Details
-
Bug
-
Status: Closed
-
Normal
-
Resolution: Outdated
-
None
-
None
-
None
-
None
Description
Two parts need (better) caching
1) The facet value count map objects are cached per mutli-index reader instance. Instead, they should be cached per index reader. This can most likely be easily achieved by iterating through all index readers in the fac nav engine, similar to how the CachingMultiReaderQueryFilter this does
2) Jackrabbit Queries are not cached at all yet. This is complex because
- they do not have proper hashCode and equals impls
- they keep references to the backing index readers, easily resulting in memory leaks
Attachments
Issue Links
- relates to
-
REPO-707 Combine docidsets efficiently
- Closed