Changes with no impact on existing DDL/API functions

Teradata

• Teradata support (MVP): SqlDBM now supports the Teradata dialect end to end — Forward Engineering, Reverse Engineering, Compare, Direct Connect, and property editors. You can create Teradata projects and model Teradata-specific features (PERMANENT space clause for CREATE DATABASE/USER, NO PRIMARY INDEX, MULTISET tables, unique and hash indexes, IDENTITY columns), and round-trip DDL between SqlDBM and Teradata

Concurrent Working

• Merge between branches: Concurrent Working now supports merging between any two branches, not just to main, so teams can mirror deployment environments (dev → test → prod) with controlled promotion. You can also create a new branch from another branch (not only from main), use a split "Merge" button, and see richer branch metadata (latest revision, lock state)
• You can now create a new branch from a locked source project or branch; the new branch starts from the last saved revision (locking still prevents merges, but no longer blocks new branch creation)

Logical Projects

• Global References are now available for Logical projects: add objects between physical and logical projects in any direction, with names shown according to the current view mode

AI Copilot

• Copilot has a dedicated tool for editing default semantic properties on entities and columns, so you can update them straight from the AI chat instead of the property editor

Databricks

• Removed the Bloom Filter Index option from Databricks projects, since Databricks has deprecated it
• (Fix) Databricks Reverse Engineering via Unity Catalog now handles data type names that previously could cause import failures

Semantic Modeling

• Semantic YAML Reverse Engineering now recognizes filters written in Snowflake's fact-with-label form (labels: [filter]) as well as the standalone filters block; Snowflake Forward Engineering emits filters in that same form to match Snowflake's spec, while non-Snowflake output preserves the imported filter format for round-trip
• (Fix) Semantic-model carry-forward no longer leaves stale default overlays behind, which had caused false-positive diffs on later compares

Forward Engineering

• (Fix) ALTER scripts between two revisions now reflect only the true delta — a compare-engine issue that added ~20 unrelated DROP/CREATE/ALTER statements for unmodified objects has been fixed
• (Fix) Newly created tables are no longer left without a database/schema assignment, which previously caused Forward Engineering to skip them or produce incorrect DDL

Global Standards

• (Fix) Saving Global Standards / naming-convention templates no longer fails when the glossary contains an entry without "Match whole word only" explicitly set

OpenAPI

• Projects created through the OpenAPI now come with the standard naming-convention templates already seeded, matching UI-created projects
• The OpenAPI now accepts 'sqlServer' (camelCase) as a dbType value when creating a project, matching the naming used elsewhere in the API
• Every public API endpoint now has a description in the generated OpenAPI spec, improving discoverability for API consumers
• Semantic-model seeding during OpenAPI project creation is now best-effort: if it fails, project creation still succeeds (the failure is logged) instead of blocking the request

Communication Page

• Refreshed the design of the Communication (in-app messaging) page

Reliability

• (Fix) Pages in the Tool app now load without a ChunkLoadError when navigating during a release — if you have the app open while a new version deploys, navigation no longer breaks