db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Knut Anders Hatlen <knut.hat...@oracle.com>
Subject Re: need theories about errors seen in the jdk 1.4 test runs
Date Tue, 15 Nov 2011 14:46:52 GMT
Rick Hillegas <rick.hillegas@oracle.com> writes:

> Thanks for that explanation, Knut. This then is the first casualty of
> our decision to stop supporting JDK 1.4 in Derby 10.9. This appears to
> be the state of
> org.apache.derbyTesting.functionTests.tests.jdbcapi.ProcedureTest:
>
> a) It never ran on supported small devices.
>
> b) It raises an error now on the unsupported JDK 1.4 platform.
>
> c) It runs cleanly on the supported Java 5, 6, and 7 platforms.
>
> Here are some options for handling this noise:
>
> 1) Change the constructor to use a String argument rather than a
> numeric. That should compile to a form which will run on JDK
> 1.4--although some other discrepancy may be lurking behind this one.
>
> 2) Adjust the build script so that we compile this class using
> compile.classpath rather than java15compile.classpath. For extra
> credit, we could also use compile.classpath for other classes in that
> package which we expect to run on small devices.
>
> 3) Remove the whole test from runs on JDK 1.4. File a JIRA to fix the
> test so that it runs on small platforms. This would involve some form
> of (2).
>
> I am leaning toward option (1) because it is simple. Other opinions?

+1.

Alternatively (or additionally) I could stop the nightly runs on Java
1.4.2 so that we won't get similar noise in the future.

-- 
Knut Anders

Mime
View raw message