Fix/v0.5.2#77
Closed
benoitcayladbx wants to merge 68 commits into
Closed
Conversation
Bumps [codecov/codecov-action](https://github.com/codecov/codecov-action) from 6 to 7. - [Release notes](https://github.com/codecov/codecov-action/releases) - [Changelog](https://github.com/codecov/codecov-action/blob/main/CHANGELOG.md) - [Commits](codecov/codecov-action@v6...v7) --- updated-dependencies: - dependency-name: codecov/codecov-action dependency-version: '7' dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com>
Bumps [actions/setup-node](https://github.com/actions/setup-node) from 5 to 6. - [Release notes](https://github.com/actions/setup-node/releases) - [Commits](actions/setup-node@v5...v6) --- updated-dependencies: - dependency-name: actions/setup-node dependency-version: '6' dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com>
Bumps [actions/download-artifact](https://github.com/actions/download-artifact) from 5 to 8. - [Release notes](https://github.com/actions/download-artifact/releases) - [Commits](actions/download-artifact@v5...v8) --- updated-dependencies: - dependency-name: actions/download-artifact dependency-version: '8' dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com>
…ked-desp-DT Fix/submit for review blocked desp dt
_BuildPipeline._complete_task() — the single success exit shared by both the UI Build button (build_kind=session) and the REST endpoint (build_kind=api) — never called write_version() after completing a build. Only the scheduled-build path (scheduler._persist_domain_metadata) wrote the domain_versions.last_build column. ReviewService.submit() reads that column; the Consistency-checks panel reads live triple-store state — so both indicators on the Validation page disagreed and Submit for Review remained permanently blocked. Fix: add _persist_last_build_to_registry() called from _complete_task() after _record_build_run(). Best-effort (exceptions are logged, never propagated). Also stamps last_build when empty for the API build path. Backport of the same fix applied to fix/v0.5.1.
_BuildPipeline._complete_task() — the single success exit shared by both the UI Build button (build_kind=session) and the REST endpoint (build_kind=api) — never called write_version() after completing a build. Only the scheduled-build path (scheduler._persist_domain_metadata) wrote the domain_versions.last_build column. ReviewService.submit() reads that column; the Consistency-checks panel reads live triple-store state — so both indicators on the Validation page disagreed and Submit for Review remained permanently blocked. Fix: add _persist_last_build_to_registry() called from _complete_task() after _record_build_run(). Best-effort (exceptions are logged, never propagated). Also stamps last_build when empty for the API build path. Backport of the same fix applied to fix/v0.5.1 and develop.
Both files were present on master but absent from develop (dropped by the revert of PR #61). Synced back to keep the branches aligned.
…tory Resolves the visual noise in GitHub's compare view caused by the merge+revert of PR #61. Net code diff between master and develop remains the same (version bump, deploy config, one DatabricksAuth change, v0.6.0 changelogs). Kept develop's 0.6.0 version strings and deploy config; kept the v0.6.0 changelog file.
Drops the experimental route→plan→clarify→dispatch loop introduced in feat/Collaborative-Comments-Tasks (agent_task_runner, agent_task_planner, agent_task_router, agents/registry) before the v0.6 merge. - Deleted: 8 source files + 3 test files + 2 design docs - CommentService: removed AI-Agent assignee, kickoff block, start_agent_task call, _maybe_resume_agent - Mapping: removed apply_agent_mappings() - Frontend: removed all AI-Agent JS state/polling/UI and CSS rules - Docs: removed AI Agent task assignment section from architecture.md - Tests: 2132 passed (1 pre-existing lakebase skip)
Feature branch takes priority throughout. Resolved: - modify/delete (×4): kept CommentService.py, comments-panel.js, test_comment_service.py, test_discussion_panel_assets.py — develop never had them; they are new additions from this feature. - Restored 13 feature-branch files (API router, domain collaboration JS/CSS, templates, tests, upgrade scripts, changelogs) that were staged for deletion because develop's master-merge-back never included them. - Content conflicts (Mapping.py, review-modals.css, ontology-init.js, docs/architecture.md, changelog): kept feature branch version in all cases. Tests: 2350 passed, 15 skipped.
The git auto-merge silently took develop's version of 4 files, stripping all collaboration UI wiring: - base.html: lost comments-panel.js <script> include - domain.html: lost domain-collaboration.css, mytasks-section, discussions-section includes, and domain-collaboration.js script - menu_config.json: lost the Collaboration group (My Tasks + Discussions) - routers/internal/__init__.py: lost comments_router registration Restored all 4 files from the pre-merge feature commit (c2416cc). Tests: 2350 passed, 15 skipped.
Documents the silent auto-merge data-loss pattern discovered during PR #67 conflict resolution: git resolved base.html, domain.html, menu_config.json, and __init__.py without conflict markers while discarding the feature branch's collaboration UI wiring. Six rules added: feature-branch priority, auto-merge risk, diff checklist for the 4 most-at-risk files, restoration procedure, test gate, and commit message guidance.
…logy-map.js The auto-merge silently dropped the openOntologyDiscussion() function (~10 lines) from ontology-map.js. The function was defined at the pre-merge tip (c2416cc:1528) but absent in develop, and the merge result landed between the two versions without it. The Information-page Discussion button called it directly via onclick= with no guard, causing an uncaught ReferenceError on click.
The earlier conflict resolution auto-merged 50+ frontend files and silently stripped collaborative feature content from every template and JS file. Restored all affected files from the pre-merge feature tip (c2416cc): Templates (25): all ontology, mapping, and dtwin partials JS (9): ontology-map.js, domain-review.js, home-tasks.js, mapping-shared.js, mapping-entity/relationship-modal.js, query.js, query-sigmagraph.js, registry-my-tasks.js Backend (8): _build_pipeline.py, RegistryService.py, ReviewService.py, store/base.py, schema.sql, store.py, test_registry_store.py, test_review_service.py All discussion buttons (ontology×9, mapping×8, dtwin×7) verified present. openOntologyDiscussion() verified defined at ontology-map.js:1528. Tests: 2372 passed, 15 skipped, 4 pre-existing psycopg failures.
Exposed openDomainDiscussion() globally from domain-collaboration.js (delegates to resolveDomainContext + OntoComments.openThread, same pattern as openThreadFor but callable from onclick). Added the onto-discuss-btn to: - _domain_metadata.html (Data Sources) — alongside Reset + Add Data Source - _domain_documents.html (Document Management) — new header action
Previously the panel only appeared after async context resolution + data fetch, causing a noticeable blank delay on click. comments-panel.js: - Added showLoadingPanel() — opens the offcanvas right away with a spinner in the list area; exposed on window.OntoComments. - openThread() — skips show() if panel is already visible (no flicker). - openForSelection() — calls showLoadingPanel() first, then resolves domain context and fetches comments behind the open panel. domain-collaboration.js: - openDomainDiscussion() — calls OntoComments.showLoadingPanel() immediately before the async resolveDomainContext() call.
When 'Create task' is opened from a comment thread, a clipboard button appears next to the title input. Clicking it copies the comment's raw text (stripped of tags) into the title field, truncated to 200 chars with a trailing ellipsis (…) if longer. Implementation: - threadHtml: stores raw comment text in data-raw-text attribute - taskFormHtml(heading, withCancel, commentText): renders copy button with data-tk-copy-text when commentText is provided - wireTaskForm: wires copy button → fills [data-tk-title], focuses it - toggleTask: reads data-raw-text from thread element, passes to form - TASK_TITLE_MAX = 200 constant
…s-Tasks feat(collab): collaborative comments, tasks & discussions (domain-wide)
Add a checkbox column to the Attributes table in the Status tab of the entity and relationship panels. All attributes are included by default; unchecking marks an attribute as excluded — it is stripped from attribute_mappings on save and no longer flagged as an unmapped gap. - Frontend (mapping-design.js): EntityPanelState/RelPanelState gain excludedAttributes[]; Status-tab rows become interactive checkboxes with strikethrough styling for excluded attrs; save functions write excluded_attributes into the mapping object. - Backend (Mapping.py): build_entity_mapping / build_relationship_mapping propagate excluded_attributes; compute_mapping_gaps skips them.
When unchecking (excluding) an attribute, do not delete it from attributeMappings — the save function already filters exclusions. On re-check, read attributeMappings to restore the correct icon (green checkmark if the attribute was mapped, red cross if not).
…appings exist When ontology classes lack dataProperties (e.g. sync_class_data_properties was not applied to the stored config), entities with existing attribute_mappings still show the Attributes table by synthesizing attribute objects from the mapping keys. Applies to both entity and relationship panels, and to EntityPanelState/RelPanelState so the Mapping-tab dropdown also works.
Two-layer guard against the LLM inventing attribute_mappings for
columns that have no corresponding ontology data property:
1. Prompt (engine.py): explicit rules — if entity 'attributes' list is
empty, submit attribute_mappings:{} and select only ID+Label columns.
Never invent mappings for attributes not in the list.
2. Server-side (mapping.py / tool_submit_entity_mapping): filter
attribute_mappings against the entity's declared attributes from
ctx.ontology. If the entity has no attributes, all invented mappings
are discarded with a warning log. Ensures Invoice/Payment etc.
(no dataProperties) get a clean SELECT id AS ID, name AS Label query.
updateMappingCompletionStatus in mapping-information.js now skips excluded_attributes when computing totalAttributes/mappedAttributes, matching the Auto-Map page behaviour. - Attribute gauge percentage and count only cover included attributes. - Entity/relationship/attribute count spans show "· N excl." when items are excluded (same _excl() helper pattern as mapping-autoassign.js). - Entity detail-list badge shows attrMapped / includedProps.length and appends "· N excl." when the entity has excluded attributes. - isMapped now requires sql_query (stub-only mappings no longer count). 2418/2418 unit tests pass.
The Information page was counting from config→ontology direction without requiring m.sql_query, which could: 1. Include stub mappings (excluded_attributes only, no SQL) as "mapped" 2. Miss a property if the config m.property URI differed slightly from p.uri Fix: rewrite to use the same ontology→config direction as Auto-Map, requiring m.sql_query for both entity and relationship counts: assignedRelUris = config.filter(m => m.sql_query).map(m => m.property) mappedProperties = activeProperties.filter(p => assignedRelUris.has(p.uri)) 2418/2418 unit tests pass.
Add automatic retry with progressive backoff (5s, 10s, 20s) in _get and _post helpers for 502/503 responses, which occur when the OntoBricks app is sleeping or starting up on Databricks Apps.
…actions/setup-node-6 Bump actions/setup-node from 5 to 6
…codecov/codecov-action-7 Bump codecov/codecov-action from 6 to 7
…actions/download-artifact-8 Bump actions/download-artifact from 5 to 8
…actions/checkout-7 Bump actions/checkout from 6 to 7
Both _get and _post now retry up to 3 times on 502/503 with back-off 5s → 10s → 20s (~35s total), covering the typical Apps cold-start window. Each retry is logged at WARNING level. All other status codes still fail immediately. Also adds pytest/pytest-asyncio/anyio as dev dependencies to the mcp-server package so `uv run pytest` works out of the box.
…d-start fix, MCP retry - Mapping designer: add Exclude all / Include all toggle button (entity & rel panels) - Mapping reset: call _stampExcludedFlags() so excluded state clears on reset - Build pipeline: force=True on GlobalConfigService reads to prevent managed_synced being silently skipped after a cold-start cache miss - MCP server: retry with exponential backoff on 502/503 (Databricks App cold-start) - Docs: user-guide and lakebase-graphdb updated to reflect the above changes
docs(contributors): use full name for Hugues Journeau in contributors table
…exclude-include Feature/mapping attribute exclude include
UI-only label rename across the entire codebase: - Top-level nav section "Digital Twin" is now "Knowledge Graph" - Sigma.js viewer sub-item "Knowledge Graph" is now "Graph Viewer" Internal code identifiers (digitaltwin, dtwin, DigitalTwin, routes) unchanged. All 2418 unit tests pass.
…gating - Add L2 subnav (#obSubnav) with Domain / Ontology / Mapping / KG dropdowns, inline breadcrumb, and Save Domain button; hidden until a domain is loaded (hasDomain) - Breadcrumb moved into the L2 subnav row as a flex <li>; no longer a standalone <nav>; icons removed, italic 0.65rem style - Save Domain button relocated from Domain Information page to L2 subnav (#menuSaveDomain); click delegation moved to document.body so it fires outside #navbarNav - domainNew() replaced with showNewDomainDialog() popup (name + description + LLM endpoint) followed by automatic UC save dialog - L1 Domain link (#domainL1Link) disabled (.ob-nav-disabled) when no domain loaded; L2 tab toggles unlocked by hasDomain, not hasRegistry - updateMenusForDomainStatus() now driven by hasDomain so all domain sub-pages are accessible as soon as a domain is open in session - New Domain button added to home page All Domains section - Fix recursive_delete treating a 404 UC directory as an error, causing spurious "partially deleted" on domains with no UC files - Update .cursor/11-frontend-design.mdc with new two-level nav and breadcrumb architecture, gating rules, and component catalog entries
Feature/two level navbar
…ection - Add graph analytics pipeline: PageRank, Betweenness, Degree, Closeness, Clustering Coefficient via NetworkX; GraphBuilder/GraphMetrics services - Analytics UI: sidebar section, Chart.js charts (4 metrics), entity-type filter, PageRank detail table, metric info modals, clickable chart bars - AI graph interpreter: full agent lifecycle (SPEC, eval harness, tools), tool-enabled LLM with loopback get_entity_details, markdown insight display - Discussion / Audit Trail: add-to-audit-trail button, markdown rendering in audit timeline, Discussion panel button - Flat / time-series entity-type detection: EntityTypeProfile dataclass, single-pass heuristics (predicate diversity, temporal keywords), Data Model Health card in Analytics UI, flat_entity_types surfaced to AI prompt - Refactoring: URIHelpers.extract_local_name, deduplicated JS helpers, centralised GraphBuilder literal filtering, tighter error handling - Docs: user-guide Analytics section, features.md updates
- Add pytestmark = pytest.mark.unit to test_graph_metrics.py (--strict-markers) - Promote _local() closure to module-level _local_name() in GraphMetrics.py - Document Data Model Health card in docs/user-guide.md (Analytics section) - Append code-review entry to changelogs/v0.6.0
…trics Feature/graph analytics metrics
feat(mapping): Serialize R2RML in deterministic order
Remove the domain-design group (Ontology + Mapping quick-links) from menu_config.json; pages remain accessible via top-level navigation.
added 3 commits
June 24, 2026 09:52
- menu_config.json: rename "Validation" menu item label to "Publication" - _domain_review.html: rename page h4 title to "Domain Publication Workflow"
Add a dark pill style to the Domain nav item whenever a domain is loaded. Driven by `updateDomainMenuVisibility(hasDomain)` so it stays in sync with the existing domain-state logic on every page.
1 task
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Linked Issue / milestone
Closes #. Part of
M<n>.<phase>in.planning/ROADMAP.md.Plan
.planning/<slug>/PLAN.mdType of change
Author checklist
<type>(<scope>): <subject>).changelogs/<today>.logupdated with title + context + numbered changes + files + test result.uv run pytest tests/<scope>/green locally.pre-commit run --all-filesclean (or skipped hooks documented).src/agents/**or any MLflow-traced LLM path changed:SPEC.mdpresent in.planning/<slug>/.tests/eval/datasets/<agent>/dataset.jsonlhas ≥20 examples (new) or ≥10 (change).src/mcp-server/**changed:uv run pytest tests/mcp/ -m mcpgreen.gsd-*references re-introduced.MLflow eval run
Test plan
uv run pytest tests/ -m "not e2e and not property and not eval" --cov-fail-under=90scripts/check_coverage.py) green for touched packages.uv run pytest tests/mcp/integration/test_tool_schemas.pyincludes it.Reviewer hint
See
docs/PR_REVIEW_CHECKLIST.md. Numbered items map 1:1 to comments —#3: missing OntoBricksError subclass for new conditionis more useful than "fix error handling".