Releases: oracle/opengrok
Releases · oracle/opengrok
1.8.1
- fix handling renamed directories in Subversion
- handle null input files for xref ctags processing in the webapp
1.8.0
- Lucene 9.5.0
- fix negative LOC counts and duplicate search results for multi-segment indexes (the default)
NOTE: this release requires reindexing from scratch (i.e. wipe out data directory and reindex)
1.7.42
- add check to both indexer and webapp that ctags can actually process files under source root
- note that the source root directory must be writable by the indexer
- fix repository scan depth handling
1.7.41
- better signal handling in Docker
1.7.40
- support Tomcat 10.1.x
1.7.39
- change the format of annotation cache to Smile
- requires reindexing from scratch (for the projects that already have the annotation cache created; the annotation cache is off by default)
- more SCMs in the Docker image
1.7.38
- decrease log level for annotation problems
- fix long standing index performance problem
- in the 2nd phase indexing the CPU was underutilized
1.7.37
- annotation cache
- disabled by default, as it increases indexing time significantly
- on the other hand decreases annotation view times in the UI, at least for the current revision of given file
- can be enabled via indexer option
--annotationCache on
globally or per project
- disabled by default, as it increases indexing time significantly
- constrained Tomcat to 10.0.x (until #4075 is resolved)
opengrok-mirror
Python tool configuration now allows to ignore a project completely- the
/reindex
endpoint should work again in Docker image
1.7.36
1.7.35
- removed the "optimization" (a.k.a. reduction of index segments to 1) step at the end of indexing
- improves indexer performance
- one can still perform this operation using the
--reduceSegmentCount
indexer option- Lucene documentation recommends this is only done for indexes that are to be archived and made read-only
- the option might go away in the future (to be replaced with tunables for Lucene segment merging)
- the Docker image now has SSH client
- so the repositories can have
ssh://
URI- one has to make sure the SSH authentication works in the container
- so the repositories can have
- usename/password is now configurable on project level
- this is currently used only for Subversion repositories
- the Subversion related environment variables were removed
- thread names now have common prefix
- this is useful for debugging or observing indexer/webapp (e.g. via
jstack
or such)
- this is useful for debugging or observing indexer/webapp (e.g. via
- avoid reopening index directory for various operations
- should improve performance of the webapp a bit