Conversation
chore(sync): main -> develop
Record the two-branch model and the release flow in CLAUDE.md and docs/DEVELOPMENT.md: bump the version in the four spots, open a develop->main PR, and merge it as a merge-commit. Emphasize that release-tauri auto-tags from Cargo.toml on the main push, so tags must never be created manually (pre-tagging stalls the build chain). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Design doc capturing the decisions for merging Praetorium into Token Dashboard as a single v5.0 desktop app: drop the Praetorium CLI, port praetorium-core + the live tauri modules, mount Live as one rail item with a Console/Cockpit/Explorer sub-switcher and an optional pop-out window, keep the read-only framing with Live as opt-in. https://claude.ai/code/session_015h2hEjpj6S2pisDcdsc9nH
Stage 1 of the V5 Praetorium merge (backend). - Add praetorium-core as a workspace member (pure parser/vault library, its 26 unit + 4 integration tests come along and pass). - Port the live Tauri modules into token-dashboard-tauri under src/live/: process.rs (spawns the claude CLI, streams events over an IPC Channel), session_watch.rs (notify watcher on ~/.claude/projects/), plus the sessions/vault command wrappers. Drop praetorium-cli (desktop only). - Register the live commands + shell/dialog/opener plugins + RunRegistry in the Tauri builder; add tokio process/io-util + notify deps. - Add open_live_window/close_live_window/is_live_window_open commands and a #live-window route for the pop-out window, mirroring the existing widget/setup-help secondary-window pattern. Emits live-window-opened/ -closed to the main window so the docked tab can re-dock. Workspace builds clean; cargo clippy -D warnings, cargo fmt --check, and cargo test --workspace all pass. https://claude.ai/code/session_015h2hEjpj6S2pisDcdsc9nH
Stage 2 of the V5 Praetorium merge (frontend + version). - Port Praetorium's frontend under frontend/src/live/ (components, stores, lib, themes, assets), preserving internal structure so relative imports stay valid. Add d3-force/d3-hierarchy/marked/@tauri-apps deps. - New frontend/src/live/index.jsx wraps Praetorium's Console/Cockpit/ Explorer view-switcher as a Token Dashboard sub-surface, with a pop-out control. LiveTab (docked) collapses to a placeholder while the standalone window is open; LiveWindow is mounted by entry.jsx on #live-window. - Register the 'live' route (app.jsx), nav rail item (nav-items.js, gated at level 2 in levels.js), and update the nav-order test. - session_watch: make watch_sessions idempotent (single owning watcher stored in a startup-managed handle) so the pop-out re-dock cycle can't double-manage and panic. - Ship live.css (Praetorium's stylesheet) with the few global/element rules (html/body, form controls, .a-ambient, reduce-motion banner) scoped under .pr-root so the Live subtree can't repaint the host dashboard; linked in index.html. Add the .pr-live-* container styles. - Bump version to 5.0.0 (4 canonical spots); reframe the README so the analytics tabs stay read-only and Live's claude-launch is opt-in; note the live module + praetorium-core in CLAUDE.md. cargo fmt/clippy -D warnings/test --workspace and the 192-test frontend suite (TD + ported Praetorium) all pass. esbuild bundles clean. https://claude.ai/code/session_015h2hEjpj6S2pisDcdsc9nH
- Add the 'live' pop-out window to the capability windows list so IPC (invoke/Channel/event) works from the detached window. - Add dialog:allow-open for the vault / working-directory pickers. - Drop the shell and opener plugins: claude is spawned via tokio::process (not the shell plugin) and the live frontend imports neither, so they were unused. Keeps the plugin surface minimal. cargo clippy -D warnings clean. https://claude.ai/code/session_015h2hEjpj6S2pisDcdsc9nH
Two CI failures from the merge: - cargo-deny: praetorium-core had no license field -> error[unlicensed]. Inherit the workspace package metadata (MIT) like the sibling crates and bump it to 5.0.0. 'cargo deny check advisories licenses sources bans' now passes. - node --test: the 'src/**/*.node-test.mjs' glob relied on node's internal glob expansion, which only exists in node 21+. CI runs node 20, so the literal path matched nothing and failed. Use a find-based file list, which is node-version-agnostic and still recurses into src/live/. 192/192 frontend tests pass. https://claude.ai/code/session_015h2hEjpj6S2pisDcdsc9nH
Folds Praetorium in as the Live tab (Console/Cockpit/Explorer) with an optional pop-out window; analytics tabs stay passive/read-only. Ports praetorium-core + the live tauri modules, drops praetorium-cli, bumps to 5.0.0. All CI green; manual desktop QA still pending.
…ettings (#152) Make the ported Praetorium "Live" feature a first-class native dashboard tab. - Fix Live IPC: declare app commands in build.rs (AppManifest::commands) and grant them in the capability, so the remote-served webview can actually call them — sessions, chat and vault were all denied by the Tauri ACL. Resolve the claude binary path explicitly in process.rs so a stale PATH does not break runs. - Rewrite Console/Cockpit/Explorer into native .a-* dashboard components; drop the embedded Praetorium shell for a secondary nav rail merged with the main bar. - Pop-out window: pre-create it hidden during setup() (WebView2 will not navigate a window spawned later from a command), then show/hide on demand. Native titlebar with drag + minimize/maximize/close, plus active-theme and glass/acrylic parity with the main window. - Add a vault-folder picker to Settings for the Live Explorer. - Unify typography (weight-400 values/titles, plain JetBrains Mono glyphs). Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
The three Live sub-views sized to content, leaving dead space below and to the right. Make the shared .a-live-view a full-height flex column and have each view fill it: console drops the magic calc(100vh - 130px) for height:100%, cockpit's graph swaps the fixed 70vh for flex:1, and the previously-unstyled .a-explorer gets a flex column + scrollable pane. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
DESIGN.md documented 4 themes (dark/light/forge/forest); theme.js actually ships 19 (10 dark, 6 light, 3 special). Rewrote the §2 theme tables by mode with real token anchors, added a Live-tab Components subsection, and carved out the Live Cockpit's 48px metric as a scoped exception to the No-Hero-Type rule. Regenerated the DESIGN.json sidecar (themes 4->19 + live subsystem) and corrected the stale "14 themes" line in CLAUDE.md. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
DESIGN.md's "22px apex / no font >24px" rule was fiction: .a-metric is clamp(28px,4vw,36px), scaling to 56px in spacious density and special themes by design. Renamed to the No-Hero-Metric Rule (the ban is the SaaS centered-hero template, not large numerals) and fixed the metric token + prose in DESIGN.md/DESIGN.json to match the real responsive scale. Brought side-stripes into spec (<=1px colored left/right borders): main dashboard .a-side-list/.a-banner/.a-budget-banner-row and the Live .pr-* list/turn/answer/blockquote rows reduced 2-3px -> 1px; .pr-line-prompt stripe dropped (tinted bg + ">" glyph carry it). Also #fff hover -> var(--bone) and .a-api-group h2 uppercase -> 10px label tier. Removed frontend/src/live/themes/styles.css: a dead, divergent copy of the served frontend/live.css (nothing imports it; missing the pop-out window palettes live.css carries). live.css is canonical. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* perf(db): tune connections (WAL/mmap/cache) and pool read connections Tier 1 of the performance roadmap. The analytics read path opened a fresh, untuned SQLite connection per query — the day view opened six per click on a default-journal-mode DB. - db::tune() applies WAL, synchronous=NORMAL, mmap_size=256MB, cache_size=64MB, temp_store=MEMORY on both the writer open() and the read-path open_ro(). WAL stops reader/scanner contention; mmap+cache keep the hot DB warm across opens. - open_ro() now hands back a pooled, tuned connection via a Deref guard (PooledConn) that returns itself to a process-wide pool on drop, so the PRAGMA setup and page cache stay warm. Callsites compile unchanged thanks to Deref coercion. Measured on a real 245MB / 147k-row DB: /api/day 477.7 -> 130.7 ms (3.6x), tags-summary 73 -> 45 ms, daily 88 -> 65 ms. Adds a perf_probe example to keep each tier honest with before/after numbers. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * perf(db): index substr(timestamp) so date filters seek instead of scan Tier 2 of the performance roadmap. Every day/* query filters `WHERE substr(timestamp,1,10) = ?`, which made the plain timestamp index unusable and full-scanned messages — six scans per calendar day-click. Add expression indexes idx_messages_day (substr(timestamp,1,10), type) and idx_tools_day. SQLite uses an expression index when the WHERE clause repeats the expression, so the existing SQL needs no rewrite. Indexes ship in SCHEMA with IF NOT EXISTS, so init_db builds them once on next launch for existing DBs. day_by_model EXPLAIN: SCAN -> SEARCH ... USING INDEX idx_messages_day. /api/day 130.7 -> 14.0 ms (34x vs baseline), under the 50ms target. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * perf(db): keep planner statistics fresh so tag aggregate seeks Tier 3 of the performance roadmap. With no sqlite_stat1, the planner drove tag_aggregates from messages (SCAN m over 147k rows) instead of the small session_tags table. ANALYZE flips it to SCAN st -> SEARCH m USING idx_messages_thread. Rather than build precomputed rollup tables (write-path complexity + drift risk), refresh stats in two cheap spots: - scan_dir runs `PRAGMA analysis_limit=400; PRAGMA optimize` after any scan that added rows. - init_db runs a one-time sampled ANALYZE when sqlite_stat1 is absent, so existing DBs get stats on next launch. tags-summary 46 -> 39 ms here; much larger win for users with few tagged sessions (no longer scans every message). All hot queries now under the 50ms target, so no rollups needed. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * perf(scan): only broadcast scan_complete when rows were ingested Tier 4 of the performance roadmap. The background scan loop ticks every 10s and emitted scan_complete unconditionally, so every connected frontend refetched its whole endpoint registry every 10s even when the user generated no new transcripts. Emit only when the scan ingested rows. Liveness is held up by the server 15s keep-alive ping, the frontend watchdog is one-shot, and the manual /api/scan route returns stats over HTTP, so suppressing idle events is safe. The planned query cache and SSE debounce were dropped as YAGNI: after T1-T3 every hot query is 13-40ms warm. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
…bottleneck (#154) Follow-up to PR #153 suspected first_prompts dominated /api/day (~95ms of ~108ms end-to-end). Measured directly: first_prompts is ~6ms for ~188 sessions and the endpoint is ~31ms server-side (raw HTTP). The earlier ~108ms was PowerShell Invoke-RestMethod deserializing the 4.6KB response into objects — a client-side artifact absent in the app, whose JS frontend parses the payload natively in <1ms. No query change warranted: rewriting the windowed query to GROUP BY MIN(timestamp) would risk the identical-timestamp tie case for ~4ms on an endpoint already under the 50ms target. Add the first_prompts timing + EXPLAIN to perf_probe so the finding stays measurable. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
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
2 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.
@
Release v5.0.0
Promotes
develop→main. The merge-commit auto-tagsv5.0.0(release-tauri reads the version fromtauri.conf.json).Headline: native Live tab (Praetorium integration)
praetorium-core+ live backend into the workspace (84c909f)8b2fef2) — bumps to 5.0.013a810a)6f71f5f)eb8b49e)Performance & design
122e7fe)e6dadfc)5324af5)476611d)CI / docs
a08d868)🤖 Generated with Claude Code
@