Releases: neo4j/graph-data-science
Releases · neo4j/graph-data-science
2.6.4
Bug fixes
- Fixed a bug in the Arrow server token cache where authentication could fail sporadically with correct login credentials.
- Fixed a bug in the Arrow server where a wrongly configured encryption key could cause the database server to fail to start.
- Adapted an internal buffer resize strategy to avoid potential overflows when creating graphs with very high degree nodes (e.g. in Louvain).
2.6.3
Bug fixes
- Fixed a bug where
gds.graph.project
would fail when a default value for an array property was used across multiple labels. - Fixed a bug where
GraphSage
memory estimations would fail. - Fixed a bug where
gds.graph.project
would sometimes not read all data when the database was using the block storage format.
2.6.2
2.6.1
2.6.0
Breaking changes
- Some of the nested fields of the config returned by the new Cypher Projection were invalid for CP and have been removed.
- Drop support for Neo4j DBMS v5.6..v5.10
New features
- Weakly connected components can now be used to improve the speed of Node Similarity by comparing only nodes belonging to the same component.
The feature can be enabled by either passing the component directly as a node property key (String) via theuseComponents
config parameter, or by setting the same config parameter totrue
, which will compute components internally using WCC as a pre-procesing step.
Bug fixes
- Fix a bug where WCC fails on node filtered, inverse indexed graphs.
- Fix a bug where non-write mode procedures would fail on composite databases.
- Fix a bug where Random Walk could fail on weighted graphs.
Improvements
- Added
degree()
function togds.graph.subgraph
node predicates. - Cancel all arrow server threads on server shutdown. This results in the client being correctly notified of the server shutdown.
Other changes
- Move GDS Arrow features out of alpha status, which most importantly includes:
- Introduction of versioned GDS Arrow endpoints
- Deprecation of the old GDS Arrow endpoints for removal in ~ 6 months
- Refactoring, more testing and regression testing
2.5.7
2.5.6
Graph Data Science 2.5.5
New features
- Add support for Neo4j 5.14
Bug fixes
- Fixed a bug where the new Cypher projection would fail when being used from the new parallel Cypher runtime.
Graph Data Science 2.5.4
Other changes
- Updated Netty dependencies to
4.1.100.Final
to fix CVE-2023-44487
Graph Data Science 2.5.3
New features
- Add support for Neo4j 4.4.27
Bug fixes
- Fixed a bug that lead to an unresponsive DBMS or even OS when running GDS projections in Neo4j on MacBooks (x86 or ARM).
- Avoid a possible race condition random walking and training in Node2Vec. Also removes the need for a timeout in Node2Vec.
Improvements
- Improved state synchronization during Arrow graph import to avoid errors due to out-of-sync messages.