KNIME 4.5.0

What's new in this version:

New nodes:
- Python Script (Labs) ("zero copy" data exchange giving significant performance improvements)
- Workflow Service: New nodes to build and call (remote) workflows, incl. support for dynamic and large data (Labs)
- Set Files/Folders Permissions
- MongoDB Run Command
- KNIME Mountpoint Connector
- MongoDB Aggregation node
- KNIME Workflow Data Area Connector
- SharePoint Online List Reader
- Local File System Connector
- Binary Object to Model and Model to Binary Object nodes
- Excel Cell Updater
- Parameter Optimization Loop Start with table input
-Generic S3 Connector

Enhancements:
- "Column Backend" table storage format moved out of KNIME Labs (ready for production use)
- Add option to read POSIX attributes to Files/Folders Meta Info node
- Support code completion in MongoDB nodes
- Support bulk execution in MongoDB for better performance
- Support Azure AD login in Snowflake Connector
- Support Azure AD login in MS SQL Server Connector
- WEKA models should not be kept in memory
- Support Kerberos authentication in PostgreSQL Connector node
- Add node decorator for single selection widgets with active re-execution setting
- Configure re-execution in single selection widget node
- Feature Request: Add "Reply to" field in Send Email node
- Support database connection initialization statement on KNIME Server
- JDBC parameter type DELEGATED_GSS_CREDENTIAL to support Kerberos constrained delegation for Microsoft SQL Server
- Deprecation of "Pipeline Pilot Connector" (users instructed to consume Pipeline Pilot's REST API)
- Recursive Loop Start and End Node to support dynamic ports
- Power BI: Improve error message in dialog if the authentication is not working
- Add dynamic port support to Counting Loop Start, Interval Loop Start and Generic Loop Start nodes
- Numeric Scorer: add adjusted R Square
- Cleanup Views available in the AP
- Support Java-based constrained delegation in DB Connector nodes
- Configure re-execution in widgets with clear defined action
- REST nodes: Additional column with error cause
- Support more recent Python version(s)
- Sunburst Chart with Option to show Tool tip text
- Platform-aware fuzzy logic / heuristics for Conda Env Prop node
- Option to check for missing values in a DB WHERE clause
- Store time zone with component/metanode template
- Case Switch (replacing old "Case Switch (...)" nodes, now with dynamic ports)
- Add option to append rows to an existing sheet in Excel Writer
- Add option to update/remove links on linked components/metanodes in captured Workflow Segments
- Add dynamic port support to Loop End and Variable Condition Loop End nodes
- Microsoft SQL Server driver updated to version 9.4.0
- Pattern Filter Panel with option to exclude items that match the pattern
- Redshift support for DB Loader
- Table Editor (JavaScript): Support dropdown selection for string columns
- Add Kerberos authentication support to REST Client Nodes (GET, POST, ...)
- (Big Data Extensions): Add support for new type annotations in Parquet Reader
- (Big Data Extensions): Support legacy LIST format in Parquet Reader

Fixed:
- Workflow Executor does not apply new port configuration when configured in executed/green state
- Azure Blob Storage Connection does not allow listing/browsing container root despite SAS token permission
- Threads for shuffling the training data have no NodeContext
- Python Serializers create tables in threads without NodeContext
- AbstractDataInputDelegator does not fulfill InputStream#read(byte[],int,int) contract
- UpdatedColumnarValueSchema doesn't preserve table spec name
- Index Query node (knime-lucene) doesn't catch DuplicateKeyException on BufferedDataContainer#close
- SHAP and Shapley Values Loop fail because of already closed resources
- Document Parsers don't support KNIME URLs
- Class Cast exception when materializing Table for Python3
- Re-execution does not show error although workflow execution failed
- "Regression Tree" appears twice in any port type selector (one is deprecated and should be hidden)
- NPE when releasing MappedByteBuffer
- External XML entity injection in workflow.knime files
- IllegalArgumentException in Testworkflows
- Unnecessary call exceptions spam log
- Error when attempting to close open closeable
- Memory leak with Fast Tables and "Add Empty Rows"
- MongoDB manipulation nodes do not check column name in configure
- Memory leak in Voting Loop End causes hanging workflows
- Polynomial Regression Learner: Multiple R-Squared is incorrect
- Division by zero when creating cursor of empty AbstractColumnarContainerTable
- NER Tag set lacks PERCENT tag value
- Wait... node should not block a worker thread from the thread pool when waiting for file
- R conda environment: Handle different capitalization of PATH on Windows
- Text Output node throws NPE for string flow variables with null value
- Division by zero leads to "NaN" in the scorer
- “Interval Loop Start” can cause infinite loops due to floating point precision problems
- Suboptimal estimation of default thread count (#cores + 2) -- new default to be 3 / 2 * #cores
- Allow leaving R preference page with invalid R home
- Component could not be updated if name contains a special character
- Error when opening views (most likely CEF only)
- String Manipulation logic breaks if KNIME open too long
- Deep learning still relying on long-deprecated pandas.DataFrame.convert_objects method
- Donut Chart shows title in the middle with weird default color
- Keybinding conflicts when starting KNIME AP
- Relative to current mountpoint/workflow and should allow to delete/copy/move workflows within the mountpoint
- Python model size is limited by serializer buffer size
- Workflow temp directory deleted while workflow is running
- (Big Data Extensions): Hive Loader should fail with a human readable error message if the file system connection is not supported