3.2.25

Core

• Add support of @fieldTypes in the json from new query engine
• Add support for multiple json or parameter in insert content in the query engine

Lucene

• Correct lucene search on class function
• Fix query engine integration with geospatial lucene

3.2.24

Changes

Core

• Optimized OResult to load records lazily
• Optimized OResult to avoid early load of link in properties
• Minor optimizations in the query engine runtime
• Fix insert into index step to use the right value
• Minor dependencies updates
• Fix regression in rid comparison in case of embedded query
• Fix regression in direct insert into index query
• Fix regression in update index step
• Avoid storage to go in read only mode if index key type validation fails

3.2.23

Core

• Optimization to avoid load record from indexes that do not match the target class
• Correct delete query in new query engine to consider also binary record
• Avoid to create temporary records in query engine equal operator
• Correct the delete edge with single from or to
• Correct optimization of count queries, issue #10021
• Moved logic for truncate cluster to the database layer
• Avoid to scan all the database files for identify a database, issue #10024
• Add check to avoid to try lock not existing ridbags

Remote

• Avoid to cast to the underlying specific type while reading a ridbag from the network, issue #9499
• Minor fix for lookup of remote session from thread local

Lucene

• Make sure that lucene functions do no fail when no class is defined

Distributed

• Correct distributed metadata query

3.2.22

Core

• Optimizations of escaping in serialization of json strings
• Avoid to store in result metadata not network serializable values, issue #10004
• Re-enable transaction methods in scripts issue #9690
• Optimization index rebuild with transactions
• Minor optimization in free space map logic
• Fix space reuse, only use spaces in pages that have enough of it, issue #10005

3.2.21

Core

• Make sure to reload correctly view index informations on context restart
• Changed logic for tracking the cluster and index views in the queries
• Optimized view refresh using batch transactions
• use every time the most specific index when multiple are found
• Update build dependencies to most recent versions

3.2.20

Core

• Refactor in query engines for execution simplification
• Corrected failure to wait for migration when storage is reused, issue #9974
• Made last refresh time of views persistent
• Minor fix in views to skip refresh of deleted views

Distributed

• Reduced logging logic that is not used anymore
• Inverted order of finish check for chunks to avoid leaving data in the buffer, issue #9985

Client

• Make sure that client executor pool size never have size 0

3.2.19

Changes

Core

• Correct the class for scanning vertex on graph repair, issue #9970
• Reduced some dependencies from external libraries
• Removed not needed implementations in native logic
• Removed the need to load a native library to check some system settings
• Moved sources in main modules to avoid split packages, issue #8001
• Minor fix for new internal experimental configuration store
• Revert upgrade of minor dependency that cause issues on older OS
• Minor fixes from static analysis
• Make sure to unblock other storage open operations only when migration is complete, issue #9974

Distributed

• Deduplication and simplification of distributed logic
• Removed operation waiting flow not needed any more
• Avoid startup failure in case of rebalancing failure
• Avoid that failure on load of database block the server starting
• Add command to fetch distributed configuration from SQL

3.2.18

This release contains two major changes to be aware of, first the orientdb-gremlin as been upgraded to the last public release of gremlin, as well the agent has been trimmed of dependencies so it has a way smaller jar, at following the list of changes in this release.

Changes

Core

• Add a specific executor, for all the io operations
• Make sure to use the context executor for all background ops
• Fix automatically remove null links on write of embedded ridbags, issue #9767
• Make sure to use the plugin class loader to load the plugin class
• Avoid repeated distributed database shutdowns
• Fix race in registering installing database to guard against concurrent installs
• Upgraded dependencies
• Minor fix on embedded storage reload
• Increased separation of schema logic for remote and embedded cases

Remote

• Corrected edge iteration in remote after commit, with relative test case issue #9805

Gremlin

• Upgraded to the last release of gremlin libraries

Agent

• Removed not needed dependencies, to reduce the agent bundle size

3.2.17

Core

• update dependencies
• make sure to make view query wait for drop and create operations
• changed security checks for the import logic to avoid spurious failures
• optimizations in disk cache implementation
• pre-check cluster existence to avoid storage to go in read only mode in case of duplicate cluster
• add check to avoid failure when engines are unloaded, issue #9629
• removed not used configurations
• moved clean-up of old structures out of the view refresh window
• replaced global executor with context executor
• use correct index in immutable view metadata
• add configuration for define the max number of threads for the context executor

Server

• make sure to print the right protocol for studio address when ssl is enabled

Client

• minor fix on view metadata serialization and demineralization over network

Distributed

• Removed not used single cluster management in distributed sync

Spatial

• Fix spacial function filtering when used trough the new query engine

3.2.16

Core

• removed the need to buffer changes before adding the to the transaction
• make sure to stop scheduling of view refresh before closing the OrientDB instance
• improved refreshing and cleaning of views
• minor fix to handle miss-typed records in delta serialization
• minor correction on metadata snapshot logic on release of index manager lock
• make sure to use transaction aware stream also with legacy index implementations, issue #9914
• reduced the number of times the storage goes in read only mode, and allow to delete not empty data structures
• make sure that on index load the schema snapshot are available, issue #9915
• changes in index manager and view implementation to avoid cross locking during updates
• avoid to re-index new cluster when swapping the cluster in views.
• make sure to initialize the timeout checker before the system db

Remote

• made remote index changes go through the transaction instead of the query

Distributed

• make sure to have the correct transaction in the context for distributed phases execution
• managed correctly the check of all value of a unique index during first phase checks