| Summary: | Index should not be in workspace/all workspaces | ||
|---|---|---|---|
| Product: | z_Archived | Reporter: | Roel Spilker <R.Spilker> |
| Component: | m2e | Assignee: | Project Inbox <m2e.core-inbox> |
| Status: | CLOSED WONTFIX | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | igor |
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | Windows XP | ||
| Whiteboard: | |||
|
Description
Roel Spilker
m2e already shares remote index gz file under localRepo/.cache/m2e folder, so download is expected to happen only once no matter how many workspaces you have. m2e also performs gz->lucene index processing only once and caches the results under localRepo/.cache/m2e too. fyi, this is very cpu expansive operation which can take several minutes even on modern hardware. beyond that, maven indexer is based on lucene and as far as I understand it is not possible to share the same lucene index directory among multiple JVM instance, so we create workspace-specific copy of lucene index. I don't see what else can be improved here but I am open for suggestions. All documentation I've read suggests that is is possible to use lucene from multiple JVM's at the same time. That said, you cannot write from multiple JVMs at the same time. Lucene will create an index.lock file when an IndexWriter is active. See also http://wiki.apache.org/lucene-java/LuceneFAQ#What_is_the_purpose_of_write.lock_file.2C_when_is_it_used.2C_and_by_which_classes.3F We do welcome patches ;-) See also http://wiki.apache.org/lucene-java/AvailableLockFactories#NativeFSLockFactory The text mentions possible failures involving different classloaders in the same JVM. That suggests that using different JVM's are not a problem... I understand you welcome patches :-) I'll see what I can do, but not before the end of May :-( Closing old bug reports. Please reopen if you have patch to contribute. |