The Add-tile menu would stop appearing once enough tiles were on the
dashboard — a different count each time. The button's handler awaited
GET /api/modules before revealing the menu, and that request could never
complete: browsers allow about six HTTP/1.1 connections per host, an open
EventSource holds one for as long as it lives, and every stream-fed tile
(ProPresenter, Timers, PCO Plan, ProdCom) opened its own. With the shell's
own /api/events that is the whole allowance at five or six such tiles, and
every request after that — the picker's module list, a new tile's /state,
its stream — sat in the browser's queue for a connection that was never
handed back. The count varied because a stream in its 3-second retry loop
frees a connection for a moment, and a clock tile holds none.
Reproduced with six stream-fed tiles: the sixth stream and the picker's
/api/modules both pend forever, and the menu stays hidden after the click.
Three changes, all in the shell:
- Dropdowns open the moment they are clicked, from what the page already
knows, and slot in what their rebuild fetches when it lands. A rebuild
that fails leaves the previous contents rather than an invisible menu,
and a generation token stops a late fetch from repainting a menu that has
since been closed or reopened (the old await also let two quick clicks
cancel each other out). The ＋ picker repaints only when the module list
actually changed; the layout menu shows Save and Reset at once and
inserts the server's layouts above Reset when they arrive.
- Tiles asking for the same stream share one connection. moduleApi.sse()
now subscribes to a shell-wide pool with one EventSource per URL, fanning
events out to every subscriber and reconnecting on the same 3-second
timer while anyone listens; a tile joining an open stream is told `open`
straight away, so its backfill runs as before. Six tiles of three modules
now hold three connections instead of six. Modules see no difference.
- /api/modules and /api/layouts fetches give up after 8 s instead of
pending forever, so nothing awaiting them can hang.
Verified in the browser pane against a scratch server: the same six-tile
layout now opens the picker instantly with all five modules listed and
/api/modules completes; with the pool deliberately starved (three extra
EventSources) the picker still opens from its cached list.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The Timers module becomes the one place a production tech looks for every
timer and clock: ProPresenter's timers as before, the LTC timecode, a
time-of-day clock (the Clock module's format / seconds / date, folded in as
a sub-part — modules/clock itself is untouched), and timers other modules
offer over the module guide's provider contract: PCO Plan's service /
current item / countdown, or anything else declaring provides: ["timers"].
Server: a small `sources` abstraction. Each source turns its own world into
the guide's standard timer objects and reports health — ProPresenter (the
existing poller, wrapped, behind ppEnabled), LTC (the existing listener,
untouched, as a kind:"clock" timer whose status.text is the timecode, with
fps/df/ageMs in an `ltc` extension), Clock, and one module source per
provider found in the shell's /api/modules over loopback (shell.port):
/timers/stream when it answers as SSE, else /timers polled every 2 s;
re-discovered every 30 s; backoff 3 s → 15 s; a provider's timers survive a
10 s grace and then go. PCO Plan sits behind pcoEnabled, the rest behind
moduleTimersEnabled. The module is itself a provider: GET /timers and
GET /timers/stream serve the merged set. The header comment says how to
add a platform in a dozen lines.
Client: one tile, every timer, grouped under source headings — label, the
big time derived locally from the snapshot and a server-offset-corrected
clock (so digits tick smoothly between snapshots), a status line, a detail
line. Countdowns count down, elapsed count up, clocks show the time; the
LTC card still counts frames. Feedback colours (running / warning near the
end / overrun / idle) are per-tile colour settings applied as custom
properties on the tile root, so they win over dashboard themes; left at
their defaults they follow the theme. Gear settings: Clock, ProPresenter
timers, LTC timecode, Module timers, Source headings, Status lines, the
clock's format / seconds / date; a "Timers ▾" header menu hides any single
timer (source:id, saved with the layout; the old hiddenItems and
timer:<uuid> / ltc picker variants keep working).
Manifest: 1.3.0, proddash >=1.4.0, provides timers, configGroups with a
toggle switch per integration (ProPresenter / LTC Timecode / PCO Plan /
Other modules) and group help; switch labels are nouns.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The service clock now belongs to one service at a time: a plan's services
take turns by window (the next one takes over pre-roll before its start,
once the previous one has ended or was due to end), and moving into the
next window resets the service elapsed clock, the item timing and the
countdown. A service starts at its scheduled time or, under the new
"Service timing" admin group, when a chosen ProPresenter document goes
live (scheduled time as the failsafe); it ends when the chosen end
document goes live, or at the next service / plan end + 60 min.
The pure logic lives in service-timing.js and matching.js (cue documents
are matched like plan items, with a stricter 0.75 bar so the pre-service
loop never ends a service) with service-timing.test.js covering windows,
precedence and failsafes. GET /timers and /timers/stream offer the
service, current-item and countdown timers per the module guide;
GET /documents lists the documents the Now/Next feed has shown or listed
for the new selects. Loopback calls use shell.port, with the old
proddash.json guess only when the shell hands no port.
Manifest: 1.4.0, proddash >=1.4.0, provides ["timers"]; Client ID
wording, shorter labels, group help tightened, field help trimmed.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
configGroups[name].help now renders as an ⓘ beside the group title (.group-info / .group-info-btn / .group-info-pop): a real labelled button — click pins a popover with the text, hover and focus peek at it, Escape or a click elsewhere closes it, and inside a collapsed group's <summary> it never folds the group. The paragraph above the fields is gone; per-field help stays one small line under its control.
configGroups[name].toggle names a switch/boolean field of the group: it moves into the title row, right-aligned (.field-group-toggle), as the group's on/off. While off the body dims to 0.45 (.field-group.is-off) but every field stays editable; the field is not repeated in the body, is saved exactly as before and still drives showWhen. Works with collapsed (the switch never folds the row) and columns. A toggle naming a non-boolean field is ignored.
The module card's status line is a .module-status readout: 1px --border, a background a shade lighter than the card, monospace, small radius and padding, with ok / connecting / error as the text (and border) colour.
New Theme section near the top: five swatches (.theme-swatch with a miniature bg/panel/accent preview), the current one marked, click POSTs /api/admin/theme and applies at once, errors show on a status line; other admin pages' changes arrive live over /api/events, and the login card already wears the server's theme.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Server: one THEMES list; `theme` is read from the data directory's proddash.json like port/adminPasscode; GET /api/theme is public (dashboards need it before anyone logs in) and returns { theme, themes: [{ id, name, colors: { bg, panel, accent } }] }; POST /api/admin/theme { theme } validates against the list, writes the key into proddash.json keeping every other key, updates the in-memory value and broadcasts a `theme` shell event over /api/events.
style.css: four html[data-theme] variable sets appended at the end — Harbor (navy/cyan), Graphite (charcoal/blue), Ember (warm black/orange) and Daylight (light, for daytime FOH) — each swapping exactly the eight palette variables; :root stays Booth, the default.
shell.js: a self-contained theme section — the last-known theme (localStorage proddash:theme) goes on synchronously, then the server's answer, then live changes via the SSE event; one listener line added in subscribeEvents. Tile-configured colours are set on the tile and still win.
README: a Themes subsection under Configure and the `theme` row in the proddash.json key table.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Four pieces of work are about to run in parallel — the shell, the admin
page, the Timers module and the PCO Plan module — and three of them meet at
things that didn't exist yet. This writes those down first, so they are built
once, the same way.
- **Offering timers to the Timers module.** A module declares
`"provides": ["timers"]` and serves `GET /timers` (a cheap snapshot with
startedAt/elapsedMs/targetMs, an optional status line, optionally an SSE
stream). The consumer extrapolates locally, so no fast polling. Modules meet
over loopback through each other's routes, as PCO Plan already does to
follow ProPresenter.
- **Admin group controls.** `configGroups[name].help` becomes an ⓘ beside the
group title rather than a paragraph under it; `configGroups[name].toggle`
names a switch that renders in the title row as the group's on/off.
- **Switches, labelled with nouns.** On means shown or enabled; the label
shouldn't say so twice.
- **Themes** swap the shell variables; a tile's own configured colour is set
closer and wins.
Two shell hooks make it real: `/api/modules` forwards `provides`, and module
servers get `shell.port` in init/routes/tiles instead of guessing it from
proddash.json.
Clock: descriptions gone, "Show seconds" is "Seconds" (1.0.1).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Dragging a seam moved exactly the two tiles touching it, so a column edge
shared by several rows had to be dragged row by row and drifted out of line
in between.
A drag now moves that column boundary wherever it appears: every pair of
tiles meeting at the same x, in every row, travels with it. Each pair still
trades width within its own fixed span, so no row can push into another, and
the whole column stops at whichever row runs out of room first — one row's
minimum width holds the line for all of them. All rows move or none do; a
column edge that bends halfway is worse than one that won't budge.
Rows are only grouped by a shared seam. A row split somewhere else keeps its
own boundary, and a row where one tile simply spans across has no seam there
and stays put. Hovering a seam lights up the others that travel with it, so
what is about to move is visible before the drag starts.
Verified in the browser: two rows split at the same column move together from
one drag; rows split at different columns stay independent; a tall tile beside
two stacked ones resizes all three; and a 4-column drag stopped at 1 because
the lower tile's minimum width said so.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
It was a filled rounded pill among two corner brackets drawn as 2px lines —
three marks along the same edge in two different languages. Now it is a 2px
line in currentColor on the same 4px inset, with the same muted-to-text
hover, so the bottom edge reads as one set.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three things the dashboard was making people work for.
**The menu handle.** It was the ProdDash wordmark on a pill hanging past the
bar's bottom edge, which read as a step cut out of the bar. The wordmark goes
back in the bar where it belongs, and the handle becomes a small accent lip on
the bar's bottom edge — the same idea as a tile's notch — that grows under the
pointer so it stays easy to hit without being loud at rest.
That also fixes the animation Safari made a meal of: the old handle
transitioned font-size and line-height, animating text metrics frame by frame
on the way out. Nothing here animates text. The lip rides `--topbar-h`, the
bar's live height, so it tracks the bar's own collapse instead of running a
second animation against it, and its hover growth is width, height and
opacity only.
**Fullscreen.** Fullscreen is for watching, not arranging, so every tile's
title bar now folds away on the way in. A notch still peeks one open, and it
folds back ten seconds after the pointer leaves that tile — coming back to the
tile cancels the countdown. Peeking is a transient class, so a tile's own
hidden/shown preference is untouched and comes back with the screen.
**Tile settings.** Gear settings are display choices — a text size, a filter,
a colour — so they apply as they are changed and save with the layout. Apply
and Cancel are gone; close by clicking away, pressing Escape, or the gear
again. Typing is debounced so a remount doesn't land on every keystroke, and
an unchanged value remounts nothing (a text field fires input and then change
for one edit). The module guide says so now too.
Verified in the browser: settings apply live and persist, no-op changes cost
nothing, all three close paths work, and in fullscreen both heads fold, one
peeks, it folds back at ten seconds, returning the pointer cancels it, and
leaving fullscreen restores everything with no peeks left behind.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Installing is dragging ProdDash to Applications. The app carries the
dashboard and the Node that runs it, so nothing has to be installed first and
nothing can go missing underneath it.
ProdDash.app/Contents/
MacOS/ProdDash the launcher
Resources/node/bin/node Node 24.21.0 LTS, fetched and checksum-verified
Resources/app/ server.js, public/, modules/
Everything mutable stays where it already was, in ~/Library/Application
Support/ProdDash — and that now includes app/, the copy of ProdDash that
actually runs. The copy is not ceremony: ProdDash updates itself in place,
and nothing may write inside a signed .app. So the bundle holds a read-only
baseline and the data directory holds the running copy, which means both
routes to a new version keep working and don't fight — installing a newer
.app replaces the copy, while a copy that has already updated itself past the
bundle is left alone.
build.sh now assembles all of it: fetches Node against nodejs.org's published
checksums (cached in .cache/), embeds the server, compiles each module's
native helper so the Mac running ProdDash needs no compiler, signs nested
code before the bundle, and builds the disk image. --dev skips the payload
for launcher work, which still runs against a checkout.
Signing is set up for a Developer ID with the hardened runtime and
notarisation (--notarize), and falls back to ad-hoc when no such certificate
is installed. Entitlements are split by what each binary actually does: JIT
for node, audio input for the launcher and for module helpers that read one.
Built and run end to end: first-run install, bundled node, dashboard and
admin serving, re-seed in both directions (newer app replaces an older copy;
an updated copy survives), disk image mounts and its app verifies.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A pass for less prose and fewer boxes.
- Header shows the version and node; the port is already in the addresses
below it, and the module list belonged in the log.
- Settings: "Network" over the port (no explanation under it), "Startup"
instead of "Starting up", and no node chooser — the path stays visible,
since which node runs is the first thing worth knowing when it won't.
- Permissions: one line per permission — a status dot, its name, and the one
thing you can do about it. The standing explanation is gone and the
modules' own reasons moved to each row's tooltip. Local network results
still appear under their row after a check, with a short line and a link
only when nothing answered at all. Re-check sits centred at the bottom,
outside the group.
The app-level "notify me" switch goes with it: macOS's own per-app
notification setting already does that job, so notifications now follow the
permission alone, asked for by the row's Enable button rather than on first
run.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A native Swift app (launcher/macos, ~2 MB, no dependencies) that starts,
stops and watches over the server from the menu bar — and holds the macOS
permissions its modules need.
The permissions are the point as much as the buttons are. macOS grants
privacy permissions to the application that owns a process, and a child
inherits its parent's identity, so with the server started from here node —
and the Timers module's ltc-capture — belong to this app bundle and the
microphone and local-network prompts are asked, and remembered, as ProdDash.
Started over SSH or from a launchd job there is nobody to ask: the permission
is silently absent rather than refused, which is the "endless zeroes" failure
ltc-listener.js warns about.
The shell was already built for a supervisor: with PRODDASH_LAUNCHER=1 it
exits 75 instead of respawning itself when the admin page applies an update,
which the launcher reads as a restart request rather than a fault. Crashes
get a backed-off restart (1s, 2s, 5s, 10s) and four in two minutes stops the
cycle. The server is spawned into its own process group, so stopping it also
stops anything it started — an orphaned ltc-capture would otherwise hold an
audio device indefinitely — and signals, a logout and Quit all stop it first.
- Menu bar: ProdDash's own 2x2 tile mark, its accent tile lit green while
serving, amber mid-move, red when something needs a person. Start, stop,
restart, open the dashboard or admin, open at login, quit.
- Window: status and both addresses, plus Log (also written to
~/Library/Logs/ProdDash, kept a week), Settings, and Permissions.
- The port is written to the data directory's proddash.json — ProdDash's own
file — so the port set here holds however it is next started. Something
already on the port is reported, and offered as a takeover when it answers
like a ProdDash started another way.
- Modules declare what they need in their manifests ("permissions": [...],
documented in docs/MODULE-GUIDE.md and ignored by the shell); the launcher
lists exactly that, with each module's reason, asks for the microphone,
and — since macOS reports nothing about local network access — offers a
check that browses for services and tries every configured upstream.
Timers, ProdCom and ProPresenter modules declare their permissions (patch
bumps). Build with launcher/macos/build.sh; output is gitignored.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Resolves the admin form conflicts in favour of this branch's grouped /
collapsible / dependent-select additions (the base hunks were the timers
lane's, ported verbatim), keeps both sides of the README and module-guide
edits, and adopts main's versioning rules: the shell moves to 1.3.0 for the
new schema features (configGroups / instanceGroups, help, color fields,
grouped and dependent selects) and pco-plan 1.3.0 declares
"proddash": ">=1.3.0".
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Three changes to the ProdCom tile, all per-tile and persisted with the
layout (version 1.2.0):
- A title-bar button flips the transcript between top-down (newest at the
bottom, still the default) and bottom-up. It is CSS only — the entries
container becomes column-reverse, DOM order is untouched — but everything
that follows "latest" flips with it: auto-follow, jump-to-latest, the
near-latest scroll test, and the "New messages" pill, which moves to the
top of the tile and points up.
- The "Channels ▾" text button is now a funnel icon. It keeps what the
label carried: while any channel is hidden the icon takes the accent
color and a compact "3/4" count sits beside it, with the full wording in
the tooltip.
- Tile settings drop the channel-name filter and text size, both of which
the title bar already handles, and gain "Show channel icons" for the
per-entry avatars. Removing the channel filter made its code dead, so
channelFilter(), filteredOut() and the data-channel-name attribute went
with it — the menubar's hidden-channel set is now the only filter.
Text size keeps working without a schema entry: the A−/A+ buttons persist
it through saveInstanceSettings, the shell's settings popover only writes
schema keys so it survives an Apply, and the client falls back to 15px when
it is absent. Existing tiles have no saved showAvatars value and the check
is `=== false`, so they keep their icons.
Verified against tools/prodcom-mock.js: both directions render and
auto-follow correctly, the pill appears at the right end and jumps to the
newest entry, the count badge tracks the channel toggles, a tile bumped to
19px kept it across an Apply and a reload, a freshly added tile started at
15px with icons shown, and two tiles held independent state.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Section headers lose the bracket look (no accent bar, no rounded corners)
and span the full tile width, with text still aligned to the other rows.
A "Colors" group in the gear menu holds swatch pickers for the song key,
section headers and the current item, a switch for fading the header color
to the right (on by default), and the item-type tint. Colors apply as CSS
custom properties on the tile.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Renders <input type="color"> with the label beside it; the stored value is
always a #rrggbb string (3-digit hex widened, anything else falls back to
the schema default). Same styling in the gear popover and the admin form.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Top bar (template, item-type tint, key color) and the whole show/hide
checklist move from admin config to instanceSchema, grouped in the gear
popover ("Top bar" open, "Show / hide" folded, three columns). They are
browser-side choices stored with the layout, not server state. Admin now
holds only credentials, service type, plan, poll interval and the
ProPresenter follow switch. noteCategories moves to the tile too.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The gear popover learns the admin form's conventions: an instanceSchema
field's `group` starts a labeled subsection, `help` puts a line of guidance
under a control, and a manifest `instanceGroups[name]` adds
{ collapsed, columns, help } for that group (rendered as <details> with a
grid body; container queries drop to two or one column in narrow tiles).
/api/modules now exposes instanceGroups. Lets a module keep a long
show/hide checklist per tile — where display choices belong — without
the popover turning into a scroll of thirty loose checkboxes.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
- Drops the separate setup page: the service type is a select grouped by
Planning Center folder path (GET /service-types), the pinned plan a
select of upcoming/recent plans that follows the chosen service type
(GET /plans?serviceTypeId=), credentials sit in a group that folds once
saved.
- Show/hide moves from per-tile settings to a server-wide checklist in
Admin (three columns, folded by default): top bar, series, part, service
type, date, service time, total length, running clock, countdown,
headers, item numbers, start times, length, type badge, key, arrangement,
author/CCLI, description, notes, pre/post-service items, now indicator,
item runtime, dim completed, ProPresenter line. Tiles keep only text
size and auto-scroll.
- Top bar text is a template with placeholders ({series} • {part} |
{serviceType} • {date} by default; • separates droppable segments, |
starts the second line).
- Song key: larger, admin-chosen color (blue by default), always right
after the title and before the type badge. Section headers are
full-width accent bands. Optional background tint per item type.
- tools/pco-mock.js plan titles are part numbers, as in a real series.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Ports the timers lane's admin schema additions unchanged (field `group`,
`showWhen`, type `switch`, `optionsRoute`) so the two branches merge
cleanly, and extends them for modules whose settings are richer than a
handful of fields:
- manifest `configGroups[name]` = { collapsed: true | false | "whenSet",
columns: 2–4, help } — a credentials block that folds once it holds a
saved secret, a show/hide checklist laid out as a compact grid
- `help` on any field: one line of guidance under the control
- select options may carry `group` → rendered as <optgroup>s (a folder
tree browsable inside the select)
- `optionsDependsOn: "<key>"` refetches a select's options with that
field's current value as a query parameter, now and on every change
- a save re-runs the card's dynamic option fetches, so a list that needed
the just-saved credentials fills in without a page reload; a blank
"— choose —" is no longer added when the module offers its own blank
option
Documented in docs/MODULE-GUIDE.md.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Square at the top, rounded at the bottom, dropping 8px past the bar's
bottom edge with "ProdDash" centred in the bar part — the shape the mockup
asked for, instead of a pill floating mid-bar. Hidden state unchanged: a
slim lip at the top-left edge.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RNYUv9EyrGSFmCnmH7WYKH
A new Display-group select (Default / Monospace) sets the font for the
timer and LTC digits. Monospace keeps the numbers from shifting sideways
as they change — easier to read on a fast-moving countdown or timecode.
Flows through server state → client applies a `tm-mono` class and re-fits
the digit sizing; both timer and LTC values (.tm-time) honor it.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The show/hide handle is now the logo itself: a translucent accent pill
wrapped around "ProdDash" at the left of the bar, centred on the text, with
"Dash" in the background colour so it reads as a cutout inside the green.
The pill is far less opaque than the old solid tab (38% idle, brighter on
hover). When the bar hides, the text folds away and the pill shrinks to a
slim lip at the top-left edge, exactly where the old tab sat. An invisible
in-flow twin of the logo keeps the bar's layout unchanged.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RNYUv9EyrGSFmCnmH7WYKH
Simplify LTC to its one working source — the built-in audio listener — and
make the admin UI reflect it.
Module:
- Remove the stage-display websocket source (stage-ws.js, stagePassword)
and the /v1/timecode/status API probe — both dead on ProPresenter 21.4.
- Remove the remote-reader ingest (POST /ltc, ingestToken, tools/
ltc-reader.js): the module owns LTC end to end now. The decoder selftest
moves to modules/propresenter-timers/ltc.selftest.js so the engine stays
covered.
- composeLtc collapses to the listener: running/stopped/nosignal, or a
clear note when disabled / no device / capture failed / starting.
- Manifest: ltcEnabled is now a `switch`; ltcDevice and ltcChannel carry
showWhen:"ltcEnabled"; ProPresenter and LTC fields carry `group` labels.
Shell (admin) — backward-compatible additions, coordinate with the shell lane:
- New `switch` field type (boolean shown as the enable-toggle control).
- `group`: consecutive fields render under a labeled subsection.
- `showWhen: "<key>"`: a field is shown only while that boolean/switch is
on (and actually collapses — .field[hidden] override, since .field's
display:flex outranks the [hidden] UA default).
- applyConfigPatch coerces `switch` as boolean. Documented in MODULE-GUIDE.
Verified: engine selftest (29 checks); listener decodes BlackHole loopback
(running 30fps, ageMs); device picker lists inputs; POST /ltc now 404s;
admin shows two labeled groups, the switch, and device/channel appearing
only when the switch is on (computed display none↔flex confirmed).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The LTC card repainted only on server anchors (~4x/s), so the frame
counter jumped in ~7-frame steps. Now the client counts every frame
locally between anchors and re-syncs on each one — frame-accurate and
smooth, with no extra network traffic.
- Server anchors carry ageMs (how old the decoded timecode is) so clients
can pin the anchor to their own clock. ltc-listener reports it while
running; composeLtc passes it through (listener) or derives it (reader).
- Client does exact SMPTE frame<->timecode math (parseTc/tcToFrames/
framesToTc, incl. 29.97 drop-frame minute counting) and advances the
display in a requestAnimationFrame loop at the LTC's own rate.
- Robust when rAF is throttled (backgrounded tab): render() paints the
extrapolated frame at the 4x/s anchor rate too, so the display never
freezes — the rAF loop only refines it to every frame when visible.
- Anchor jitter can't tick the counter backwards (small regressions
clamped; genuine seeks/re-strikes pass); freshness cap stops freewheel
into phantom frames when anchors stop arriving.
Verified: SMPTE math against the encoder over 200k drop-frame frames plus
round-trips; the extrapolation pipeline simulated at 30/25/29.97DF showing
strict +1-per-frame advance and correct DF boundary crossing.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adopts the new contract: the manifest declares "proddash": ">=1.2.0"
(this module's admin device picker needs the optionsRoute select support
this branch adds), module version 1.2.0, and the shell version bumps
1.1.0 → 1.2.0 for that admin capability. The optionsRoute paragraph
merged into the guide's new schema section cleanly.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
ProdDash now has a version (package.json, 1.1.0) and modules declare the
shell versions they need ("proddash": ">=1.1.0" in the manifest). A module
whose requirement this shell doesn't meet is listed in the admin page but
never loaded, and cannot be installed from the catalog.
Modules come from two places: bundled with the checkout (modules/) and
installed from the admin page into the data directory (<data dir>/modules/).
Same id in both → the newer version runs; a downloaded copy the bundled one
has overtaken is removed. The admin page's new "Available modules" section
lists modules in the ProdDash repo that aren't installed here — one small
tar.gz of the branch is fetched (at most every few minutes, only while the
admin page is open or for the periodic update check) and read with a
dependency-free tar reader (ustar, GNU long names, pax headers; the pax
global comment carries the commit sha). Install downloads the module folder
into the data directory and mounts it immediately; a module card offers
"Update to vX" when the repo has a newer version. Uninstall removes an
installed module's files; a bundled module is hidden instead (its folder
belongs to the app) and can be restored from Available modules without a
download. Settings are kept across uninstall/reinstall.
Shell updates: the server compares the repo's package.json version and
commit sha with its own (git rev-parse when the app folder is a checkout),
shows a banner in the admin page, and "Update now" brings the app folder up
to date — git pull --ff-only for a checkout that is on the repo branch with
no local changes (file modes ignored), otherwise the archive is unpacked over
the folder (runtime state and .git untouched). The server then exits with
code 75 and starts its own replacement unless PRODDASH_LAUNCHER is set; the
replacement waits for the port to free up (EADDRINUSE retry). The page polls
until the new version answers. PRODDASH_REPO (owner/repo#branch),
PRODDASH_ARCHIVE_BASE and PRODDASH_NO_REMOTE_CHECK configure/disable it.
Docs: MODULE-GUIDE gets the rules every module follows (all settings in the
admin page, dependencies shipped inside the module, everything runs inside
the module's server entry — no helper programs, declare the ProdDash
requirement and bump the version), the proddash/version manifest fields and
the new installation story; README gets "Installing modules and updating";
CLAUDE.md summarises the rules.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RNYUv9EyrGSFmCnmH7WYKH
Main now keeps a per-machine data directory whose proddash.json can override
the port, so reading the repo's config file was no longer a safe way to find
the ProPresenter module's loopback feed. Guess the port the way the shell
does (PORT, data-dir proddash.json, repo proddash.json, 24500), then confirm
it from the Host header of the first tile request; a learned port that
refuses the connection (a reverse proxy) is blacklisted and the guess kept.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
New module modules/pco-plan: reads the current plan for a chosen Planning
Center Services service type (PAT or OAuth token), streams items with start
times, lengths, keys, descriptions and item notes to tiles (every detail a
per-tile show/hide setting), and follows the propresenter-now-next module's
live item over loopback SSE to highlight the current plan item and track
elapsed vs planned runtime. Ships its own setup page
(/modules/pco-plan/setup.html) to browse folders and pick the service type
through the shell admin API, plus tools/pco-mock.js for development.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The menu bar's show/hide tab was centred on the screen, exactly where a
centred tile's own title-bar tab lands once both bars are hidden: two
identical tabs on top of each other, with the tile's unreachable. The tab
now sits at the left under the ProdDash brand and wears the accent colour,
so it never coincides with a tile tab and reads unmistakably as the
dashboard's. To keep it off a top-row tile's status dot and title it
straddles the bar's bottom edge while the bar is shown (mostly on the bar,
a short lip below) and becomes a slim tab at the very top edge when the bar
is hidden.
The Admin link is hidden in fullscreen, including when the menu bar is
brought back there.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RNYUv9EyrGSFmCnmH7WYKH
Start ProdDash.command now prefers ./runtime/bin/node (production machines
have no system Node) and kills whatever already holds :24500 before
starting — so relaunching from the GUI (which the LTC listener needs for
microphone permission) is one double-click, not a port-conflict hunt.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The LTC listener now lives inside the module: it spawns the native
ltc-capture helper (moved into the module folder, compiled on first use,
exact-name device matching) for the admin-selected input device, extracts
the configured channel, and decodes SMPTE LTC in-process (ltc.js — the
engine, shared with tools/ltc-reader.js, which remains the remote-machine
option posting to /ltc).
Admin config: an "LTC listener" switch, an audio-device select populated
live from the new GET /ltc/devices route, and a 1-based channel number.
The switch off means LTC is gone entirely: no LTC entry in the Add-tile
picker, no LTC row in the Timers menu or All-timers grid, no stage/API
probes, and POST /ltc answers 409. Source precedence with it on:
listener-with-data, fresh remote reader, listener's own no-signal verdict,
then the legacy API/stage chain.
Shell (admin page): `select` fields learn `optionsRoute` — options fetched
from a module route at render, saved value kept selectable when the route
or device is gone. Documented in MODULE-GUIDE; backward compatible.
Coordinate with the shell lane before its next admin.js change.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Tiles get a third resize handle: a pill on the bottom edge (hover-only,
between the two corner handles) that moves that edge. With nothing directly
below, it changes just the tile's height. When other tiles share the edge,
the drag trades rows with them so the column keeps its total height — the
vertical counterpart of the seam pill. Clamped to every tile's minimum
height; blocked by collisions like the corner handles.
Fullscreen auto-fit reworked so columns actually reach the bottom of the
screen. The old rule only stretched tiles whose bottom edge was within two
rows of the viewport bottom, which on a display taller than the browser
window the layout was designed in left a blank band below every column.
Now the bottom-most tile of each column stretches to the bottom regardless
of distance. Short tiles (h <= 2, the clock) are never stretched into a
ribbon: under another tile they drop to the bottom edge and the tile above
grows into the gap (stopping at anything else below it); alone in their
columns they stay put. The fit re-runs from a ResizeObserver on the grid
area, so the browser's animated fullscreen transition and the menubar
collapse both settle into a correct fit (replaces the resize/transitionend
listeners).
README documents the pill and the fullscreen behaviour.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RNYUv9EyrGSFmCnmH7WYKH
Real Dante capture produced a bit-error frame whose frames field read 35
yet passed the lax range check (f<=59, hours<=39), and max-frame-based
rate detection then reported 36 fps to the dashboard permanently. Now:
every BCD digit is validated and fields respect hard ceilings (f<=29,
h<=23); nominal fps locks only from two agreeing second-rollovers (the
frame number preceding f=0 with seconds+1), so one glitch can't fake it
and a genuine rate change re-locks within two seconds; the drop-frame
flag gets two-frame hysteresis. Selftest reproduces the forged-35 frame
plus PCM glitches and pins fps at 30 throughout.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Production machines rarely have ffmpeg/sox, and the LTC reader only needs
"device N → s16le stdout". This ~200-line Swift CLI (compiles with the
Xcode CLT already on the target: swiftc -O -o tools/ltc-capture
tools/ltc-capture.swift) enumerates input devices (--list), pins an
AVAudioEngine tap to one picked by name substring (--device), and streams
interleaved s16le at the device's own rate/channel count, printing the
matching --sr/--channels reader flags to stderr.
Verified by a real loopback on BlackHole 64ch: synthesized LTC played into
the device, captured back (channel 2 of 64 @ 44.1 kHz), decoded to
running/stopped on live tiles with zero drift over 14 s.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Connection settings, API keys, enabled flags and named layouts were written
to config/ inside the checkout, so a re-clone or reset from main started the
server with nothing configured. They now live in a data directory outside
the app folder:
Windows %APPDATA%\ProdDash
macOS ~/Library/Application Support/ProdDash
other $XDG_CONFIG_HOME/proddash, else ~/.config/proddash
PRODDASH_DATA_DIR overrides the location (a relative path resolves against
the app folder, so "config" restores the old spot). The first start after
upgrading copies an existing config/modules.json and config/layouts/ into
the data directory, so nothing has to be re-entered. If the directory cannot
be created or written, the server warns and falls back to config/ rather
than refusing to start. A proddash.json in the data directory overrides the
checked-in defaults (port, adminPasscode).
The startup banner prints the directory in use, /api/admin/state exposes it,
and the admin page shows it under the Modules heading. Admin config and
layout writes that fail on disk now report the path instead of a generic
server error. README, MODULE-GUIDE and CLAUDE.md updated.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RNYUv9EyrGSFmCnmH7WYKH
