The way junit test runs via ant are setup is a single target
junit-oneclass and multiple antcall calls to that target, each supplying
a different test class.
I wondering if there's a good reason to do it this way? The ant junit
task can take multiple <test> arguments which would achieve exactly the
same effect.
E.g. (extract)
<junit printsummary="on"
fork="yes" forkmode="perTest"
jvm="${derby.junit.jvm}"
showoutput="yes"
dir="junit_${derby.junit.timestamp}/testout"
errorproperty="tests.failed"
failureproperty="tests.failed">
<formatter type="xml"/>
<test name="org.apache.derbyTesting.functionTests.tests.tools._Suite"
todir="junit_${derby.junit.timestamp}"/>
<test
name="org.apache.derbyTesting.functionTests.tests.jdbcapi.JDBCDriversEmbeddedTest"
todir="junit_${derby.junit.timestamp}"/>
etc.
This would be a more natural way of writing the ant file and would allow
the "tests.failed" property to work. I added that the other day but it
doesn't work because a property set in an antcall is not visible to the
callee.
Dan.
