Sorry, should have said: maven job. Thanks!

Ok, I'll see what I can do. I don't currently have any Maven stuff to test with.

Here is the code for Regression Trigger. I'll have to try and setup something with Maven to test out your issue.

public boolean trigger(AbstractBuild<?, ?> build) {
    if (build.getPreviousBuild() == null)
        return build.getResult() == Result.FAILURE;
    if (build.getTestResultAction() == null) return false;
    if (build.getPreviousBuild().getTestResultAction() == null)
        return build.getTestResultAction().getFailCount() > 0;
    return getNumFailures(build) > 
            getNumFailures(build.getPreviousBuild());
}

I'm trying to figure out the history behind the getNumFailures method, it was implemented before I started working on the plugin.

I'm seeing something similar in email-ext 2.24.1 and Jenkins 1.480.1.

My failed and unstable triggers work ok, but it doesn't seem to catch regressions.
The logs does not contain anything useful either...

Could you post any workarounds if you find any?

There wouldn't be any workarounds...there seems to be a code issue.

I would also like to suggest that if one test is fixed and another is broken between builds it should still count as a regression I guess...

Maybe we could check if there is a failing test with age 1 and then trigger the regression email?

If you are familiar with Java, you could fork the plugin on GitHub, make and test the changes and issue a pull request. It may get done sooner that way

Code changed in jenkins
User: Alex Earl
Path:
src/main/java/hudson/plugins/emailext/plugins/EmailTrigger.java
http://jenkins-ci.org/commit/email-ext-plugin/52dae3a93c5acb1a8e0a8f3149c49a5a2c4abc2e
Log:
Fix JENKINS-16404

Added a check for the value of getFailCount if the child
reports did not have any fails.

–

Added check of getFailCount if the child reports don't have fails