opengrok

• fix opengrok-deploy so that it can create all path components
• avoid warning on empty index directory when performing index check

• history based reindex
  • this should speed up indexing of projects that consist solely of Git/Mercurial repositories. The history and history cache have to be enabled.
• fix CVS history parsing

• new highlight colors for the UI (now 6 in total)
• Python tools: change in format of API call configuration see https://github.com/oracle/opengrok/wiki/Repository-synchronization#api-call
  • this affects both opengrok-sync and opengrok-mirror
• fix revision sorting for CVS
• avoid building suggester data for a project twice during per project indexing

• add the ability to strip outgoing changesets in opengrok-mirror with --strip-outgoing command line option and strip_outgoing configuration directive (global and/or per project)
• fix handling of RESTful API errors in the Python tools

• fix handling of the repo repository type in opengrok-mirror
• fix author parsing for CSSC
• fix timeout default for opengrok-sync

• more strict configuration syntax check for opengrok-sync
• fix async API call handling for Python tools
• fix SCCS date parsing to include time

• use negative cache for LdapUserPlugin
• if the last revision cannot be retrieved from the index, try to get it from history cache and only then fall back to the repository method
• more RESTful API endpoints were converted to asynchronous
  • notably the /{project}/indexed endpoint

• better UTF-8 detection in plain text files
• asynchronous RESTful API support
  • added connectTimeout and apiTimeout indexer tunables/command line switches

• more printable characters (#3827)
  • it is recommended (however not necessary) to reindex from scratch
• fix one case of history cache corruption
  • collateral change: the historyCacheTime tunable was removed.
• better Terraform support
• slew of fixes for problems identified by SonarCloud
• add serverName web app configuration tunable to support RSS/OpenSearch in reverse proxy environment

• ctags regexp fixes
• code cleanup and input data sanitization

• fix for IllegalStateException thrown in the UI when displaying history view of files that belong to older index
  • older index is one that does not store the last changeset of a file in its document

• Lucene 8.11.0
• bug/review links in history view are now configurable per project
  • see https://github.com/oracle/opengrok/wiki/Webapp-configuration#configuration-tunables
• repository invalidation concurrency level for the web application can now be tuned using separate property repositoryInvalidationParallelism
  • see https://github.com/oracle/opengrok/wiki/Webapp-configuration#configuration-tunables
• UI fixes

Note: there is a bug introduced in this build, causing history view to end with IllegalStateException. This happens for older indexes so the workaround is to reindex the affected indexes from scratch or skip this release. Will be fixed in 1.7.23 (PR #3779).

• history cache creation progress reporting (with --progress)
• opengrok-projadm now uses per project lock
• UserWhitelistPlugin strips leading/trailing whitespace
• history cache handling threads are now named accordingly

• Lucene 8.10.0
• enable literalMidWordUnderscores option in Showdown
• fix project/repository property setting via RESTful API
• indexer progress reporting now works better on busy systems

• fix for the spurious xref rename problem when indexing
• fix overflowing header in the UI
• fix XML insertion for the deploy Python tool (and Docker image)

• fix opengrok-mirror configuration check for configs with overridden commands

• fix commands check in opengrok-mirror
• fix regression in selecting projects with space in the UI
• add timeout for generating single xref
  • added xrefTimeout indexer tunable

• fix UI problem with displaying revisions in annotation view

• fixed Git submodule detection
• insufficient permissions are now properly reported in the web app

• added historyCachePerPartesEnabled indexer tunable to disable history cache generation in chunks
  • use this if you have giant repository (say hundreds thousands of changesets) and care about indexer performance and can assign lots of heap memory to the JVM running the indexer (say tens of gigabytes)
• commands section of the opengrok-mirror configuration is now properly checked
  • also in Docker: failure leads to sync being disabled
• fix e-mail address detection in Go sources
• Subversion no longer required for building OpenGrok