v0.10.0

Changes since 0.9.3

• delta handling is now in a separate thread
• delta handling of deletes has been improved
• proper support for nested objects in delta handling (note still does not take nested type into account)
• performance improvement by using construct queries to fetch properties for document
• moved to jruby 9.4 (ruby 3.1) by bump of ruby-template
• added arm64 builds
• Added query param to enable exact counts of search results
• Add query params :id: and :uri: to filter by id and URI
• Allow wait interval less than a minute
• Group updates sent to Elasticsearch per delta message
• support large yaml files

Full Changelog: v0.9.3...v0.10.0

v0.9.3

changes since v0.9.2

🐛 Fixes

• fix: typo in regex for kebap-cased property names

v0.9.2

changes since v0.9.1

🐛 Fixes

• fix: Support kebab-cased property names with modifiers: previously :gte:order-date resulted in ['gte', ['order']] while the expected result is ['gte', ['order-date']]

Changes

• change: pin base docker image to semtech/mu-jruby-template:2.13.0

v0.9.1

What's Changed since 0.9.1

🐛 Fixes

• fix: correctly handle deletion of a resource. due to a bug not all deletions in the DB also resulted in removal from the search index

v0.9.0

What's Changed since 0.8.0

🚀 Features

• better support for composite types: now properly handling delta messages
• experimental support for having multiple rdf types in one (non composite) index
• basic highlighting support
• search indexes defined in eager_indexing_groups can now partially match the user's allowed groups
• experimental support for multilingual strings
• improved handling of delta messages

🐛 Fixes

• correctly respond to SIGTERM, so containers stops quickly
• the admin endpoint to delete indexes should now work correctly
• search should no longer be blocked when a new index is created

v0.8.0

What's Changed since 0.7.0

🚀 Features

• Allow to specify wildcards and per-field boosts in fields parameter
• Multi match support for fuzzy filter
• Basic config validation on startup
• Include URI of nested objects in the document
• Extracting file content using external Tika service
• Configurable log levels per scope
• Improved error logging
• Add documentation on eager indexing groups and update queues

🐛 Fixes

• No longer throw an error in the delta handler when composite types are used (composite types are still not handled though)
• Improved, but far from perfect composite type support
• Using connection pool for all SPARQL queries
• Error handling in case a file is not found for indexing
• Allow all permutations of lt(e),gt(e) in search query params
• Taking authorization headers into account on index management endpoints
• Indexing of boolean values
• Allow dashes and underscores in search property names
• Ensure same index name independently of order of auth group keys

Contributors

• @erikap
• @wschella
• @nbittich
• @madnificent
• @smessie
• @nvdk

Changelog: https://github.com/mu-semtech/mu-search/blob/master/CHANGELOG.md#v080
Commits: v0.7.0...v0.8.0

v0.6.1

Fixes since 0.6.0:

• if a search is executed and the user does not have allowed groups yet, mu-search now fetches those first.
• better error response if no matching indexes could be found (now returns a 400 bad request)