KNIME 5.5.0

What's new in this version:

New nodes:
- Microsoft Fabric Data Warehouse Connector
- Google Workload Identify Federation node
- Message Part Extractor
- Message Creator
- Agent Chat View
- Workflow to Tool node
- Agent Prompter Rewrite
- Microsoft Fabric Workspace Connector
- Vertex AI Connector node
- Google AI Studio Authenticator node
- Gemini Embedding Model Connector node
- Gemini Chat Model Connector node
- IBM watsonx Embedding Model Connector node
- IBM watsonx Chat Model Connector node
- IBM watsonx Authenticator node
- Anthropic Chat Model Connector node
- Anthropic Authenticator node
- Microsoft OneLake Connector
- Power BI Model Reader
- Power BI Model Refresher
- BD-1350: (Big Data Extensions): Delta Table Reader (Labs)
- Date&Time Rounder
- Duration Format Manager

Enhanced:
- Call Workflow Nodes: Set parent job identifier when creating callee job
- Row Filter to allow equality check between compatible numeric types
- Row Filter to support "is not equal " variant that retains missing cells
- Expressions: Keep selected text as argument when inserting functions
- Expression node should auto-complete boolean operators "and ", "or ", and "not "
- Allow removal of arbitrary port from extensible port groups
- Web UI for Column Name Replacer (aka Column Rename (Regex))
- Web UI for UNIX Timestamp to Date&time
- Trim native (outside JVM) memory usage from the Watchdog periodically and on low resources
- Make boolean operators case insensitive in expressions to allow "AND ", "OR ", and "NOT "
- Node name improvements and other minor enhancements in AI Extension
- Tableau Writer to be streamable (performance enhancements)
- Allow multi-modal message column as input for LLM Prompter
- Allow python node view to be completely defined by static resources
- Support multi resources OAuth authorization in Microsoft Authenticator
- Add description-field to Workflow Service Input/Output nodes
- Introduce multi-modal Message data type
- Introduce Workflow Tool data type
- Allow underscores in number strings going into parse_int and parse_float
- More robust "state notification " when workflow/node state changes (e.g. notifications may break state in executor)
- Improve "KNIME > Customization Profiles " preferences page
- Ability to "soft lock " mount point preferences using AP customization
- (Internals) Invisible ThreadPool threads to not stay invisible for the remainder of their lifetime
- Constant Value Column node: Improve Performance with Columnar Backend
- Add date and time expression types and functions to K-AI’s prompt
- LLM Prompter: Support JSON prompt columns
- Secrets Retriever node should show a proper exception if an extension is missing for a selected secret e.g. Databricks
- "BufferedDataTable Reference Reader "-Nodes not copy-paste-able between Workflows
- Run REST Request nodes' I/O without consuming KNIME thread pool slot (i.e. "invisibly ")
- (Internal Metrics:) Track KNIME Executor’s internal thread pool queue length
- K-AI Build Mode: Display K-AI’s explanation of its actions
- Breakpoint node: Option to fail on non-empty table
- Expose health metrics via standard Java monitoring API
- Extend Expression nodes to handle Date&Times
- Improve Python extension installation robustness, speed and size by using pixi
- Cancel node execution if watchdog detects memory pressure
- Report watchdog metrics to prometheus
- Make KNIME Explorer synchronization less resource intensive on remote end
- "KNIME Audio Processing " marked legacy (due to low usage) - still available via update site but uncategorized
- Support Markdown in Quick Build Mode
- K-AI Quick Build Mode: Support building from scratch, or from multiple nodes
- K-AI Build Mode: Provide port type information to K-AI
- AP to be more error robust (=better error message) when system runs out of temporary disk space, default 100MB space left
- Secrets Retriever node should fetch secrets in parallel
- REST Client Nodes: Split default timeout to 5s connect and 120s read defaults
- Data types to have user-friendly names (e.g. "Set " ? "Collection (Set) " or "Period " ? "Duration (Date-based) ")
- Google Ads Query node automatically converts column names and values to human-readable formats.
- Legacy H2O model versions (3.36-) separated into separate hidden extension (no longer installed by default)
- Add most nodes with modern dialog to K-AI build mode
- Decision Tree Learner: Fix links in description
- Ability to load preference profiles from multiple providers in AP
- Become verified publisher of KNIME Analytics Platform Azure App
- BD-1346: (Big Data Extensions): KNIME H2O Sparkling Water Integration uncategorized (to be removed); after marking legacy in 5.3
- BD-1342: (Big Data Extensions): Update Databricks JDBC driver to version 2.7.3
- Update Equo Chromium 128.0.13
- Introduce Reset-Before-Upload Handling for Hub/Server Workflows
- "Reveal in space explorer " to unfold sidepanel
- Enable embedded dialogs by default
- "Info " panel for metanodes is empty
- Enable Copy/Move for items in local space
-Align nodes
- Web UI for Integer Configuration
- Web UI for Double Configuration
- Web UI for String Configuration
- Don’t allow renaming the same column twice in the Column Renamer
- Web UI for Boolean Configuration
- Enable Embedded Component Dialogs
- Web UI for Row Sampler (aka Row Sampling)
- Web UI for Nominal Value Row Splitter
- Web UI for Reference Row Splitter
- Web UI for Reference Column Splitter
- Web UI for Column Splitter
- Web UI for Variable Creator
- WebUI for DB Query Extractor
- WebUI for DB Query Injector
- WebUI for DB Column Filter
- WebUI for DB Column Name Replacer (aka Column Rename (Regex))
- Web UI for Table Column to Variable
- Web UI for String to Duration
- Web UI for Duration to String
- Web UI for Duration Part Extractor (aka Extract Duration Fields)
- Web UI for Duration to Number
- Support binary/integer columns as category
- Web UI for Date&Time Difference
- Web UI for Constant Value Column
- Web UI for Table Partitioner (aka Partitioning)

Fixed:
- Python node dialog moves and resets file stores of incoming row-based table
- K-AI code suggestions in multi-expression context doesn’t know about columns added before the current expression
- Manual Aggregation via GroupBy on a Network column produces error
- Chemistry type value factories do not adhere to KNIME type system contract
- 'Temperature' parameter in HF Hub LLM Selector node does not affect responses
- GraphCell is not registered at extension point
- Parallel Coordinates Plot: Selection not working in component with other views
- ColumnFilter overwritten by flow variable behavior changed
- Port view factory and port spec view factory are flipped for CredentialPortObject
- Excel Reader throws IllegalStateException when reading XLSB and parsing is canceled quickly (e.g. in loop)
- LLM Chat Prompter requires an empty table to be provided when starting a new conversation
- KnimeExtensionArray.to_pylist called with unexpected argument with pyarrow 20
- String flow variables with the value null cause Expressions to fail with NullPointerException
- GPT4All Connectors unnecessarily contact the gpt4all website
- DB Pivot can make AP unresponsive and takes a long time to react on an attempt to cancel execution
- Attempts to log to closed appenders (again)
- Metanode name in Description view data is not encoded (Classic UI)
- Terminal stays open when starting AP on Windows via knime-hidpi.bat
- Expression If function returns wrong values for large INTEGER values
- "Build with K-AI " button in Quick Node Add dialog should be bottom-anchored
- Parts of K-AI’s messages are formatted differently
- Python extensions cannot be uninstalled on Windows if there’s a space in the path
- Framework sets NO_PROXY environment variable incorrectly
- Expression diagnostics do not update reliably when inserting columns or from function catalog via double click
- Expression editor steals focus if the node is selected in the embedded mode
- Concatenate: The reported number of skipped rows is not accurate, off by one
- Python Extension Framework hides relevant stack trace if extension import fails
- Very long scripting AI prompt makes AI popup unusable
- Very large expressions can cause a StackOverflowError
- Cannot use modules or packages named "ports " in Python extensions because the name clashes with ports.py in our Databricks extension
- Expression editor not accessible
- String representation of unknown python types show internal types
- Deleting a not-selected expression changes the selection to the previous expression to the deleted expression
- Remove Top-Level Domain Check in Web Interaction Navigator Node
- Invalid regexes show ugly error message, error obstructed in large embedded mode
- When appending the same table twice using VirtualTable, the second copy gets stuck on the last row
- Only render node or component view tab outputs for nodes which have a view
- Table View: Opening view in disabled branch raises com.fasterxml.jackson.databind.JsonMappingException
- Casing of space items cannot be changed by "rename " in Space Explorer on macOS
- Retain correct port description on port removal for components
- Importing KNWF or KNAR file into local space keeps temp files until AP is shut down
- Component loading can’t be cancelled
- Cannot Drag and Drop Workflow from Non-Authenticated Hub
- Opening Temp Copies of Remote Workflows Not Cancelable
- Ctrl + Enter in node not always triggers node execution
- Inline error messages are not reset when new errors occur
- Unsetting flow variable not applyable in some cases
- Missing (null) valued control flow variable causes a non-user-friendly error message in BarChart
- File chooser component cannot handle non existent drive on windows
- Unsetting deprecated flow variables does not make dialog dirty
- Table View: Illegal Group Reference when used with String Formatter