Peter Keller created XMLBEANS-608:
-------------------------------------
Summary: Memory leak in interface extension handling
Key: XMLBEANS-608
URL: https://issues.apache.org/jira/browse/XMLBEANS-608
Project: XMLBeans
Issue Type: Bug
Affects Versions: Version 5.1.0
Reporter: Peter Keller
The [following
code|https://github.com/apache/xmlbeans/blob/e684ff4332aab1aba1e42cb5fb50daa9cd79e3cb/src/main/java/org/apache/xmlbeans/impl/config/ChildSolverCollectionStrategy.java#L104]
in
{{/xmlbeans/src/main/java/org/apache/xmlbeans/impl/config/ChildSolverCollectionStrategy.java}}
can leak memory:
{{ private class FileVisitor extends SimpleFileVisitor<Path> {}}
{{ @Override}}
{{ public FileVisitResult visitFile(Path file, BasicFileAttributes
attrs) throws IOException {}}
{{ if (javaMatcher.matches(file)) {}}
{{ if (roots.stream().noneMatch(file::startsWith)) {}}
{{ getRoot(file).ifPresent(r -> {}}
{{ getSolver().add(new JavaParserTypeSolver(r,
getParserConfiguration()));}}
{{ roots.add(r);}}
{{ });}}
because the paths in {{roots}} are absolute paths at this point, but {{file}}
may be a relative path. This means that the {{file::startsWith}} test always
fails, and IIUC this means that a new {{JavaParserTypeSolver}} instance is
created for every Java source file, for every interface extension. These
instances can never be garbage-collected.
For the project that I am currently trying to migrate from XMLBeans 3.1.0, I
can run the schema compiler with -Xmx50g on a large memory server, and the heap
still fills up in a few minutes.
In my hands, the following patch fixes the problem and gives a huge performance
boost as well:
{{index 25003989..a19d495a 100644}}
{{---
a/src/main/java/org/apache/xmlbeans/impl/config/ChildSolverCollectionStrategy.java}}
{{+++
b/src/main/java/org/apache/xmlbeans/impl/config/ChildSolverCollectionStrategy.java}}
{{@@ -105,7 +105,7 @@ class ChildSolverCollectionStrategy implements
CollectionStrategy {}}
{{ @Override}}
{{ public FileVisitResult visitFile(Path file, BasicFileAttributes
attrs) throws IOException {}}
{{ if (javaMatcher.matches(file)) {}}
{{- if (roots.stream().noneMatch(file::startsWith)) {}}
{{+ if
(roots.stream().map(Path::toAbsolutePath).noneMatch(file.toAbsolutePath()::startsWith))
{}}
{{ getRoot(file).ifPresent(r -> {}}
{{ getSolver().add(new JavaParserTypeSolver(r,
getParserConfiguration()));}}
{{ roots.add(r);}}
The difference can also be seen by giving absolute paths, not relative paths,
for the directories containing Java source files to the schema compiler.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
