Coalesce Customers,

We’re writing to let you know about a planned maintenance window on Saturday, July 26, from 8:00 AM to 4:00 PM Pacific Time.

During this window, the application will be fully offline and unavailable while we perform infrastructure re-deployment work to improve platform consistency and reliability.

We expect the work to take less than 4 hours, but we are reserving the full maintenance window until 4:00 PM Pacific in case unexpected circumstances arise.

What to expect

• The application will be unavailable for the duration of the maintenance window

• Users will not be able to access the platform during this time

• Service will be restored as soon as the work is complete

We recommend planning work accordingly to avoid disruption and subscribing to updates on our Status Page:

• https://status.coalesce.io/

We’ll share a follow-up notice once maintenance has been completed and the application is back online.

If you have any questions or concerns, please reach out to our Support Team. Thank you for your patience and understanding.

Mike McCune

VP of Engineering, Coalesce.io

Updates

• Copilot Preferred Node Type Version: You can set a V1 or V2 Node Type preference in AI org and project settings. The default is V1. When you choose V2, Copilot prompt context updates so it reasons about and generates the correct Node Type.

• External Node and Column Descriptions: Organizations can configure table and column descriptions to be managed outside Transform. The description field can’t be edited with this setting enabled.

• Deploy Plan Details Layout: Deploy plan details open in a larger drawer so you have more room to review changes.

• Deploy Dashboard Job Filter: You can search and multi-select Jobs on environment cards in the Deploy dashboard. Job names are listed alphabetically with their IDs, and deep-linked Jobs stay visible.

• Scheduled Jobs Cron Expression: The Scheduled Jobs table includes a Cron Expression column between Schedule Name and Last Execution. Hover the info icon to see a human-readable schedule description.

Packages

Snowflake

• Streams and Tasks 2.7.2: Adds a Task Time Out option for all task-related Node Types, including Dimension with Task, Fact with Task, Work with Task, Stream and Insert or Merge, Task DAG Create Root, Delta Stream Merge, and Insert or Merge with Task.

• Streams and Tasks 2.7.0 : Adds the Task DAG Custom-SQL Node Type for executing custom SQL within Task DAGs. Includes Task DAG Resume Root case-retention and trimming improvements.

• Data security package 2.0.0: Adds the Masking Enabled Materialized View Node Type.

• Base Node Types Advanced Deploy 2.5.7: Resolves a minor issue in the Dimension Advanced Deploy Node Type discovered during validation of related bug fixes.

Bug Fixes

• Git Input Fields During Loading: Git modal input fields are disabled while commit, push, or force checkout actions are in progress, preventing edits during loading.

• Databricks Intermittent HTTP 400 Errors: Fixed intermittent HTTP 400 responses when creating or running Nodes against Databricks with token authentication (Cloud or Browser).

• Missing Database or Schema During Plan: Planning no longer blocks when a Node's database or schema no longer exists during managed-access checks. Missing or invalid locations surface as deploy warnings.

• Deploy Dashboard Cross-Environment Jobs: Jobs no longer incorrectly show under other environment cards when you expand multiple platform environments on the Deploy dashboard.

• Results Pane Sort Order: Fixed inconsistent ordering in the Workspace results pane where the latest result could appear at the bottom instead of the top.

• V2 SQL Comment and UNPIVOT Parsing: The V2 SQL parser now respects line and block comments and parses UNPIVOT. Commented-out SQL no longer compiles into DDL or DML.

• Node Run Double-Click: Fixed an issue where rapidly double-clicking Run in the Node menu could throw a "Nodes are already running" error.

• Node Type Editor Monaco Error: Fixed a "undefined" is not valid JSON error in the Monaco editor when editing Node Type templates.

Google SSO

We've added Google as a Single Sign-On (SSO) provider. You can now log into Coalesce using Google Cloud or Google Workspace credentials Learn how to set up in Google SSO.

Updates

Copilot Create and Update V2 Nodes: When the SQL editor is enabled for your org, Copilot can now create and update V2 Node Types. Copilot writes V2 Nodes in raw SQL and automatically prevents mixing V1 and V2 Node Types.

Copilot Column Datatype and Nullability: Copilot can now change a column's datatype, precision, and nullability.

V2 Node Mapping Grid View: V2 Nodes now include a read-only mapping grid, accessible from the SQL editor. As you write SQL, the grid displays each column's transform, data type, source, nullability, description, and related fields.

Packages

Snowflake

Semantic View 1.0.1: Adds support for AI_SQL_GENERATION and AI_QUESTION_CATEGORIZATION on the Semantic View Node Type.

Dynamic Tables 2.1.8: Fixes Validate Select failures caused by the template rendering error 'ns' is undefined.

Base Node Types Advanced Deploy 2.5.5: Adds incremental loading and soft delete handling for Fact Advanced Deploy and Factless Fact Advanced Deploy Nodes.

Bug Fixes

Multi Source setNode Source Mapping: Fixed a bug where column sources and transforms were lost from all but the first source when using the setNode API on a Multi Source Node with a transformed column.

Force Checkout Data Loss Warning: Force checkout now correctly shows a data-loss warning when your version control repository format is stale, preventing accidental loss of uncommitted Workspace changes.

Azure DevOps Personal Access Token Expiration Error: When an Azure DevOps personal access token expires, you'll now see a clear error message with a direct link to the version control setup documentation.

AI Version Control Commit Message Size Limit: If a diff is too large for the AI commit message generator to process, you'll now see a specific error message instead of a generic 500 error.

Bug Fix

Table truncation not applied before insert (V2 SQL Node) - V2 SQL Nodes with the`@truncateBefore` annotation were silently ignoring it, causing rows to be appended on every run instead of replacing the table's contents. This is now fixed and `@truncateBefore` works as expected.

• Metadata Differences in Plan: When you generate a plan, the Deploy wizard includes a Changes tab on each altered Node. The tab shows property-level before and after values for DDL, metadata, source mapping, and other changes.

• COA CLI and API Environment Management: coa environments update is available in the COA CLI and the API. It supports creating, updating, and deleting deployment environments.

• Environment List Project Filter COA CLI and API adds a project filter for environments. coa environments list --project or ?project=.

Packages

Snowflake

Base Node Types 1.6.2: Removes the Treat NULL as Current Timestamp toggle from Persistent Stage, Dimension, and Fact Nodes, and includes Persistent Stage bug fixes.

Base Node Types - SQL 1.0.0 : Initial release with INSERT and MERGE support for SQL-first V2 Node authoring in Coalesce.

Base Node Types Advanced Deploy 2.5.4: Same Treat NULL as Current Timestamp toggle removal as Base Node Types 1.6.2, plus a validation stage when Last Modified is enabled that stops execution if NULL values exist in the selected timestamp column, and includes Persistent Stage Advanced Deploy bug fixes.

Cortex 4.1.0: Adds task enhancements to the AI EXTRACT Node Type: notifications, execute-as user, suspend after failures, auto-retry, overlapping execution, and expanded schedule units, and adds Node Type Switch support to CSS and AI EXTRACT Node Types.

Iceberg Tables 3.3.0: Adds the External Iceberg REST Table Node Type and fixes Iceberg Dynamic Tables validate-select behavior.

Fabric

Fabric - Base Node Types 1.0.0: Initial release with core Fabric Node Types and basic pipeline functionality.

Bug Fixes

• Duplicate Column IDs on Git Pull: Coalesce blocks checkout when a commit introduces duplicate or empty column IDs in a Node.

• COA Init Refresh Token Guidance: coa init now points you to the Deploy tab for refresh tokens instead of the Build tab.

• Column Name Editor Console Errors: Editing column names in the Node grid no longer routes non-SQL fields through the SQL web worker, which removes spurious JSON.parse errors in the browser console.

• COA Validate False Missing-Column References: Fixed an issue where coa validate reported missing column references on V2 SQL Nodes that select from V1 YAML sources when the column UUID counters were valid.

• V2 SQL Deploy with Package Node Types: When you plan or deploy V2 SQL Nodes that use a Node Type from an installed Package, Coalesce now generates tables with the correct columns. Previously, deploy could produce empty table definitions that your warehouse rejected.

Updates

• Job Scheduler Parameters: Define parameters directly in the Job Scheduler when creating or editing scheduled Jobs.

• COA Init Command:coa init is an interactive setup flow for project configuration, Snowflake credentials, storage location mappings, and default V2 Node scaffolding. Runs coa doctor at the end.

• COA Environment Management from the CLI: Create and delete environments with coa environments create and coa environments delete.

• Git checkout progress: Branch checkout in version control now shows a progress modal.

Bug Fixes

• Parameter errors during plan: Planning catches invalid parameter references before deploy or refresh.

• Join tab performance: Join tab editing on Nodes with large source mappings is faster and more responsive.

• Column lineage view focus: Column lineage keeps zoom and pan position when switching tabs.

Packages

Snowflake

• Semantic View 1.0.0: Initial release of Semantic View and Semantic Query Node types for Snowflake semantic layer workflows.

• Incremental Loading 2.1.1 : Adds Node Type Switch support in advanced deploy templates.

Databricks

• Databricks Base Node Types 1.1.1: Adds unmatched record strategy options to Persistent Stage and Fact Nodes.

• Lakeflow Declarative Pipelines 1.1.0: Adds file format and advanced file format options, with support for AVRO, ORC, TEXT, and XML.

This release includes bug fixes and performance improvements.

We fixed an issue that causes browsers to crash when building and testing subgraphs.

Updates

• Job Scheduler cron timezones: Scheduled Jobs accept a timezone so runs align with the clock you intend.

• New UI navigation polish: We refined sidebar and Build workspace visuals and labels, including Workspace Settings and Build Settings tooltips, a higher-contrast graph background, updated secondary grey theme transparency, and a black logo mark in the build workspace.

• Git validation for SQL Nodes: Validation during Git operations catches SQL Node and Node Type mismatches and duplicate Node IDs before they cause downstream errors.

• CLI sources: CLI source handling improvements support local-first workflows.

• Projects list search: You can search the Projects list to reach a Project faster in large organizations.
  

Packages

Snowflake

• Base Node Types 1.6.1: Adds a delete option for Persistent Stage and includes bug fixes in base Nodes.

• Base Node Types Advanced Deploy 2.5.3: Persistent Stage delete option, Metadata Stage handling for Jinja in text inputs across advanced deploy Node types, and other bug fixes.

• Functional Node Types 4.3.3: Adds a ddlColumncasing macro and updates Pivot and UNPIVOT Node types to quote pivot columns consistently.

Bug Fixes

• View deployment ordering: Views that depend on one another are dropped and re-created in the correct order, ensuring upstream views always exist before downstream views are updated,

• Git checkout after browser refresh: If you refresh the browser while the app is syncing files with Git checkout, the Git UI can still write files.

• Snowflake MODEL() parsing: Snowflake ML MODEL() invocation parses correctly in SQL Nodes.

• Job-spec selector case sensitivity: Job-spec selector name matching is now case-sensitive.

• coa doctor Snowflake key pair path: coa doctor now accepts snowflakeKeyPairKey .