Skip to content

CHANGELOG.org

Latest commit

 

History

History
executable file
·
1558 lines (1367 loc) · 84.5 KB

CHANGELOG.org

File metadata and controls

executable file
·
1558 lines (1367 loc) · 84.5 KB
Raw

Changelog

[Unreleased]

Breaking: project renamed from Org-Supertag to Supertag

  • The repository, package entry, feature, current Lisp interfaces, Customization group, and documentation now use supertag exclusively.
  • Removed org-supertag.el, compatibility aliases, and the old Babel language names. Load the package with (require 'supertag) and use supertag-query-block.
  • The default data directory is now ~/.emacs.d/supertag/. It is not moved automatically: startup refuses the retired default path or an ambiguous pair of old/new paths before creating or loading a database.
  • See doc/MIGRATING-TO-SUPERTAG.md for the explicit upgrade checklist.

Data safety

  • New references now write one forward Org link in the source node only. Backlinks in Node/Table views are derived from the relation index; relation and node-reference field operations never insert reciprocal text into target files.
  • Existing ambiguous reciprocal links now have a confirmation-only migration: read-only preview, exact occurrence selection, a second confirmation, adjacent file snapshots, and automatic file/Store restoration on projection failure. Nothing is selected or deleted by default.

Breaking: database format version bump 5.0.0 -> 6.0.0

  • supertag-data-version is now 6.0.0. A review of the git-sync S2 canonical (line-per-entity) serialization found its “old versions can read the new format” claim false: a 5.9.x (or earlier) build’s loader does exactly one read of the database file, so against the line-per-entity format it only ever sees the first line and silently ends up with an empty-looking store – not a parse error. The format itself is unchanged (still frozen for git merging); this is an honest acknowledgment that reading it requires >= 6.0.0.
  • New one-time snapshot: the first canonical save over a still legacy-format on-disk database is preserved to backups/supertag-db-preformat6-<timestamp>.el before the atomic rename, covering the case a stale :version alone would miss (format changed on disk without the stored version being out of date). Like premigrate snapshots, never touched by daily-backup cleanup.
  • Every canonical save’s root line now also carries :supertag-format and a plain-language :incompatible-notice sentinel, and the format header comment embeds the current data version – so a raw glance at the file (or an old build’s own inspection tooling) has a chance of surfacing what happened instead of an inert, silently-empty plist.
  • M-x supertag-doctor now reports the on-disk format (legacy vs. canonical) and counts both snapshot kinds.
  • See README “Data storage” / “Migration from older versions” for the user-facing upgrade/downgrade notes.

View framework

  • Stream nodes are now grouped under one creation-day heading instead of repeating a timestamp on every row. Rows remain title #tags summaries with no body, file path, companion index, or underlined buttons.
  • Stream title navigation now preserves natural window scrolling instead of pinning every selected row to the top. The e edit buffer explicitly expands the source title and body; C-c C-c confirms the edit and C-c C-k restores the pre-edit text and aborts.
  • Source-backed node upserts now preserve an existing :created-at value, so editing a Stream node cannot silently move it to the end of creation order.
  • Added M-x supertag-view-stream: choose a tag to browse its nodes and all transitive :extends descendants as a creation-ordered, single-column title stream. n=/=p navigate, e opens the complete source-backed narrowed node, and v opens Node View for field changes. The View Runtime remains unchanged and no dependency or persistence format was added.
  • Progress Dashboard, Effort Distribution, and Priority Matrix now use the same View Runtime as Search, Table, Kanban, Node, and the Widget DSL.
  • The Widget DSL is now a production renderer backend with dynamic render trees, stable keyed selection, native text buttons/links, editable fields, and layout-safe interaction inside columns and cards. The three built-in dashboards now use this backend and no longer keep separate buffer renderer functions. No external package dependency was added.
  • Removed the legacy developer lifecycle (define-supertag-view, supertag-view--with-buffer, and supertag-view-render). Custom views now register with supertag-view-register and open through supertag-view-open; refresh has one Runtime-only path.

Bug fixes

  • Tag completion now shows full parent paths such as diary/happy in the candidate column, so every row shares the same left edge. Selecting the display path still inserts and stores the real Tag ID.
  • Fixed inline #tag completion waiting for Corfu’s generic two-character prefix threshold in normal Org buffers. Once the leading # identifies the Tag CAPF context, existing candidates can now appear immediately without changing the user’s global completion threshold.
  • Fixed background sync storing custom Org TODO keywords such as DOING in node titles. The shared current-buffer parser now initializes Org before parsing, and supertag-sync-full-rescan reparses unchanged files so existing bad :title, :todo, and :olp values can be rebuilt.
  • Fixed recurring org-element--cache parser warnings after embed sync. The embed writer now disables stale-cache reads while replacing node content, resets the target buffer’s cache even when the write fails, and no longer leaks suppressed modification hooks into global save callbacks or other open Org buffers.
  • Fixed supertag-sync-validate-nodes mass-orphaning legacy file nodes that predate persisted identity metadata. Current file nodes are validated by their own link type: Org-ID nodes match the top-level :ID: and Denote nodes match #+IDENTIFIER:. Legacy nodes without :link-type fall back to file existence, while deleted files and missing headings are still orphaned. Identity changes now retire the old file node instead of leaving two nodes pointing at one file.
  • Fixed Symbol's value as variable is void: --cl-block-...-- errors (seen from the async sync worker when destructive sync is disallowed): eight functions used cl-return-from inside a plain defun, which lacks the implicit cl-block — now cl-defun. Affected: supertag-sync--verify-file-nodes, supertag-sync--check-and-sync-legacy, supertag-sync--check-and-sync-guarded, supertag-sync-full-rescan, supertag-field-rename, supertag-field-remove, supertag-schema--cleanup-inherited-field-associations, and the restore-heading-ids script. A byte-compiled behavioral regression test locks down the guarded sync-worker path.
  • Fixed heading deletion cleanup being lost across restart when the same file also updated another node: deferred cleanup now leaves both the old content hash and mtime intact until destructive sync is allowed.
  • Fixed deferred cleanup becoming permanently stuck after an async worker error: complete snapshots re-enqueue deferred files and rely on the async queue’s existing deduplication instead of a stale :queued marker.
  • Fixed test/aggregate-test.el being unloadable: a stray paren closed test-aggregate-column-all-functions early, leaving unbalanced top-level forms. The repaired tests now run in the default suite.

[5.9.0] - 2026-07-12

Data safety (highlights)

Atomic database saves

  • supertag-save-store now writes to a same-directory temp file, optionally verifies it re-reads with the expected node count (new option supertag-db-verify-after-save, default t), then renames over the target. A crash mid-save can no longer truncate the database.
  • Fixed a defun-boundary bug that made the legacy field migration run at load (require) time.

Multi-instance protection

  • New option supertag-db-lock (default t): the database is advisory-locked after load. A second Emacs instance loads read-only with a clear warning and M-x supertag-db-retry-lock to take over once the first instance exits.

Automatic migration on load

  • New option supertag-db-auto-migrate (default t): out-of-date databases are migrated automatically after load, preceded by a timestamped supertag-db-premigrate-* snapshot that daily-backup retention never deletes. Read-only (locked) sessions skip migration.
  • Fixed version-only migrations never being persisted (they re-ran on every load).

Health check

  • New command M-x supertag-doctor: one-stop report over database files, guards, lock, version, integrity, and backups, with optional guided repairs.

Discoverability

  • New command M-x supertag-menu (transient): all major commands grouped and keyed in one place.
  • New command M-x supertag-setup: first-run wizard for sync directories, file-ID source, persistence, and the initial scan.
  • New command M-x supertag-automation-insert-template: nine ready-made automation rule templates with guided parameters.

Query language

  • Dynamic query blocks: #+BEGIN: supertag-query :query "..." re-render on C-c C-c / org-dblock-update. New command supertag-insert-query-dblock.
  • Both block forms honor :sort, :order, :limit, :columns.
  • New supertag-query-library: save/run/insert named queries, a guided query builder (supertag-query-build), and supertag-query-describe-syntax. Full reference in doc/QUERY.md.
  • Implemented recent-days / in-month / in-year operators that older docs advertised but the parser never supported.
  • Fixed absolute dates (YYYY-MM-DD) signaling “Invalid time specification” in after / before / between.

Fixes

  • cl-return-from inside plain defun in three supertag-ops-tag.el functions signaled void-variable on their early-exit paths.
  • Auto-migration whose save is deferred by a guard no longer re-runs and overwrites its premigrate snapshot on the next load.

Engineering

  • GitHub Actions CI (Emacs 28.2 / 29.4) running the full ERT suite; tests grew from 81 to 110.
  • All root-level test scratch files moved under test/; fixed a bare test/ .gitignore rule that had kept four suite files untracked.
  • Removed the unused gptel dependency and the abandoned archive/ code; renamed supertag-test.el to supertag-diagnostics.el.
  • New reproducible performance baseline (test/perf-benchmark.el): at 10k nodes the O(N) paths stay well inside interactive budgets.

Features (from community PRs and pre-release work)

File node reference / backlink support

  • supertag-add-reference now accepts file nodes (level-0 nodes representing entire files) as both source and target.
  • When a file node is the source, the forward link duplicate check scans the whole file instead of a subtree.
  • When a file node is the target, the reciprocal backlink is inserted at the marker position (after file metadata, as positioned by supertag-ui--find-node-marker).
  • New helper supertag-ui--find-node-marker provides unified node locating for both heading and file nodes, replacing scattered org-id-find calls.

Backlink timestamps

  • New option supertag-reference-backlink-include-timestamp. When set to t, creating a reference appends the current time as an inactive Org timestamp to the reciprocal backlink, e.g. [[id:abc][Title]] [2025-06-22 Sun 16:07]. Default is nil to preserve existing behavior.

Fixes

Table view

  • Parent headings now inherit theme-appropriate subdued styling instead of using a hard-coded dark background.
  • Suppress schema fields whose slug is refs so the reserved Refs column is not displayed twice.

Reference insertion cursor position

  • supertag-add-reference now saves the cursor as an Emacs marker instead of a raw integer position. This ensures the forward link is inserted at the original cursor location even when the reciprocal backlink is inserted earlier in the same file, which would otherwise shift subsequent text and leave the saved position stale.

[5.8.2] - 2026-03-05

Fixes

Persistence

  • supertag-save-store no longer raises user-error when invoked by the auto-save timer with an uninitialized store. Timer-driven calls now log a single Supertag auto-save skipped: ... message instead of spamming the echo area whenever the store origin is missing (e.g. after a half-finished initialization). Interactive calls still surface the full guard error.
  • supertag-sync-full-initialize now stamps supertag--store-origin and triggers an immediate save once the rescan completes, so users who run the full initializer without going through supertag-init end up with a persisted database and no recurring timer error.

Scheduler

  • The daily scheduler now respects the :days-of-week filter that was already accepted by supertag-scheduler-register-task but silently ignored at dispatch time.
  • A failing scheduled task no longer aborts the master scheduler tick; the error is logged and the remaining tasks keep running.

Automation

  • supertag-automation-create now preserves the :schedule plist on the stored rule. Previously it was silently dropped, which meant any :on-schedule rule was registered with :time nil and the scheduler rejected it with “Invalid time format” — so scheduled automations had never actually fired in this release line. Added test-automation-scheduled.el as a runnable batch self-check.

#tag Completion

  • Made the new-tag candidate survive orderless enumeration. orderless enumerates the candidate set by calling the CAPF table with action=t and STR””, then filters with its own regexp. Our CAPF gated the new-tag candidate on the passed-in =STR being non-empty, so when orderless asked with STR””= the candidate was absent from the enumerated set — orderless could not “reveal” what it never saw, and no user-visible [Create New Tag] entry ever appeared in the popup. This was the root cause behind every “I still don’t see it” report. Switched the action=t branch to enumerate based on the captured prefix (the live user input that triggered the CAPF), so the candidate is always present in the enumerated set and orderless filters from there.
  • Stopped the “[Create New Tag]” popup entry from being silently auto-inserted into the buffer when it is the only surviving candidate. The CAPF table’s `try-completion’ branch used to return the full labeled candidate string (“PREFIX [Create New Tag]”), and corfu — especially with `corfu-preview-current’ set to `’insert’ — treated that as “already complete” and committed the suffix to the buffer (often as the lowercase residue “[Create new tag]”) without ever showing the popup. The CAPF now detects when the longest common prefix contains the new-tag marker and returns the user’s raw input instead, forcing the popup to appear.
  • `test-completion’ branch hardened the same way: a sole new-tag candidate no longer reports the input as “exactly complete”, so `completion-at-point’ will not skip the popup either.
  • Made the “create new tag” option actually visible in the popup across UI/filter combinations. Two earlier shapes failed:
    • [Create New Tag: <name>] candidate: orderless filtered it out because the leading [ and trailing ] did not match user input;
    • bare <name> candidate with metadata affixation-function: affixation is not honored by every corfu version / marginalia setup, so users saw no label at all.

    Switched to a self-labeled candidate: the candidate string itself is <name> [Create New Tag], carrying is-new-tag and new-tag-name text properties. Because the candidate starts with the user’s prefix, orderless/basic/substring matchers all keep it visible. post-completion-action strips the suffix back off the buffer after commit, leaving only the bare tag name.

  • Added supertag-tag-insert as a UI-stack-agnostic fallback: M-x supertag-tag-insert reads the tag via the minibuffer with completion against existing tags, and a freshly typed name creates the tag immediately. Use when the popup completion misbehaves with exotic corfu/cape/marginalia configurations.
  • Added supertag-completion-debug for popup diagnosis. Run it with the cursor right after a #prefix to dump bounds, live prefix, raw candidate list, post-filter candidate list, candidate width, and the active capf chain to *Messages*.
  • Typing #newtag and continuing to write (without explicitly selecting the popup candidate) still records the tag immediately via a buffer-local post-self-insert-hook. corfu/company quit their popup silently in that case and never call the CAPF :exit-function, so the tag used to be lost until the user re-triggered completion with C-M-i. The new hook fires only when the just-inserted character ends a #tag context, and skips tags already on the node.
  • Loosened the CAPF :exit-function status guard to accept any successful exit code (everything except unknown), so default completion-at-point paths that report nil also commit.
  • Added test-completion-newtag.el covering candidate shape, affixation label, the four exit-status paths, the ignored-candidate hook path, the prose-no-tag path, and the duplicate-tag path.

Sync

  • supertag-sync-save-state now creates its parent directory before writing sync-state.el, fixing a file-missing error on first-time setup when ~/.emacs.d/org-supertag/ did not yet exist. Users no longer need to create the directory manually before running supertag-sync-full-initialize.
  • Demoted the per-file “Skip sync” / “Enqueued external modification” messages to fire only under supertag-sync-smart-detection-verbose, reducing echo-area noise during normal editing.

#tag Completion

  • Reworked inline #tag completion: the “create new tag” candidate is now the typed text itself (propertized) instead of a [Create New Tag] placeholder. This keeps it visible under corfu/orderless filtering and lets the UI insert the real tag name directly.
  • Completion candidates now track the live input string rather than the prefix captured when completion started, so the new-tag candidate updates as you type under corfu’s incremental filtering.
  • supertag-completion--post-completion-action no longer rewrites the buffer after insertion (the inserted text is already the real tag name), removing the fragile delete-region/re-insert step.

[5.8.1] - 2026-03-04

Fixes

Board UI

  • Fixed stale tag field values in board cards: Emacs now pushes fresh board-data directly to the frontend on any store change, replacing the fragile store-changed → open-board round-trip that could silently drop refresh signals.
  • Added supertag-board--current-board-id to track the open board on the Emacs side; cleared on WebSocket disconnect and board deletion.
  • Removed orphaned supertag-board-open autoload from org-supertag.el (function was never implemented; caused autoload errors on startup).
  • Replaced Chinese toggle labels (“展开内容” / “收起内容”) in board node cards with English (“Expand” / “Collapse”).

[5.8.0] - 2026-03-03

Features

Board (Whiteboard) Subsystem

  • Added ext/board-ui/: React Flow based whiteboard UI (Next.js).
    • BoardCanvas, BoardNode, BoardGroup, FloatingEdge components.
    • NodePalette for drag-and-drop node creation.
    • EdgeRelationMenu for edge type selection.
    • Toolbar with zoom/layout controls.
    • Zustand store for board state management.
  • New supertag-board.el: board data model and rendering in Emacs.
  • New supertag-board-ops.el: board CRUD operations.
  • New supertag-graph-ui.el: graph visualization bridge.
  • Added :boards collection to core store schema.

Force-Directed Graph Visualization

  • Added ext/graph-ui/: org-roam-ui inspired graph visualization frontend.
    • Force-directed graph layout with d3-force via react-force-graph.
    • Sidebar with node preview, backlinks, and tag bar.
    • Tweaks panel: physics, visual, filter, behavior settings.
    • Theme support (light/dark) with customizable colors.
    • Tag-based node coloring and filtering.
    • WebSocket integration for real-time Emacs communication.
    • Org content rendering via uniorg parser.

Semantic Relation Types

  • Added supertag-relation-type-register for user-defined semantic relation types.

Performance

In-Memory Relation Indexes (O(N) → O(1))

  • New supertag-core-index.el: secondary hash-table indexes over :relations collection.
    • Two indexes: from-id → {relation-ids} and to-id → {relation-ids}.
    • Maintained incrementally on create/update/delete; full rebuild on store load.
  • supertag-relation-find-by-from, supertag-relation-find-by-to, supertag-relation-find-between now O(k) instead of O(N) full-table scans.
  • supertag-relation-delete-for-node, supertag-relation-delete-for-tag, supertag-relation-get-database-relations also accelerated.
  • Benchmark (1000 iterations avg):
    RelationsScan (μs)Index (μs)Speedup
    500900.5~180x
    5,0007200.7~1,100x
    20,0002,8800.7~4,000x
    50,0007,3000.8~10,000x
  • Index rebuild cost: 0.1ms (500 rels) to 62ms (50k rels), once on load.

Fixes

  • Fixed grammar error in codebase.

[5.7.0] - 2026-02-10

Features

Virtual Columns & Views

  • Added virtual column runtime and built-in virtual views.
  • Table view now supports virtual columns and force refresh.
  • Core now loads dashboard and view framework modules.

Schema / Field Editing

  • Added interactive global-field editor with defaults.
  • Improved schema-view keymap layout and field edit workflows.

Automation

  • Added multi-tag predicates and a `move-node` action.

Fixes

Node / Capture / Relations

  • Guarded node-view field count against nil values.
  • Kept capture `batch-entries` within `let*` scope.
  • Hardened relation lookups against nil entries.
  • Propagated `add-reference` failure reasons.
  • Refreshed node hash when OLP changes.

Service Org Safety

  • Hardened node move extraction and safety checks.

Docs

  • Added view framework and virtual columns guides.
  • Clarified automation `:condition` semantics and quoting.
  • Recorded issue007 resolution commit.

Chore

  • Ignored/untracked local beads and phrase artifacts.
  • Added agent-relay Gemini snippet lock metadata.

[5.6.3] - 2026-01-21

Improvements

Sync

  • Smart sync detection is now always on; content-hash checks skip unchanged files and prevent accidental database clears from mtime-only touches.

Docs

Sync

  • Documented smart scan behavior and safety note in README and README_CN.

[5.6.2] - 2026-01-16

Improvements

Persistence / Vault Safety

  • Hardened DB loading with candidate fallback and `read-circle` support; record store-origin metadata for diagnostics.
  • Added safeguards to refuse unsafe saves when runtime persistence configuration appears inconsistent.
  • Added a runtime config guard to prevent manual changes to vault persistence variables; use vault activation APIs for switching.

Automation

  • Added trigger matching and opt-in commit-hook integration to avoid double-triggering; improved verbose logging.

References / View

  • Added reference field ownership helpers to keep node-reference fields consistent with :reference relations.

Tests

Persistence

  • Added ERT coverage for load-store fallback and `print-circle` scenarios.

Docs

Architecture

  • Added an ontology architecture overview (data/logic/behavior) and clarified trigger spelling pitfalls in automation docs.

[5.6.1] - 2025-12-26

Improvements

Sync Integrity

  • Added snapshot guard (availability/completeness) to block destructive sync when directories are unavailable or partial.
  • Deferred orphan verification and GC remember pending files; full rescan now aborts if snapshot is not complete.
  • Bound sync-state to the active data directory (per-vault) and auto-reload on data-dir changes to prevent state mismatch.
  • Added `supertag-sync-snapshot-guard` toggle.

View/Fields Resilience

  • Resolve node tags via relations with fallback to node `:tags`, improving field rendering when relations are missing.
  • Node view refreshes on `:field-values` change events.
  • Table view pins a “Refs” node-reference field at the end (auto-created/associated in global-field mode) so references can be edited in-table; it merges `add-reference` relations into display/selection; `C-o` jumps to referenced nodes.

Bug Fixes

Inline Tag Parsing

  • Broadened inline `#tag` matching to support non-ASCII and slash-containing tags across parsing, capture, and migration.

[5.6.0] - 2025-12-17

Features

View Data API and Plugin System

  • Introduced `supertag-view-api.el` as a unified, UI-agnostic data access layer for all views.
  • Established a formal View Plugin system with comprehensive documentation (`ORG-SUPERTAG-PLUGIN-GUIDE.md`).
  • Added a complete View Plugin demo: `supertag-view-demo-dashboard.el` (Dashboard UI).
  • Added Schema Registry to support custom entity type registration at startup.

Improvements

Architecture

  • Decoupled `supertag-services-ui.el` and all view modules (`view-table`, `view-node`, `view-kanban`, `view-schema`) to use the new stable Data API.
  • Clarified plugin architecture: View Plugins are now the primary extension point.

[5.5.0] - 2025-12-17

Features

Smart Sync Detection

  • Implemented content-hash based change detection to reduce unnecessary parsing.
  • Sync now only parses files if `mtime` changes AND content hash differs.
  • Added automatic sync state reconciliation when `org-supertag-sync-directories` changes.

Improvements

Performance & Observability

  • Optimized buffer usage in sync to avoid double-reading files.
  • Added verbose logging (`supertag-sync-smart-detection-verbose`) and decision debugging.

[5.4.0] - 2025-12-16

Features

Queue-Based Sync & Auto-Maintenance

  • Introduced file-level queue system for sync operations (default for full rescan).
  • Implemented configurable automatic database maintenance (validation + GC) triggered by sync ticks.

Improvements

Safety & Reliability

  • Enabled “Safe Auto Mode” by default for sync.
  • Fixed `supertag-sync-check-state` compatibility with plist-style state entries.

[5.3.1] - 2025-12-16

Features

Extractor Plugin Pipeline

  • Refactored node parsing to use an extensible Extractor Plugin system.
  • Built-in extractors now handle tags, properties, content, and OLP parsing.

Improvements

Architecture

  • Moved parsing logic into registered extractors to allow future extensibility.

[5.3.0] - 2025-12-14

Features

Property to Field Migration System

  • Added comprehensive Org properties to structured fields migration system
  • Introduced `supertag-migration-add-ids-to-org-headings` command to add :ID: properties to all Org headings
  • Enhanced `supertag-convert-properties-to-field` with automatic tag insertion in Org files
  • Added `supertag-batch-convert-properties-to-fields` for batch property conversion
  • Implemented plist format compatibility for property parsing and storage
  • Added automatic tag insertion in Org headlines during property conversion
  • Improved property collection with format-agnostic parsing (alist/plist support)

Code Cleanup and Legacy Removal

  • Removed deprecated `supertag-import-fields-from-properties` and related import functions
  • Cleaned up 300+ lines of obsolete import code from `supertag-service-org.el`
  • Streamlined codebase by removing redundant property import functionality

Improvements

Enhanced Property Format Compatibility

  • Fixed plist/alist format inconsistencies in property parsing
  • Updated `supertag–parse-properties` to consistently return plist format
  • Improved `supertag-node-hash` with format-agnostic property handling
  • Enhanced automation sync with proper plist-member usage

Migration Workflow Optimization

  • Optimized property-to-field conversion workflow for both Org-roam and non-Org-roam users
  • Added intelligent tag insertion that respects Org file structure
  • Improved error handling and user feedback during migration
  • Added comprehensive documentation for migration paths

Bug Fixes

Fixed Property Conversion Issues

  • Resolved “No org properties found” error in property conversion
  • Fixed plist-member parameter order in automation sync
  • Corrected property format detection in migration functions
  • Fixed tag insertion logic to prevent duplicate tags

Documentation

Added Migration Documentation

  • Comprehensive English and Chinese documentation for property-to-field migration
  • Step-by-step guides for different user scenarios (with/without Org-roam)
  • Before/after examples showing migration benefits
  • Batch conversion workflow documentation

[5.2.0] - 2025-12-01

Features

Global Field Model: Fields as First-Class Entities

  • Introduced an opt-in global field model guarded by `supertag-use-global-fields`, making fields independent entities instead of being nested under individual tags.
  • Extended the core store (`supertag-core-store.el`) with dedicated collections:
    • `:field-definitions` – global field definitions keyed by field-id.
    • `:tag-field-associations` – per-tag ordered lists of associated field-ids.
    • `:field-values` – node → field-id → value flat storage for all field values.
  • Added `supertag-core-schema.el` as a light-weight schema module:
    • Provides `supertag-sanitize-field-id` to normalize field ids (name → slug).
    • Maintains global field caches (`supertag–global-field-cache`) and tag→field-order caches, with `supertag-schema-rebuild-global-field-caches` and `supertag–maybe-rebuild-global-field-caches`.

Ops Layer Refactor for Global Fields

  • `supertag-ops-field.el`:
    • `supertag-field-set` / `supertag-field-set-many` now route to `:field-values` when `supertag-use-global-fields` is non-nil, using sanitized global field ids.
    • Maintains compatibility with legacy nested `:fields` storage when the flag is off.
  • `supertag-ops-tag.el`:
    • Tag field definitions now bind to global field ids; associations are stored in `:tag-field-associations` with explicit ordering.
    • Field CRUD on tags updates both legacy definitions and global associations where appropriate.
  • `supertag-ops-node.el`:
    • Tag add/remove on a node initializes and prunes corresponding global field value entries in `:field-values`.
    • Node operations are now aware of global field storage, ensuring tag operations and field values stay consistent.

Migration: Legacy Tag-Scoped Fields → Global Fields

  • Extended `supertag-migration.el` with a global field migration pipeline:
    • Deduplicates per-tag field definitions into global field definitions in `:field-definitions`.
    • Builds tag↔field association lists (with order) into `:tag-field-associations`.
    • Rewrites node-level field values from nested `:fields` into flat `:field-values`, with basic conflict handling and idempotent behavior.
  • Added dry-run and stats/reporting helpers to summarize created fields, associations, migrated values, skipped entries, and conflicts.

Improvements

Automation – Global Field Integration & Field-Centric DSL

  • Unified global field changes with the automation engine:
    • `supertag-core-store.el` emits `:store-changed` events on `(:field-values node-id field-id)` when global values change.
    • `supertag-automation.el` passes the full path as metadata into `supertag-automation-sync-handle-event`, allowing sync handlers to recover `(node-id field-id)` reliably.
    • `supertag-automation-sync.el` consumes commit `:path` metadata and handles `:field-values` events via `supertag-automation-sync–handle-global-field-value-event` / `supertag-automation-sync–process-global-field-change`.
    • `supertag-field-set` (global branch) explicitly calls into the sync pipeline so UI updates that bypass `supertag-ops-commit` also trigger automation.
  • Extended the automation condition DSL for the global field model:
    • Added `global-field-equals`, `global-field-changed`, and `global-field-test` so rules can explicitly target global fields.
    • Updated `field-equals` / `field-changed` to:
      • Prefer global field storage (via sanitized field ids and `:field-values`) when `supertag-use-global-fields` is non-nil.
      • Fall back to legacy nested fields when global fields are disabled.
    • Enhanced `supertag–extract-trigger-sources` and the rule index so both legacy property keys (e.g. `:status`) and normalized global field ids (e.g. `”status”`) are indexed, allowing `:field-values` events to resolve matching rules in O(1).
  • Tests (`dev-note/automation-global-field-tests.el`):
    • Added coverage for global field conditions (`global-field-equals`, `global-field-changed`, `global-field-test`) and their interaction with `:field-values` events.
    • Introduced an integration test verifying that both `supertag-store-put-field-value` and `supertag-field-set` fire automation rules correctly when `supertag-use-global-fields` is enabled.

Org / Capture / Services Integration with Global Fields

  • `supertag-service-org.el`:
    • Import/export of Org `:PROPERTIES:` drawers is now aware of global fields:
      • Maps Org properties to global field ids according to tag associations and global field definitions.
      • Writes imported values into `:field-values` instead of legacy nested `:fields` when global fields are enabled.
    • Ensures capture and sync flows treat global fields as the single source of truth for node-level field values.
  • `supertag-services-capture.el`:
    • Capture templates and capture flows now read/write node fields via global field APIs when the flag is on, removing direct dependency on tag-scoped storage.
  • `supertag-services-query.el`:
    • Query filters that reference field names are resolved against global field ids and `:field-values` first, with legacy fallbacks only when global fields are disabled.

UI & Views – Global Field–Aware Rendering

  • `supertag-view-table.el`:
    • Column construction now deduplicates fields using global field slugs when `supertag-use-global-fields` is enabled, preventing duplicate columns when multiple tags share the same global field.
    • Column keys and exported metadata carry `:field-id` for smooth round-tripping with global field storage.
    • Cell editing (`supertag-view-table-edit-cell`) sets field values via `supertag-field-set`, which in global mode writes to `:field-values` and triggers automation.
    • Title column can now optionally render node hierarchy context: the main title on the first line and the direct parent title on the second line using a low-contrast face, while still preserving table alignment.
    • Org-style links (e.g., `Description`) inside title cells are rendered as clickable `Description` labels using the existing org-link rendering helper.
  • `supertag-view-node.el`:
    • Field rendering and in-place editing (`supertag-view-node–edit-field-value`) now resolve the effective field set using global field associations and read/write via `supertag-field-set` in global mode.
  • `supertag-view-schema.el`:
    • Schema view reads definitions from `:field-definitions` and tag associations, showing which fields are global and how they are associated to tags.
    • Adds debugging helpers to inspect raw tag data, legacy field definitions, global field associations, and the resolved schema.
  • `supertag-view-kanban.el`:
    • Kanban groupings and columns take global field values into account when determining grouping and display, avoiding duplicates from shared fields.

Documentation – Global Field Model & Field-Centric Rules

  • Existing global field design/migration docs are complemented by updates to the automation guide:
    • `doc/AUTOMATION-SYSTEM-GUIDE.md` / `_cn.md`:
      • Updated the `contact-tier-frequency` example to use `field-changed` instead of the older `property-changed` string form, aligning with the global field DSL.
      • Added a “Field-Centric Rules (Global Fields)” section explaining how `field-equals` / `field-changed` behave when `supertag-use-global-fields` is enabled, including examples of rules that are driven purely by field changes without tag predicates.

Migration & Compatibility Notes

  • The global field model is gated by `supertag-use-global-fields` (default `nil`) to avoid breaking existing setups:
    • When `nil`, behavior stays compatible with the legacy tag-scoped field model and nested `:fields` storage.
    • When non-nil, all new field operations, UI interactions, and services use `:field-definitions`, `:tag-field-associations`, and `:field-values` as the source of truth, with legacy paths used only for reading or migration.
  • A dedicated migration path in `supertag-migration.el` supports one-shot conversion from tag-scoped fields to the global model; see `GLOBAL-FIELD-MIGRATION-GUIDE.md` for background, ordering, and risk notes.

[5.1.8] - 2025-11-22

Improvements

Reference and Backlink Consistency

  • `supertag-add-reference` no longer blindly inserts a forward `id:…` link when creating a `:reference` relation. It now scans the current node’s entire subtree (including the headline) and only inserts a forward link if none already exists, preventing duplicate forward links when the user has manually added a link.
  • Added `supertag-fix-missing-backlinks` in `scripts/check-reference-integrity.el`, which uses the results of `supertag-check-reference-integrity` to automatically insert missing backlinks under the target node’s content region whenever a `:reference` relation exists in the database but the corresponding backlink is absent in the Org file.

[5.1.7] - 2025-11-18

Features

Field Export to Org Properties with Progress UI

  • Added `supertag-export-all-fields-to-properties` command to export all database field values to Org `:PROPERTIES:` drawers for nodes that have fields.
  • Field names are exported as human-friendly Org property keys (e.g. `rate` → `:RATE:`, `who` → `:WHO:`) while keeping the database as the single source of truth.
  • Node-reference fields are exported as readable titles (e.g. director names) instead of raw node IDs, improving interoperability with external tools that only see Org files.
  • The export process shows a minibuffer progress bar and a final summary message, and can optionally save modified buffers when called with a prefix argument.

[5.1.6] - 2025-11-17

Improvements

Org-Capture Integration as a First-Class Capture Path

  • Added a unified finalization API `supertag-capture-finalize-node-at-point` that turns any Org heading into a Supertag node, ensuring a stable `ID`, syncing to the database, and applying tag fields in one place.
  • `supertag-capture-with-template` and the new org-capture integration now both rely on this core API, avoiding duplicated sync logic.
  • Introduced `supertag-org-capture-after-finalize` hook integration:
    • Templates can opt in with `:supertag t` in `org-capture-templates`.
    • Optional `:supertag-template` lets org-capture templates pre-fill tag fields using the Tag/Field/Value model.

Supertag-Aware Tag Prompt in org-capture

  • New `:supertag-tags-prompt` option for org-capture templates:
    • After capture finalization, prompts `Supertag tags (comma separated):` with completion from existing Supertag tags.
    • Supports creating new tags on the fly: names are sanitized and missing tags are automatically created as Tag entities.
    • Selected tags are written both to the Supertag database and to the Org headline as inline `#tag`, then re-synced in one pass.

Movement Workflow: org-capture + Supertag Move

  • Extended `:supertag-move` options to chain capture → finalize → move:
    • `t` / `node` → run `supertag-move-node` for full file+position selection.
    • `link` / `:link` → run `supertag-move-node-and-link`, moving the node and leaving a backlink at the original location.
    • `within-target` / `:within-target` → only select a position within the capture target file, skipping the file prompt for a smoother “in-file re-positioning” flow.
  • This replaces the need to use `org-refile` for Supertag nodes, while staying compatible with org-capture’s templating model.

Capture DSL and org-capture Symmetry

  • Clarified the roles of:
    • `supertag-capture-with-template`: advanced DSL for Supertag-native capture flows.
    • org-capture integration: lightweight path that reuses existing `org-capture-templates` and adds Supertag-specific post-processing via `:supertag`, `:supertag-template`, `:supertag-tags-prompt`, and `:supertag-move`.
  • Updated `doc/CAPTURE-GUIDE.md` and `doc/CAPTURE-GUIDE_cn.md`:
    • Org-capture based capture is now documented as a first-class, recommended workflow for existing Org users.
    • Added concrete examples for `:supertag-template`, `:supertag-tags-prompt`, and `:supertag-move within-target`, and a developer section for the core finalization API.

Implementation Notes

  • Files touched: `supertag-services-capture.el`, `supertag-ui-commands.el`, `doc/CAPTURE-GUIDE.md`, `doc/CAPTURE-GUIDE_cn.md`.
  • `supertag-org-capture-after-finalize` now normalizes `:supertag-move` values from quoted and string forms and safely checks for the presence of move commands (`supertag-move-node`, `supertag-move-node-and-link`) before calling them.

[5.1.5] - 2025-11-15

Improvements

Node View ergonomics and safety

  • Auto-focus: When switching to the Node View window, the cursor automatically jumps to the value column of the first Field, reducing extra movement.
  • Keybindings restored: j/k, n/p, and page navigation keys now use their defaults, and no longer force-align to the value column when moving.
  • More robust context detection: Only calls Org API inside Org buffers, avoiding errors like “Point must be at an Org heading.”
  • Docstring corrections: Removed unescaped quotes that could cause void-variable errors, preventing `Point` from being parsed as a variable.
  • Non-destructive cleanup: Rendering no longer incorrectly deletes :node-tag relations; if a tag is unavailable, a “Not found (skipped cleanup)” message is shown instead.

Node/Table View unification and reference management

  • Unified reference creation entry point: `supertag-add-reference`, the `:node-reference` field in Node View, the `:node-reference` field in Table View, and sync-layer `:ref-to` parsing now all use `supertag-relation-add-reference` to create `:reference` relationships and automatically insert/deduplicate backlinks.
  • Shared data layer between Node View and Table View: Added `supertag-view-build-node-state` to construct a node’s general view state (fields, references, etc.) for use by different layouts.
  • New `node-detail` layout in View Table: `supertag-view-table–build-node-detail-state` + `supertag-view-table–render-node-detail` now support rendering a single node card view through the table engine.
  • Node View rendering prefers delegating to the View Table’s `node-detail` layout; if View Table is not loaded, safely falls back to a local renderer, achieving the structure where “view-node depends on view-table, and view-table only optionally enhances view-node.”

Node reference field display improvements

  • In Node View, `:node-reference` fields now show the referenced node’s title(s) instead of just raw IDs (multiple references are comma-separated).
  • When the node-reference value is empty, shows a gray italic “[Empty]” prompt, consistent with other fields.

References section is more readable and clickable

  • Each line uses text properties to bind RET/mouse click to jump to the corresponding node, implementing reference lists that “look like plain text but act like links.”

Data Integrity

When cleaning tags, also clear field values (globally consistent)

  • When a node has a tag removed, automatically clears all field values under that tag on the node, preventing “old values” from being revived and re-triggering automations.
  • Changes: Field cleanup is now always done inside `supertag-node-remove-tag`; the `supertag-change-tag-at-point` path also does field cleanup (redundantly but safely).

Reliability

Fixed recursive require error caused by cyclic dependency

  • Resolved: “Recursive ‘require’ for feature ‘supertag-core-scan’” error:
    • Removed hard dependency of ops-node on ops-field, replaced with forward declaration, and read field definitions directly from the :tags set.
    • Before calling field deletion, checks `fboundp` to avoid cycles due to load order.

Automation

Add-tag insertion location returns to “append at end of line”

  • Automated add-tag actions revert to the original behavior: if the headline does not contain the tag, append ` #tag` to the end of the line and save.
  • Dropped use of view-helper’s “intelligent” insertion location logic to avoid experience differences.

Implementation Notes

  • Files touched: `supertag-view-node.el`, `supertag-ops-node.el`, `supertag-ui-commands.el`, `supertag-automation.el`。
  • Added window focus hook: enabled on Emacs versions that support `window-selection-change-functions`, with graceful fallback on others.

[5.1.4] - 2025-11-11

Improvements

Enhanced Node View with Popup Interface and Evil Mode Support

  • Replaced side-window Node View with a stable popup child frame (posframe) for better focus management
  • Ensured proper cursor visibility and focus control in the popup window
  • Added Evil mode compatibility:
    • Disabled evil-local-mode in the popup to prevent key conflicts
    • Forced Emacs state when Evil is present
    • Set initial state to emacs for consistent behavior
  • Improved editing workflow:
    • Jump directly to first field’s value column on open (not header or last line)
    • After updating a field, restore cursor to the same field/value with preserved column offset
    • On cancel (C-g), return focus to popup and restore cursor position
  • Enhanced field targeting with text properties (:type :field-value) for precise cursor placement
  • Added supertag-value-column property to rendered value segments for accurate positioning
  • Implemented fallback behavior: return to buffer start when no fields exist

Implementation Details

  • Files modified: supertag-view-node.el, supertag-view-helper.el
  • posframe-show configured with override parameters for cursor visibility and focus acceptance
  • Focus/positioning fixes applied after popup window selection to avoid focus churn

[5.1.3] - 2025-11-11

Improvements

Safe Auto-Start for Sync and Enhanced Data Protection

  • Implemented delayed and retry-based auto-start of sync service
    • New defcustom: supertag-sync-auto-start (default: t)
    • New defcustom: supertag-sync-auto-start-initial-delay (default: 3s)
    • New defcustom: supertag-sync-auto-start-retry-interval (default: 5s)
    • New defcustom: supertag-sync-auto-start-max-retries (default: 24)
    • Waits for all org-supertag-sync-directories to exist before starting
    • Prevents early-start races when directories (cloud drives, mounts) aren’t ready yet

Enhanced Persistence Protection

  • Added safeguard in supertag-save-store to prevent data loss
  • Skips saving when in-memory node count is 0 AND on-disk DB appears non-trivial (>1KB)
  • Logs protective skip message to alert users of potential issue
  • Prevents accidental overwrite of valuable on-disk data with empty in-memory state

Orphan Node Garbage Collection Guardrails

  • New defcustom: supertag-sync-orphan-grace-seconds (default: 3600)
    • Orphan nodes marked with :orphaned-at timestamp
    • Only deleted after grace period expires
    • :orphaned-at cleared when node reappears in a file
  • New defcustom: supertag-sync-max-delete-ratio (default: 0.5)
    • Limits mass deletions to 50% of total nodes
  • New defcustom: supertag-sync-max-delete-count (default: 1000)
    • Hard cap on number of nodes deleted in one GC operation
  • Garbage collector aborts with warning when limits exceeded

Rationale

  • Prevents transient environment issues (unmounted drives, delayed cloud sync) from causing mass data loss
  • Protects against scenario where early sync misclassifies many nodes as orphans
  • Provides conservative, configurable guardrails for safe automated cleanup

Backwards Compatibility

  • Auto-start enabled by default but safe with retry mechanism
  • All GC guardrails are configurable via defcustoms
  • No breaking API changes

[5.1.2] - 2025-11-02

Improvements

Enhanced Tag Inheritance Management and UI Clarity

  • Clarified tag inheritance model: Explicitly defined `org-supertag`’s schematic field inheritance, distinct from Org-mode’s structural inheritance.
  • Improved Schema View UI: `supertag-schema–render-tag-node` now visually groups fields by origin (own vs. inherited), making inheritance explicit.
  • Renamed functions for clarity: `supertag-set-child` to `supertag-set-tag-parent` (and `supertag–set-parent` to `supertag–set-tag-parent`) to better reflect tag inheritance management.
  • Simplified Field Editing Workflow:
    • `supertag-schema–rename-field-at-point`: Prevents renaming inherited fields, guiding users to edit at the source.
    • `supertag-schema–edit-field-definition-at-point`: For inherited fields, offers “Jump to definition” (via `supertag-schema–goto-tag`). For local fields, retains full editing.
    • `supertag-schema–delete-at-point`: Prevents deleting inherited fields from child tags, guiding users to delete from the parent.
    • Added `supertag-schema–goto-tag` helper for jumping to tag definitions.

[5.1.1] - 2025-09-12

Improvements

Enhanced Schema View User Experience

  • Redesigned shortcut system:
    • Use n/p (and j/k) for up/down navigation
    • ‘a’ for “add field”
    • Capital ‘A’ for “add new tag”
    • Added ‘q’ key for quitting the window
  • Added help documentation:
    • Added footer at the bottom of the view, clearly listing all available shortcuts and their functions
    • Implemented using supertag-view-helper-insert-simple-footer function
  • Optimized cursor position:
    • Modified rendering logic to ensure the cursor automatically positions at the top of the buffer after each view load or refresh
  • Refactored supertag-schema–goto-context function:
    • Replaced cl-return-from macro with flag-based approach for improved robustness
    • Eliminated dependency on cl-lib macros

These improvements make the schema view more user-friendly with clear instructions and an Emacs-convention-compliant interface.

[5.1.0] - 2025-09-11

Features

Implement Interactive Schema View for Tag Management

  • Introduce `supertag-schema-view`, a dedicated buffer for visualizing and managing the entire tag hierarchy.
  • The view is fully interactive, allowing direct manipulation of the tag and field schema from a single interface.
  • **Hierarchical Display**: `:extends` relationships are shown as an indented tree for clarity.
  • **Interactive Commands**:
    • `a`: Add a new top-level tag.
    • On a tag: `n` (new field), `e` (set extends), `d` (delete tag).
    • On a field: `r` (rename), `d` (delete), `M-up`/`M-down` (reorder).
  • **UX Enhancements**:
    • `g`: Intelligent refresh that preserves cursor position.
    • `j`/`k`/`p`: Standard navigation keys for familiar Emacs-style browsing.
  • The field reordering capability (`M-up`/`M-down`) has also been added to the Node View (`supertag-view-node.el`) for a consistent experience.

[5.0.0] - 2025-09-06

Major Refactor

Complete Architecture Overhaul with Pure Emacs Lisp Implementation

  • Eliminated Python dependency: Removed all Python backend components (simtag/) for a pure Elisp implementation
  • Data-Centric Architecture: Introduced single source of truth with `supertag–store` hash table
  • One-Way Data Flow: Implemented strict Action -> Ops -> Transform -> Store -> Notify pipeline
  • Performance Improvements: Reduced codebase by ~53% while enhancing core functionality
  • Simplified Deployment: No external dependencies required for core functionality

Unified Module Structure

  • Consolidated Components: Merged multiple modules into cohesive, single-responsibility files
  • Eliminated Circular Dependencies: Clean module hierarchy with clear separation of concerns
  • Improved Maintainability: Reduced inter-module coupling and increased cohesion

Enhanced Automation System 2.0

  • Rule Indexing: Automatic rule indexing for O(1) lookup performance
  • Multi-Action Support: Single rules can trigger multiple sequential actions
  • Scheduled Tasks: Integrated scheduler for time-based automation
  • Relationship Rollups: Automatic calculation and synchronization of related node data
  • Formula Fields: Real-time calculated fields in table views

New Capture System

  • Template-Driven Creation: Flexible node creation with dynamic content generators
  • Smart Field Filling: Automatic population of node fields from various sources
  • Interactive Enrichment: Post-creation field value addition workflow

Query System Enhancement

  • Query Block Renaming: S-expression query functionality renamed to “Query-Block”
  • Improved Syntax: Enhanced query syntax with better logical combinations
  • Time-Based Queries: Advanced temporal query operators
  • Dynamic Table Output: Clickable results with field value columns

Improved User Experience

  • Streamlined Interface: Removed experimental and rarely-used features
  • Consistent Navigation: Unified keybindings with C-c s prefix
  • Better Documentation: Comprehensive English and Chinese documentation

[4.6.2] - 2025-08-29

Improvements

Added custom setting to enable AI services

  • Usage: (setq org-supertag-bridge-enable-ai nil) to disable AI services
  • Default behavior: AI services are enabled by default

[4.6.1] - 2025-08-13

Improvements

Simplified user experience by removing overly complex features

  • Removed auto-tag module: Eliminated automatic tag generation system to reduce complexity and user distraction
  • Removed smart-companion module: Discontinued smart companion functionality to streamline the user interface
  • Improved focus on core functionality: These changes allow users to focus on the essential features of Org-SuperTag
  • Enhanced manual tag management: Integrated tag suggestions into the node view for better user control
  • Simplified background processing: Removed background scanning in favor of on-demand processing for better performance

Completely rewrote README documentation to enhance readability

  • Full documentation rewrite: Reorganized and rewrote the entire README from scratch
  • Improved structure: Better logical flow and organization of information
  • Enhanced clarity: More concise and understandable explanations of features
  • Bilingual consistency: Ensured both English and Chinese versions are properly synchronized
  • Better user guidance: Clearer onboarding process and feature exploration path
  • Comprehensive query system documentation: Detailed explanation of the dual-query architecture (Query Blocks vs Query Buffer)

[4.6.0] - 2025-08-06

Features

Enhanced Table View with display multi-line text and display images in cell

  • Display multi-line text
  • Display images in cells
  • Easy navigation shortcuts

[4.5.0] - 2025-07-31

Features

Enhanced Chat Command System with Intelligent Mode Switching

  • Intelligent Command Mode with Immediate Execution: Implemented a new command system that allows commands to be executed immediately with parameters while maintaining persistent mode
    • **Parameterized Command Execution**: `/bs 微软` → switches to bs mode and immediately executes with “微软” as input
    • **Persistent Command Mode**: Subsequent conversations remain in the selected command mode until user switches back with `/default`
    • **Flexible Command Usage**: Support both parameterized execution and mode-only switching
  • Improved Command Parsing: Enhanced command parsing to support alphanumeric, underscore, and hyphen characters in command names
  • Enhanced User Interface: Updated prompt display to show current mode as “User [bs mode]:” for better clarity
  • Better Command Management: Sorted command list alphabetically for improved user experience
  • Simplified Command Selection: Removed complex completion-at-point mechanism in favor of direct `completing-read` approach
    • Multiple command selection methods: `C-c /` for direct selection, `/` for smart slash, `C-c C-h` for command help
    • Disabled company-mode in chat buffer to prevent conflicts
    • Reduced code complexity by removing ~150 lines of complex completion code

Enhanced /define Command with Multiple Format Support

  • Multiple Format Support: `/define` command now supports three different formats:
    • Standard format: `/define name “prompt content”`
    • Empty prompt format: `/define name`
    • Double-quoted format: `/define “name” “prompt”`
  • Syntax Validation: Added comprehensive validation with helpful error messages
  • Improved Error Handling: Clear feedback for syntax errors and usage instructions
  • Enhanced Debugging: Added debug information and better command processing logic

[4.4.0] - 2025-07-30

Features

Enhanced Embed Block System with Bidirectional Synchronization

  • Bidirectional Synchronization: Implemented complete bidirectional sync between source nodes and embed blocks
    • **Embed to Source**: Changes in embed blocks automatically sync back to source nodes
    • **Source to Embed**: Changes in source nodes automatically refresh embed blocks
    • **Smart Content Merging**: Preserves source node structure (PROPERTIES, ID) while updating content
  • Automatic Synchronization: Added `after-save-hook` integration for seamless auto-sync
    • Automatically detects changes in both embed blocks and source files
    • Prioritizes embed-to-source sync to prevent conflicts
    • Provides detailed feedback messages for sync operations

[4.3.0] - 2025-07-28

Features

Enhanced S-expression Query Engine with Dynamic Table Output

  • **Dynamic Table Output for S-expression Queries**: The `org-babel-execute:org-supertag-query` function now generates dynamic Org tables with clickable node links, associated tags, and dynamically added columns for fields specified in the query.
  • **New Time-based Query Operators**: Introduced `(before “DATE”)`, `(after “DATE”)`, and `(between “START_DATE” “END_DATE”)` operators for filtering nodes based on creation timestamps.
  • **Relative Date Support**: Date arguments now support both absolute dates (“YYYY-MM-DD”) and relative dates (“-7d”, “+1m”, “now”) for flexible time-based queries.
  • **Improved Field Querying**: The `field` operator (`(field “FIELD_NAME” “VALUE”)`) has been thoroughly debugged and refactored to correctly retrieve nodes based on field name and value, with deep integration into `org-supertag-db.el`’s link storage mechanism.
  • **Enhanced Interactive Query Context**: The `org-supertag-query-find-nodes` function now provides more intelligent context snippets. When keywords match field names or values, the context displays `Field [FIELD_NAME]: FIELD_VALUE`, prioritizing relevant field information over general content snippets.
  • **New Interactive Command**: Added `M-x org-supertag-insert-query-block` to quickly insert a pre-formatted `org-supertag-query` babel block at point, prompting the user for the S-expression.
  • **Query History and Reusability**: Query history is automatically saved and can be reused, with frequently used queries prioritized in the history. Supports both keyword-based and S-expression queries in the same interface.

Query Operators Supported

  • **Tag queries**: `(tag “TAG_NAME”)` - Find nodes with specific tags
  • **Field queries**: `(field “FIELD_NAME” “VALUE”)` - Find nodes with specific field values
  • **Logical operators**:
    • `(and QUERY1 QUERY2)` - Both conditions must be true
    • `(or QUERY1 QUERY2)` - Either condition can be true
    • `(not QUERY)` - Exclude nodes matching the query
  • **Time-based queries**:
    • `(after “DATE”)` - Nodes created after the specified date
    • `(before “DATE”)` - Nodes created before the specified date
    • `(between “START_DATE” “END_DATE”)` - Nodes created between two dates
    • Date formats support absolute dates (“2024-01-01”) and relative dates (“-7d”, “+1m”, “now”)

[4.2.0] - 2025-07-27

Features

Implement tag field extension (inheritance)

  • Tag Extension: Tags can now extend other tags, automatically inheriting their field definitions. Child tag fields override parent fields.

[4.1.0] - 2025-07-26

Improvements

Implement card-based layout with structured three-line format

Fix

Fix node navigation by implementing robust real-time location lookup

[4.0.0] - 2025-07-21

New Features

AI Chat Service

  • Users can open AI Chat interface with M-x org-supertag-view-chat
    • Input content after * User: and press RET to send message directly
    • Provides RAG system’s Context content during conversation, clicking expands to jump to source content
  • Supports /commands command system similar to Claude Code
    • Type / and command name to invoke commands
    • Use /define command to define new commands
      • Syntax: /define <command-name> “Prompt”, e.g. /define branstrom “Help me brainstorm this topic”
      • Supports recognition of $INPUT variable in Prompt
  • Supports multilingual conversation, provides org-supertag-view-chat-lang custom option
    • Can directly set English, Chinese, Japanese, Korean, French, German, Italian, Portuguese, Russian

RAG Service: org-supertag now functions as RAG service for smarter tag recommendation and search

  • RAG service automatically checks database changes and incrementally updates to SQLite-vss database
  • Users can M-x org-supertag-background-run-now to manually run RAG synchronization
  • RAG provides accurate information retrieval for LLM in background

Tags: New auto-tag recommendation feature

  • LLM will automatically generate tag suggestions for nodes without tags in the backend
  • Provides unified interface for applying tag recommendations

Smart Partner: New smart partner functionality

  • Smart Partner automatically generates tag suggestions for nodes without tags
  • Provides unified interface for applying tag recommendations

Scheduled Tasks: Used to manage background services

Refactors

EPC: Completely refactored EPC server backend

  • Less code with cleaner structure

Relationships: “Co-occurrence relationships” of tags no longer stored in separate files, but unified with other relationship types in LINK data objects

Relationships: Redesigned Org Supertag Relation Management interface

  • No longer supports Find by Groups, Isolate Tag, etc. to focus functions

Completion: Removed direct support for Company, now uses Emacs’ built-in completion-at-point function

  • Works seamlessly regardless of whether user uses Company or Corfu

Tags: Removed original “preset tag” mechanism

Tags: No longer supports org-mode’s traditional TAGS, only supports custom #inline-tag

  • Not restricted by org-mode’s character limitations for TAGS input
  • Can use #inline-tag anywhere

Properties: Removed direct modification of org-mode PROPERTIES

  • Users can now open node management interface with M-x org-supertag-view-node to modify Fields, data will be directly saved to database
  • org-supertag-tag-set-field-and-value is no longer available

Behavior System: Extracted scheduling functionality into separate timer task component (org-supertag-scheduler.el) to support other org-supertag services

Removed

Removed org-supertag-backlink.el

  • Related functionality has been integrated into org-supertag-view-node
  • Its provided commands are no longer available

[3.0.5] - 2025-06-25

Features

Decouple metadata and adopt a database-first architecture

  • Fields are no longer synchronized with the Org :PROPERTIES: drawer. All field data is now read from and written directly to the database.
  • Support for headline :TAGS: has been completely removed. The system now exclusively uses org-supertag’s inline-tag format for visual representation in files, while the actual tag relationships are managed solely by the database.
  • This view is now the central and sole interface for viewing and editing all metadata associated with a node. Users can modify field values directly within this view, creating a closed loop for data manipulation.
  • The :tag field type has been significantly upgraded to natively support one or more tag values (e.g., a movie with multiple directors). The UI accepts comma-separated input and formats the display cleanly (e.g., TagA / TagB / TagC).
  • The legacy List and Range field types have been deprecated to streamline the data model and reduce complexity.

[3.0.4] - 2025-05-30

Bug Fixes

Fixed database cleanup path matching issues

  • Fixed path normalization in database cleanup to prevent aggressive node removal
  • Normalized all file paths to absolute paths using `expand-file-name` for accurate comparison
  • Created separate `sync-files` hash table to ensure precise path matching
  • Ensured only nodes from files truly not in sync state are removed
  • Prevented accidental removal of nodes from files still in sync scope
  • Improved logging and debugging information for path comparison process

[3.0.3] - 2025-05-30

Improvements

Improved sync scope management with automatic node cleanup

  • Added automatic removal of nodes from files outside sync scope
  • Enhanced sync state management to maintain database consistency
  • Improved periodic sync to handle out-of-scope file cleanup
  • Added detailed logging for sync scope changes and node removal
  • Ensured proper cleanup when modifying sync directories

[3.0.2] - 2025-05-24

Improvements

Enhanced synchronization stability with comprehensive data protection

  • Added preservation and restoration of non-node entities during sync operations
  • Enhanced protection for field definitions, tag relations, and link data
  • Improved `org-supertag-sync-force-all` to preserve all critical data structures
  • Strengthened `org-supertag-sync-all-nodes` with comprehensive data backup
  • Ensured database consistency during intensive sync operations

Fixed tag vector database update mechanism

  • Resolved issues with tag_vector library updates
  • Improved vector database synchronization reliability
  • Enhanced tag similarity computation stability

Fixed AI auto-tagging system freezing issues

  • Resolved `org-supertag-sim-auto-tag-node` causing Emacs to freeze
  • Added comprehensive timeout protection for all AI operations
  • Replaced blocking EPC calls with asynchronous implementations
  • Added initialization state validation to prevent infinite loops
  • Enhanced error handling with automatic recovery mechanisms

Features

Added comprehensive database recovery system

  • Introduced `org-supertag-recovery` methods for database restoration
  • Provided automated recovery workflows for corrupted data
  • Added diagnostic tools for database integrity checking
  • Enhanced backup and restore capabilities

Added AI system diagnostic and emergency tools

  • Added `org-supertag-sim-diagnose` for comprehensive system health checking
  • Added `org-supertag-sim-emergency-reset` for complete system recovery
  • Added `org-supertag-sim-safe-auto-tag-node` as a safer alternative to auto-tagging
  • Provided detailed troubleshooting guidance and emergency commands

Maintenance

Codebase cleanup and organization

  • Removed deprecated protection scope checking utilities
  • Removed redundant recovery process files
  • Updated `.gitignore` to exclude development directory
  • Streamlined codebase for better maintainability

[3.0.1] - 2025-04-26

Features

Add Backlink panel for node references

  • Add org-supertag-backlink-show command to display node references
  • Show both referenced and referencing nodes
  • Display node content with proper formatting
  • Support keyboard navigation and node viewing

[3.0.0] - 2025-04-05

Features

Add AI backend for tag auto suggestion

Add Bidirectional tag relation management

Add Table View

Refactor sync-mechanism

[2.5.3] - 2025-03-11

Features

Enhanced multi-column view with relation-based tag management

  • Added org-supertag-view-add-related-tag-to-column for adding related tags to existing columns
  • Improved tag relation integration in column view:
    • Select source column and relation group
    • Choose from available related tags
    • Add selected tag to existing column
  • Enhanced user interface with clear feedback messages
  • Added comprehensive error handling for edge cases

Improvements

Better user interaction in multi-column view

  • Clearer operation instructions and feedback
  • Improved error messages and edge case handling
  • Better integration with tag relation system

[2.5.2] - 2025-03-10

Features

Improved inline tag support with enhanced visual styling

  • Improved inline tag fontification with adaptive theming
  • Enhanced tag appearance with slight elevation and proper spacing
  • Fixed compatibility issues with font-lock implementation
  • Optimized tag recognition in various contexts
  • Improved tag insertion with better completion support
  • Added support for hiding # prefix symbol with customizable option

The inline tag system now provides:

  • More reliable tag highlighting in all contexts
  • Better visual distinction between tags and regular text
  • Smoother integration with org-mode’s native styling
  • Improved performance for documents with many tags

Added comprehensive customization options for inline tags

New variables for complete control over tag appearance:

  • `org-supertag-inline-enable-fontification`: Toggle tag highlighting
  • `org-supertag-inline-light-theme-colors`: Color settings for light themes
  • `org-supertag-inline-dark-theme-colors`: Color settings for dark themes
  • `org-supertag-inline-text-properties`: Control font size, weight, and spacing
  • `org-supertag-inline-tag-regexp`: Customize pattern for matching tags
  • `org-supertag-inline-hide-hash`: Option to hide/show the # prefix
  • `org-supertag-inline-excluded-contexts`: Control where tags are highlighted

These customization options allow users to tailor the appearance of inline tags to match their preferences and theme setup, ensuring optimal readability and visual harmony with the rest of their document.

[2.5.1] - 2025-03-03

Bug Fixes

Fix org-supertag-tag-delete-at-all to ensure complete cleanup

  • Replace org-supertag-db-remove-object with direct remhash for reliable deletion
  • Add comprehensive cleanup of all tag-related database entries
  • Implement proper cleanup sequence for tag removal
  • Add debug logging for deletion process tracking

Fix org-supertag-tag-set-field-and-value for not working

[2.5.0] - 2024-03-03

Features

(org-supertag-inline.el) Add inline tag support for content tagging

  • Enable users to tag specific content within org nodes using #hashtag style
  • Seamlessly insert tags in text with C-c t i:
    • Choose from existing tags through completion
    • Create new tags on the fly by adding ‘#’ suffix
    • Quick access to preset tags marked with [P]
  • Tags in content maintain proper spacing for readability
  • Tags automatically link to current node without cluttering headlines
  • Natural writing flow with minimal disruption: “Meeting with Owen who is my #friend about the #project”

This feature allows for more granular content organization while maintaining a clean document structure. Users can now tag specific parts of their notes without adding tags to headlines, making the tagging system more flexible and context-aware.

[2.4.0] - 2025-03-02

Featuresf

org-supertag-view-discovery: Discover and explore your tags naturally

Introducing Tag Discovery View - a new way to explore and understand your tag ecosystem:

  • Start from any tag and discover related content: “Starting with #project, what other tags frequently appear with it?”
  • See the bigger picture of your tag usage:
    • How many notes use this combination of tags?
    • Which tags often appear together?
    • What are the relationships between different tags?
  • Build your tag filters progressively:
    • Add more tags to narrow down your focus
    • Remove tags to broaden your view
    • Reset and start a new exploration path
  • Preview and access content directly:
    • See matching notes with their titles and status
    • View complete note content with a single click
    • Understand tag relationships in context
  • Smooth keyboard-driven operation:
    • a - Add tag to filter
    • d - Remove tag from filter
    • r - Reset all filters
    • v - View selected node
    • m - Manage tag relations
    • g - Refresh view
    • q - Quit

org-supertag-view-column: Compare and analyze with Multi-column Tag View

A powerful new way to analyze your tagged content side by side:

  • Compare different tag combinations simultaneously: “Show me all #project tasks alongside #urgent items and #delegated work”
  • Build your analysis view flexibly:
    • Start with any tag combination
    • Add new columns for different perspectives
    • Add related tags based on actual usage patterns
    • Remove columns you don’t need anymore
  • See your content from multiple angles:
    • Compare todo states across different tag combinations
    • Track related items across different categories
    • Identify patterns and relationships in your notes
  • Stay in control with keyboard shortcuts:
    • a - Add a new column
    • A - Add a column with related tag
    • t - Add tag to current column
    • T - Add related tag to column
    • d - Remove unwanted column
    • r - Start fresh
    • v - View node details
    • m - Manage tag relations
    • q - Quit

These new views work together to help you:

  • Understand how your tags are connected
  • Find relevant content more efficiently
  • Discover patterns in your note organization
  • Make better decisions about tag usage

Change

Original `org-supertag-view-tag` chang to `org-supertag-view-tag-only`

[2.3.0] - 2025-03-01

Features

Add tag relationship management module (org-supertag-relation.el)

  • Implement sophisticated tag relationship system with symbolic representation
    • Add 10 relationship types with intuitive symbols (A ⊃ B, A → B, A ⋈ B, etc.)
    • Support relationship categories: hierarchical, causal, associative, co-occurrence
  • Enable automatic discovery of relationships based on usage patterns
    • Statistical analysis of tag co-occurrences with strength metrics
    • Real-time incremental updates when tags are added/removed
  • Provide intelligent tag recommendations based on contextual relevance
  • Create dedicated management interface for visualizing and editing relationships
  • Support customizable relation groups for organizing related tags
  • Add relationship strength thresholds to filter insignificant relationships

[2.2.13] - 2025-02-18

Bug Fixes

Fix node movement failing due to missing file path

  • Fixed node sync to properly save file path information
  • Added comprehensive debug logging for node operations
  • Improved error handling in node movement process
  • Enhanced node property validation before movement

[2.2.12] - 2024-02-16

Features

Enhanced reference creation with duplicate detection

  • Added smart node reference creation that checks for existing nodes
  • Reuses existing nodes when title matches exactly
  • Prevents duplicate node creation for same content
  • Improves reference management efficiency

Bug Fixes

Fix field edit mode keymap and key bindings

  • Fixed keymap initialization in org-supertag-field-edit-mode
  • Simplified mode definition by removing unnecessary syntax-table
  • Fixed key bindings for field editing operations (e, d)
  • Improved error handling in field operations

[2.2.11] - 2024-02-12

Features

Enhanced org-supertag-view with field value display

  • Added field value support in tag view table
  • Implemented case-insensitive field name matching
  • Added comprehensive field value lookup
  • Display field values in dedicated columns
  • Improved table formatting for better readability

[2.2.10] - 2024-02-03

Fix

Fix tag company completion not working in heading

Fix field edit mode not working

[2.2.9] - 2024-02-03

Features

Added

  • New component: org-supertag-view.el
    • Provides visualization system for tag content
    • Smart command org-supertag-view-tag that:
      • Shows tag content directly when cursor is on a tag
      • Offers tag completion when not on a tag
    • Table-based view showing:
      • Node titles
      • Node types
      • Creation dates

[2.2.8] - 2024-02-03

Bug Fixes

Fix tag deletion to ensure complete cleanup

  • Replace org-supertag-db-remove-object with direct remhash for reliable deletion
  • Add comprehensive cleanup of all tag-related database entries
  • Implement proper cleanup sequence for tag removal
  • Add debug logging for deletion process tracking
  • Add org-supertag-after-tag-delete-hook for post-deletion customization

[2.2.7] - 2025-02-02

Features

Improve node movement(org-supertag-node-move-node) with outline path display

  • Show complete outline structure when moving nodes
  • Display paths in “filename / outline-path / title” format
  • Improve target location selection with clear hierarchy view
  • Keep existing insertion options (file start/end, under/same level)

[2.2.6] - 2024-1-31

Features

Add comprehensive node finding system

  • Add org-supertag-node-find command for interactive node finding
  • Add org-supertag-node-find-other-window for other window display
  • Show complete node path in format “filename / outline-path / title”
  • Add robust error handling for missing metadata
  • Sort results alphabetically for easier navigation
  • Improve node visibility with automatic expansion

[2.2.5] - 2024-1-30

Features

Add company-mode completion for supertags

This commit adds interactive tag completion using company-mode:

  • Add company backend for supertag completion
  • Trigger completion with ‘#’ prefix
  • Integrate with existing tag management system
  • Simplify implementation by reusing org-supertag-tag-add-tag
  • Add tag deletion functionality (org-supertag-tag-delete)

The completion system provides a smoother user experience for adding supertags to org headlines, with proper tag formatting and handling of preset tags.

[2.2.4] - 2024-1-29

Refactor

Change field type tag-reference to direct node reference

BREAKING CHANGE: Completely redesign the tag-reference field type to reference nodes directly instead of referencing other tag’s field values.

Before:

  • tag-reference pointed to another tag’s field value
  • Used for value inheritance between different tags

After:

  • tag-reference now directly points to org nodes
  • Uses org-mode’s link format (TITLE)
  • Enables direct node relationships

Changes:

  1. Rewritten org-supertag-field–convert-value to handle org-mode link format
  2. Modified org-supertag-validate-tag-reference to validate node existence
  3. Added node title lookup for better link display

Note: This is a breaking change. Existing tag-reference fields will need to be migrated to the new format.

[2.2.3] - 2025-01-28

Features

Improve field type selection and display

  • Add help message showing available field types with descriptions
  • Use all field types from org-supertag-field-types instead of hardcoded list
  • Format field display with fixed-width alignment for better readability
  • Clean up help buffer after type selection
  • Improve field editing interface with consistent formatting

[2.2.2] - 2025-01-26

Bug Fixes

Fix node movement and link creation

  • Completely replace original heading with link reference
  • Remove redundant title text in reference link
  • Maintain proper heading level and spacing
  • Ensure link insertion starts at beginning of line
  • Add proper error handling for nil values
  • Update documentation for modified behavior

[2.2.1] - 2025-01-26

Refactor: Improve query history handling and fix node navigation

  • Fix history saving for multi-keyword queries
  • Store complete search queries in history
  • Fix recenter error in node navigation
  • Improve error handling in node lookup
  • Add proper return values for navigation functions
  • Update documentation for modified functions

[2.2.0] - 2025-01-25

Features

Add automatic node creation for org headlines

  • Add org-supertag-sync-auto-create-node to control auto creation
  • Add org-supertag-sync-node-creation-level for minimum level
  • Modify org-supertag-db-update-buffer to handle auto creation
  • Update org-supertag-sync--process-node for auto conversion

This change enables automatic conversion of org headlines to nodes during synchronization, making node creation more seamless while maintaining control through customization options.

Add directory scope control for synchronization

  • Add org-supertag-sync-directories for monitored directories
  • Add org-supertag-sync-exclude-directories for excluded paths
  • Add org-supertag-sync-file-pattern for file matching
  • Add helper functions for directory management:
    • org-supertag-sync-add-directory
    • org-supertag-sync-remove-directory
    • org-supertag-sync-list-directories

Improve

File organization

  • Move org-supertag-sync-state.el to org-supertag-data-directory
    • Add directory existence check in org-supertag-sync-init
    • Keep data files organized under ~/.emacs.d/org-supertag/

Bug Fixes

Fix node movement reference format

  • Now completely replaces original heading with link reference
  • Removes redundant title text in reference link
  • Maintains proper heading level and spacing

[2.1.0] - 2025-01-18

Features

Node: Enhanced node movement functionality

Added interactive node movement command

  • Added org-supertag-node-move-node for direct node movement
  • Support comprehensive insertion positions:
    • File start (after org-mode header)
    • File end
    • Under selected heading
    • Same level as selected heading
  • Added clear error messages and operation feedback

Added file start position support

  • Added helper function to locate content start after org-mode header
  • Added “File Start” option for node movement and export operations
  • Improved file header preservation during node operations
  • Better integration with existing query and export features

[2.0.0] - 2025-01-13

Features

Behavior: Added schedule trigger and deadline management system

Implemented cron-style scheduling for behaviors

  • Format: “minute hour day month weekday”
  • Examples:
    • “0 9 * * 1-5” (weekdays at 9:00)
    • “30 * * * *” (every hour at :30)
    • “0 0 1 * *” (first day of each month)

Added native org-mode timestamp integration

  • Support for SCHEDULED and DEADLINE timestamps
  • Support for absolute timestamps (<2024-03-20 Wed>)
  • Support for relative time expressions (now+2h, now-1d)
  • Support for property-based timing (${prop:DEADLINE}-2h)

Added deadline check behavior (@deadline-check)

Added derived behaviors for deadline management:

  • @overdue-urgent: Set high priority for overdue tasks
  • @upcoming-deadline: Mark tasks due in next 3 days
  • @overdue-archive: Archive overdue completed tasks

Field: Enhanced timestamp handling

  • Improved org-mode timestamp format support
  • Date fields now use active timestamps (<…>) for agenda visibility
  • Timestamp fields use inactive timestamps ([…]) for historical records
  • Added automatic format conversion between different timestamp styles

Behavior Template: Added template variable system for behaviors

Added support for dynamic variable expansion

  • ${input:prompt} for user input
  • ${date:format} for date operations
  • ${prop:name} for property access
  • ${context:key} for shared data

Added conditional execution with {if:condition}

Added relative date calculations (now+1d, now-1w)

Added simple context object for data sharing

Node Management: Added new node operations

  • Added org-supertag-move-node-and-link command for node relocation with reference
  • Added org-supertag-delete-node command for complete node cleanup
  • Added flexible node movement behaviors
  • Support for keeping links at original location
  • Interactive target selection

Sync: Added automatic synchronization system

  • Added file state tracking
  • Added buffer modification monitoring
  • Added automatic sync with conflict detection
  • Added state persistence and error recovery
  • Added node-level conflict analysis command
  • Added file analysis tools for debugging sync issues

Query: Enhanced query system

  • Added cursor-based node insertion in query buffer
  • Added history support for query keywords
  • Added customizable history size
  • Preserved last used keywords

Reference: Improved reference system

  • Added reference type tracking (ref-from)
  • Added reference count tracking

Tag: Added improved tag change functionality

Performance: Added asynchronous face refresh mechanism

Bug Fixes

Fixed tag prefix matching issues (#9)

  • Improved tag completion with custom completion function
  • Added TAB completion support

Fixed field type options insertion

Fixed behavior system event handling

Improved error handling in node operations

Enhanced synchronization stability

Refactoring

Query: Rewritten query results display using ewoc for better performance

Query: Implemented query buffer reuse for cursor insertion

Behavior: Improved behavior system architecture with detach support

Sync: Enhanced file synchronization architecture

Internal Changes

  • Added scheduler system with task management
  • Added robust file synchronization mechanism
  • Optimized reference handling with better metadata tracking
  • Optimized query result caching
  • Improved node deletion with comprehensive cleanup
  • Enhanced query system with history management
  • Improved code organization and modularity
  • Enhanced documentation and examples