-
Bug
-
Resolution: Won't Fix
-
Major
-
None
-
Jenkins 1.491 - Warnings plugin 4.18 - Static Analysis Utilities 1.48 - Violations 0.7.10 - Dashboard view 2.2
I have a maven project style with the following configuration for the maven compiler plugin
<plugin>
<artifactId>maven-compiler-plugin</artifactId>
<version>2.3.2</version>
<configuration>
<source>1.6</source>
<target>1.6</target>
<compilerId>jdt</compilerId>
<showWarnings>true</showWarnings>
</configuration>
</plugin>
The jenkins config scan for compiler warnings using two parsers: Eclipse Java compiler and Java compiler. None of the advanced options are enabled (see screnshot). I have a not so big project with around 20 compiler warnings.
I am running Jenkins 1.459 with the warnings plugin 3.26 and this is working fine (the graph is generated, new warnings are found, etc).
If I switch to the new configuration (and that's been a long time I want to upgrade), the job is stuck when parsing the warnings. This is exactly similar to JENKINS-10097. See the configuration of 1.459 with the latest warnings plugins.
- is related to
-
JENKINS-15933 Simplify configuration of maven jobs (automatic parser detection)
-
- Closed
-
[JENKINS-15895] Job stuck collecting compiler warnings with the eclipse compiler
Stéphane Nicoll
added a comment - Sorry it took so long to get those info. Let me know if you need anything else.
Stéphane Nicoll
added a comment - Sorry it took so long to get those info. Let me know if you need anything else. Ok, thanks, seems that the problem is unrelated to JENKINS-10097. The process "hangs" while scanning for warnings using a regular expression matcher. How large is your console log? The eclipse parsers is a slow multi-line parser that reads the whole console log into a buffer. Would it be possible to pipe the warnings to an additional file?
BTW: I think you do not need to activate the Eclipse parser, since the warnings seems to be in the default javac format (due to the maven plugin).
Stéphane Nicoll
added a comment - 350kb.
After 1h15, it is still on at hudson.plugins.warnings.parser.RegexpParser.findAnnotations(RegexpParser.java:86)
Seems a lot for ~4000 lines
Stéphane Nicoll
added a comment - 350kb.
After 1h15, it is still on at hudson.plugins.warnings.parser.RegexpParser.findAnnotations(RegexpParser.java:86)
Seems a lot for ~4000 lines Can you please check if that happens with the plain Java parser, too? Or why are you using the Eclipse Parser? Are there warnings not covered?
Stéphane Nicoll
added a comment - it does work with only the Javac compiler as the problem is in the eclipse parser it seems. I can only use the Javac compiler and I get a proper list of warnings.
Thanks!
(Hint: maybe we should document somewhere that Maven-based jobs using the jdt compiler do not need to use the eclipse parser)
Stéphane Nicoll
added a comment - it does work with only the Javac compiler as the problem is in the eclipse parser it seems. I can only use the Javac compiler and I get a proper list of warnings.
Thanks!
(Hint: maybe we should document somewhere that Maven-based jobs using the jdt compiler do not need to use the eclipse parser) Well, actually the warnings are different if you build with maven and tycho (then you need the Eclipse parser...).
But I think it would be worth adding a note in the documentation at least.
Are you using a maven job or freestyle job to build your project? Currently there is no special support for maven jobs in my plug-in, maybe I should add that.
Setting the resolution to won't fix since there is almost no way to improve the speed of the multi-line Eclipse parser.
I am using a maven-based job. According to JENINS-15933, eclipse would be detected automatically as far as I understand. That's probably not what I want.
By the way, I am not so sure that it's a performance problem, really. Still Parsing 4000 lines after 2 hours, something must definitely be wrong. Maybe I should left it running for the night to see how it goes.
Thanks.
Stéphane Nicoll
added a comment - - edited I am using a maven-based job. According to JENINS-15933, eclipse would be detected automatically as far as I understand. That's probably not what I want.
By the way, I am not so sure that it's a performance problem, really. Still Parsing 4000 lines after 2 hours, something must definitely be wrong. Maybe I should left it running for the night to see how it goes.
Thanks. 
thread dump