V0.6.1

[benoitcayladbx]

OntoBricks — Release Notes V0.6.1

Release date: 2026-07-03
Type: Patch release (v0.6.0 → v0.6.1)
Test status: 2990 passed, 275 skipped, 5 deselected (uv run pytest -q -m "not scenario").

---

Summary

v0.6.1 is a focused patch on top of v0.6.0. It fixes the Switch modal on published /
in-review domains, hardens Lakebase deploy diagnostics and database-name resolution,
adds Databricks CLI profile support to deploy scripts, and backports a mapping SQL
column deduplication fix that prevents AMBIGUOUS_REFERENCE errors during digital-twin
builds.

No breaking changes. No schema migrations required.

---

Highlights

• Switch modal on read-only versions — on IN-REVIEW or PUBLISHED domains, the version
  picker is now usable while the "Save my changes before switching" option is correctly
  disabled (fixes #97).
• Lakebase deploy resilience — shared diagnostics helper, underscore/hyphen slug
  tolerance when resolving db-… resource segments vs Postgres datname, and clearer
  failure hints with ready-to-run CLI commands.
• Deploy ergonomics — DEFAULT_DATABRICKS_PROFILE in scripts/deploy.config.sh
  pins the CLI profile for all deploy and bootstrap targets.
• Auto-Map SQL fix — duplicate id/label column references in agent-submitted
  mappings are deduplicated before build (backport from PR fix(mapping): deduplicate Auto-Map SQL columns colliding with ID/Label alias #98, Eric Poilvet).

---

Bug Fixes

Switch modal on published / in-review domains (#97)

On non-DRAFT lifecycle versions, the L2 Switch popup previously kept the version
dropdown disabled and still offered "Save my changes before switching" even though
read-only versions cannot be saved.

• permissions.css — exempt #switchVersionSelect from the read-only select disable rule
  (navigation, not design editing).
• navbar.js — isSwitchSaveAllowed(), configureSwitchSaveOption(), and a confirm-handler
  guard disable save-before-switch on non-DRAFT versions.
• docs/user-guide.md — documents read-only Switch behaviour (version picker available,
  save option disabled).

Auto-Map SQL column deduplication (PR #98 backport)

When the Auto-Map agent repeats a source column under its original name alongside an
"AS ID" / "AS Label" alias, and that original name is itself id or label
(case-insensitive), Databricks/Spark resolves both as the same output column, causing
AMBIGUOUS_REFERENCE when the digital-twin build creates the triple-store VIEW.

• src/agents/tools/mapping.py — call SQLWizardService._deduplicate_select_columns
  after LIMIT stripping in tool_submit_entity_mapping and
  tool_submit_relationship_mapping.
• src/back/core/sqlwizard/SQLWizardService.py — _deduplicate_select_columns is now a
  @staticmethod so agents can reuse it without constructing a full service instance.
• CONTRIBUTORS.md — Eric Poilvet (@epoilvet) added.

---

Deploy & Operations

Databricks CLI profile (deploy.config.sh)

Deploy and bootstrap scripts now honour a single source of truth for the CLI profile.
Users with multiple Databricks profiles no longer need to export
DATABRICKS_CONFIG_PROFILE manually before every make target.

• scripts/deploy.config.sh — DEFAULT_DATABRICKS_PROFILE (exported as
  DATABRICKS_CONFIG_PROFILE when set).
• scripts/deploy.sh — active profile shown in the deploy summary; auth hints mention
  --profile when configured.
• docs/deployment.md and README.md — document the new variable.

Lakebase connection diagnostics

Lakebase misconfiguration errors (wrong project/branch, stale db-… id, missing
database) now print actionable hints instead of terse failures.

• scripts/_lakebase-diag.sh (new) — shared helper printing common causes and
  ready-to-run databricks postgres list-databases / endpoints commands (with
  --profile when configured).
• scripts/deploy.sh — diagnostics on db-… resolution failure, resource checks,
  dry-run abort, bootstrap failure, and Lakebase step ERR-trap hints.
• scripts/bootstrap-lakebase-perms.sh — diagnostics on endpoint/credentials/psql
  connection failures; distinguishes psql connection errors from missing schema.

Lakebase underscore vs hyphen naming

Lakebase exposes two names per database — status.postgres_database (PG datname,
underscores) vs the resource-path database_id (RFC-1123, hyphens only). Operators
saw ontobricks_demo become ontobricks-demo in list-databases and deploy failed on
exact string match.

• scripts/_lakebase-resolve-db.py (new) — resolve db segment + datname with
  underscore/hyphen slug tolerance.
• scripts/deploy.sh — uses the shared resolver; auto-corrects LAKEBASE_DATABASE when
  slug-matched.
• scripts/_lakebase-diag.sh — documents the two-name model in failure hints.
• scripts/deploy.config.sh — comment that DEFAULT_LAKEBASE_DATABASE must come from
  postgres_database, not the hyphenated resource id.

---

Upgrade Notes

Upgrading from v0.6.0

No schema changes. Deploy the new app bundle as usual (make deploy). If you use a
non-default Databricks CLI profile, set DEFAULT_DATABRICKS_PROFILE in
scripts/deploy.config.sh once instead of exporting it per session.

If Lakebase deploy previously failed because your configured database name used
underscores while the API returned hyphens (or vice versa), v0.6.1 should resolve the
mismatch automatically. Ensure DEFAULT_LAKEBASE_DATABASE in deploy.config.sh matches
the Postgres postgres_database field from databricks postgres list-databases, not the
hyphenated resource-path tail.

Upgrading from v0.5.x

Upgrade to v0.6.0 first (see releases/ReleaseNotes_V0.6.0.md for migration steps),
then apply v0.6.1 on top.

---

Changes Summary

Area	Key files	Change type
Switch modal (read-only versions)	permissions.css, navbar.js, test_switch_domain_modal.py, docs/user-guide.md	Fix
Auto-Map SQL deduplication	agents/tools/mapping.py, SQLWizardService.py, test_mapping_tools.py	Fix
CLI profile support	deploy.config.sh, deploy.sh, docs/deployment.md, README.md	Enhancement
Lakebase diagnostics	_lakebase-diag.sh, deploy.sh, bootstrap-lakebase-perms.sh	Enhancement
Lakebase name resolver	_lakebase-resolve-db.py, deploy.sh, test_lakebase_resolve_db.py	Fix
Version	pyproject.toml	Bumped to 0.6.1

---

What is NOT changed

• External API contract — all /api/v1/ endpoints are backward-compatible.
• MCP tool contracts — no MCP tool signatures changed.
• Lakebase schema — no new tables or migrations.
• v0.6.0 features — collaboration, graph analytics, edit-lock, audit trail, and all other
  v0.6.0 capabilities remain intact.

---

This discussion was created from the release V0.6.1.