".
You may want to email Lance a description of the larger issues you are
encountering with driver autoloading. You may also want to discuss these
issues with Chris Farrar, who is both a SQLJ expert and your
representative on the JDBC committee.
Hope this is helpful,
-Rick
Harmony and Derby
- I mentioned that I had recently tried the Derby tests with the
Harmony JDK and they had fallen over completely due to security
manager issues. I will research a little more and send an email to
the Harmony group. It seems at one time they had the old harness
tests running http://wiki.apache.org/harmony/Apache_Derby, so
hopefully it wouldn't be a major effort to get them going.
In conclusion, Rick suggested for the next get together, we meet in
the east bay for beverages. I didn't mention it then, but think it
would be great to invite the user community if we do that.
Cheers,
-Rick
Thanks
Kathey
