-
Improvement
-
Resolution: Fixed
-
Minor
-
Version 5.0.1
-
-
7.0.0
We use the following setup to scan for java and error-prone warnings:
def javaIssues = scanForIssues(
blameDisabled: !blame,
sourceCodeEncoding: 'UTF-8',
tool: java(reportEncoding: 'UTF-8'))
def errorproneIssues = scanForIssues(
blameDisabled: !blame,
sourceCodeEncoding: 'UTF-8',
tool: errorProne(reportEncoding: 'UTF-8'))
publishIssues(
id: 'java-compiler',
name: 'Compiler',
ignoreFailedBuilds: false,
sourceCodeEncoding: 'UTF-8',
issues: [javaIssues, errorproneIssues],
qualityGates: [[threshold: 1, type: 'TOTAL', unstable: true]])
For a build output like this it results in duplicate warnings count:
00:03:20.475 [WARNING] /home/jenkins/agent/workspace/xxx/src/test/java/foo/SettingsControllerIT.java:[139,57] [StringSplitter] String.split(String) has surprising behavior
00:03:20.475 (see https://errorprone.info/bugpattern/StringSplitter)
00:03:20.475 Did you mean 'for (String productFamStr : Splitter.on(',').split(productFamiliesStr)) {'?
See screenshot
![2019-05-13 14_14_43- [Jenkins] - Vivaldi.png](/secure/attachment/47110/2019-05-13%2014_14_43-%20%5BJenkins%5D%20-%20Vivaldi.png)
![2019-05-13 14_16_36-[Jenkins] - Vivaldi.png](/secure/attachment/47109/2019-05-13%2014_16_36-%5BJenkins%5D%20-%20Vivaldi.png)
[JENKINS-57428] Duplicate warnings when using error-prone and java scan
Michael Düsterhus
added a comment - I would recommend to define priority for parsers and put all warnings with unique identifier like (File + Line) in a global HashMap. Then if a parser has already created a warnings entry for this location all subsequent parsers will ignore it.
Michael Düsterhus
added a comment - I would recommend to define priority for parsers and put all warnings with unique identifier like (File + Line) in a global HashMap. Then if a parser has already created a warnings entry for this location all subsequent parsers will ignore it. A simpler solution would be to use the LookaheadStream in the Java Parser as well and skip all warnings that are using multiple lines.
Christian Ciach
added a comment - Why is this issue marked as "Fixed"? We are using the current Warnings-NG 8.4.0 and the "java"-tool still records all errorprone-warnings.
Christian Ciach
added a comment - Why is this issue marked as "Fixed"? We are using the current Warnings-NG 8.4.0 and the "java"-tool still records all errorprone-warnings. Christian Ciach
added a comment - I've found this PR: https://github.com/jenkinsci/analysis-model/pull/233
It supposedly fixes this issue, but well, the issue is still present for me. I need to dig deeper...
Christian Ciach
added a comment - I've found this PR: https://github.com/jenkinsci/analysis-model/pull/233
It supposedly fixes this issue, but well, the issue is still present for me. I need to dig deeper... 
Yes, both parsers match the same warning line. Just error prone has some additional entries. I'm not sure what a good approach would be to eliminate these duplicates.