Stylesheet not working in 4.8.4

[esabol]

We use stylesheet="fancy-hist.xsl" to generate our SpotBugs reports in our CI/CD pipeline, which builds using ant. Our build.xml has the following:

    <taskdef
        resource="edu/umd/cs/findbugs/anttask/tasks.properties"
        classpath="${spotbugs.home}/lib/spotbugs-ant.jar"/>
 
    <target name="spotbugs">
        <spotbugs home="${spotbugs.home}"
                  output="html"
                  outputFile="${dist.dir}/spotbugs.html"
                  stylesheet="fancy-hist.xsl" >
            <auxclasspath path="${basedir}/../libs" />
            <sourcePath path="${basedir}/src/java" />
            <class location="${build.web.dir}/WEB-INF/classes" />
        </spotbugs>
    </target>

This all worked for us just fine with SpotBugs 4.8.3. After upgrading to 4.8.4, however, the SpotBugs report generated by the same exact build.xml is clearly not using fancy-hist.xml. And, yes, spotbugs.home is defined properly in project.properties (which hasn't changed) and we are using the spotbugs-ant.jar from the 4.8.4 release.

Any idea as to what the problem is or what the solution might be? Thanks!

[gtoison]

Hello @Vogel612, I'm not sure how this was working but I was wondering if this could be a side effect of #2894

[JuditKnoll]

Maybe it can be because of #2915. Can you please share the fancy-hist.xml? Is there any error logged about this?

[gtoison]

I think the xsl is part of SpotBugs: https://github.com/spotbugs/spotbugs/blob/master/spotbugs/src/xsl/fancy-hist.xsl
I was wondering about the version change for the filter xsd too, that might be the problem

[esabol]

Yes, the fancy-host.xsl we are using is the one provided by SpotBugs, as @gtoison indicated. We haven't modified it or moved it.

I'm not seeing any error messages. Where would they be logged to in this scenario?

[esabol]

My colleague additionally tested it with SpotBugs 4.8.2, and the stylesheet worked.

Also tested 4.8.4 with stylesheet="${spotbugs.home}/src/xsl/fancy-hist.xsl" instead of stylesheet="fancy-hist.xsl". Still didn't work.

4.8.3 works with either of those two stylesheet attribute values.

[Vogel612]

Hello @Vogel612, I'm not sure how this was working but I was wondering if this could be a side effect of #2894

The changeset you mention should only affect the filepaths, although I did touch the stylesheet assignment so I can't exclude it with any degree of certainty. Judging by the syntax of the options, the code contributed with that changeset should be working as intended and not mess with the stylesheets, assuming generating an XML report withMessages also works as intended...

I haven't checked the release yet, but unfortunately this seems like my changes are a more likely culprit than the filter XSD namespace changes...

[asheldon]

assuming generating an XML report withMessages also works as intended...

When I diff SpotBugs XML output between 4.8.2 and 4.8.5, having set output to xml:withMessages, the shortMessage and longMessage blocks are missing from the bugInstance.

<     <ShortMessage>Redundant nullcheck of value known to be non-null</ShortMessage>
<     <LongMessage>Redundant nullcheck of maxuptime, which is known to be non-null in new com.example.Test(boolean, String[])</LongMessage>

Additionally, the <SourceLine> blocks, <LocalVariable>, etc blocks are self-closing instead of including a <Message>.

[esabol]

Is there a developer who can look into this? Maybe try reverting PR #2894?

[gtoison]

@Vogel612 if the issue is indeed with #2894 I think that the improvement in the PR is not worth the regressions reported here?
Would you be able to look into it? Otherwise I agree with @esabol that reverting it might be reasonable

[Vogel612]

@gtoison I really don't have the capacity to look into it right now. I'm happy to go for a revert, the improvement is only marginal and the regression is quite a heavy thing.

I might be able to look at it at some later point with some more test coverage around this specific element of it, but not in any reasonable time frame. I would like to see the original issue that was supposed to be fixed by that PR reopened if we revert, though :)

[asheldon]

I checked out the source for SpotBugs and on commit 6cf7b2c I did "git revert 789d397" and rebuilt the jar locally. The locally built jar appears to work correctly in this scenario.

[hazendaz]

Closing as fixed on master, will be released in 4.9.0 as code reverted.

[esabol]

Thanks, everyone!