[
https://issues.apache.org/jira/browse/DERBY-3445?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12578721#action_12578721
]
Vemund Østgaard commented on DERBY-3445:
----------------------------------------
Doesn't the <junit-all>, <junit-core>, <junit-jdbc4>, etc. behave in the same
way as <junit-single> by not causing the build to fail if a test fails? At
least I cannot from the code see that there is any difference. None of these
targets have something like:
<fail if="tests.failed">Tests Failed!</fail>
>From my reading only the <junit-html>, the <junit-*-codeline-jars> and the
><emma-*> targets do this.
If all the junit targets were changed to cause build failure if a test fails,
then targets like <junitreport> would abort without producing a report if there
is a test failure? The same would be true if running with emma, no coverage
reports, at least thats what happened when I tested it with <emma-single>. I
don't know if it is possible to get ant to proceed with the next task even if
one it depends on fails?
Maybe a solution here would be to add a <junit-single-report> target to be the
equivalent of the <junitreport> target, which would then build a report after
the test is run and cause a build failure if the test/suite fails? The
<junit-single-report> target can be used when it is important to get test
failure translated to build failure, while the <junit-single> target would
remain to be used by <emma-single> and <junit-single-report> or directly if you
are not concerned about the build failure problem.
> Make it easier to use the EMMA tool to measure the code coverage of the Derby
> testing
> -------------------------------------------------------------------------------------
>
> Key: DERBY-3445
> URL: https://issues.apache.org/jira/browse/DERBY-3445
> Project: Derby
> Issue Type: Improvement
> Components: Build tools
> Reporter: Vemund Østgaard
> Assignee: Vemund Østgaard
> Attachments: 3445-emma-clean-fix-diff, 3445-general-diff,
> 3445-general-diffv2, 3445-singletest-codeline-jars-diff,
> 3445-singletest-diff, 3445-SysinfoLocaleTest-diff, 3445-testspecific-diff,
> 3445-testspecific-diffv2
>
>
> It is a bit tricky to use EMMA to measure code coverage for the derby testing.
> Modifications must be made to the source both to avoid problems with the
> SecurityManager and individual tests. It would be good if these modifcations
> could be done once and for all so that it was easier for anyone to run the
> tests with EMMA.
> It would also be good to have ant tasks that would make it even easier to run
> the tests with EMMA.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
