Skip to content

Commit 79e070d

Browse files
shai-almogclaude
andcommitted
Bump SpotBugs forked-JVM heap for codenameone-maven-plugin
The module's SpotBugs run with `<effort>Max</effort>` intermittently OOMs on the default 512MB fork heap, with the JVM crashing inside FindRefComparison / ResourceValueAnalysis when the type-frame dataflow expands the larger mojo classes. The crash returns Java exit code 1, which `failOnError=false` does NOT suppress (that flag only swallows reported findings, not analysis crashes), so the whole reactor build dies. Observed on master CI (pojo-annotation-frameworks branch run at 2026-05-27 10:40) as well as on this branch's runs. Setting `<maxHeap>2048</maxHeap>` lifts the fork heap to 2GB, which absorbs the spikes seen so far. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
1 parent 770e1c0 commit 79e070d

1 file changed

Lines changed: 11 additions & 0 deletions

File tree

  • maven/codenameone-maven-plugin

maven/codenameone-maven-plugin/pom.xml

Lines changed: 11 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -349,6 +349,17 @@
349349
<xmlOutput>true</xmlOutput>
350350
<xmlOutputDirectory>${project.build.directory}</xmlOutputDirectory>
351351
<excludeFilterFile>${project.basedir}/spotbugs-exclude.xml</excludeFilterFile>
352+
<!--
353+
The default 512MB forked-JVM heap intermittently OOMs on
354+
this module under `<effort>Max</effort>`. Observed on
355+
both PR branches and master: FindRefComparison /
356+
ResourceValueAnalysis blow up the type-frame dataflow on
357+
the larger mojo classes here. Bump to 2GB to absorb the
358+
spikes; `failOnError=false` silences findings but does
359+
NOT catch a Java exit code 1 from an OOM crash of the
360+
fork, so without the bump the build dies anyway.
361+
-->
362+
<maxHeap>2048</maxHeap>
352363
</configuration>
353364
<executions>
354365
<execution>

0 commit comments

Comments
 (0)