opengrok

• do not constrain file descriptors in Docker
• add --nolock option to opengrok-indexer Python tool, fix the logic for the option in the opengrok-sync tool

• change the order of passages in search results to be based strictly on file offsets
  • i.e. they always reflect the initial part of the search hits

• cleanup indexer resources a bit earlier

• Lucene 9.6
• suggester reinit/rebuild progress reporting
• fix Rust const warning with recent Universal ctags

• fix directory listing when path descriptions are present

• honor xref timeout in more analyzers, in particular the XMLAnalyzer
  • this speeds up AOSP indexing significantly
• unify thread naming

• fix API result codes for configuration endpoints
• directory listing improvements when dates/descriptions are fetched from history cache
• give name to ctags error buffer reader threads

• fix loading of project groups from read-only configuration after merging with runtime configuration

• project group configuration changed from list to map. In the configuration that would be from:

  <void property="groups">
   <void method="add">
    <object class="org.opengrok.indexer.configuration.Group">
     <void property="name">
      <string>foo</string>
     </void>
     <void property="pattern">
      <string>foo-.*</string>
     </void>
    </object>
   </void>
  </void>

to the following (note the change of structure and also in the method):

  <void property="groups">
   <void method="put">
    <string>foo</string>
    <object class="org.opengrok.indexer.configuration.Group">
     <void property="name">
      <string>foo</string>
     </void>
     <void property="pattern">
      <string>foo-.*</string>
     </void>
    </object>
   </void>
  </void>

• enable merge changesets by default for all repositories
• Mercurial repositories now honor the mergeCommitsEnabled tunable

NOTE: this requires reindex from scratch given the impact on the history cache contents.

• display links to file revisions in diff side view
• directory listing API
  • see PR #4308 for limitations

• improved logging
  • Repository objects are now reported in the logs along with all their properties
  • progress reporting for repository scan/invalidation when the indexer is run with --progress

• restore date display in directory listing (for the case when file dates are not fetched from history cache)

• display dates from history cache in directory listing
  • only applies for repositories with mergeCommitsEnabled set to True.
  • the dates for directories will not be displayed
  • can be disabled with the useHistoryCacheForDirectoryListing tunable (see webapp tunables)

• avoid exception to be thrown in the UI when getting history of a directory

• generate history cache also for file based repositories (SCCS, BitKeeper, etc.)
  • this will require more space but should speed up the UI/APIs

• avoid getting history twice when indexing
  • should speed up indexing a little bit when history cache is on

• change serialization scheme (Smile) for history cache
  • the cache is bigger however more speedier for both access and indexing
    • if the increased size is a problem, consider using file system with built-in compression (e.g. ZFS)
• restore environment variable propagation in Docker (needed for NOMIRROR)

NOTE: this requires reindex from scratch due to the changes in the history cache serialization scheme

• fix handling of Git repositories with large history by using full Git hashes
  • the UI will be using abbreviated/shortened hashes, where makes sense
• to make sure analysis of Terraform files still works, make sure to upgrade to the latest Universal ctags

NOTE: this requires reindex from scratch due to the changes in the history and annotation cache triggered by the Git handling changes

• move Tomcat temporary directory in Docker image to /opengrok/data/tomcat_temp/