The control server answered a line that wasn't JSON with an error and read
on. A web page can make the browser send an HTTP request to the loopback
port; its request and header lines were skipped that way and the JSON lines
of its body ran as commands. The first line that isn't JSON now closes the
connection before anything after it is read. Clients that send JSON lines
(MCP --attach, scripts) are unaffected.
The message now comes from the UI crate (dialogs::window_failed) through
tr() and trf(), with entries in every catalog, and the app picks the
language from the preferences before the event loop starts, so the box
reads in it even though no frame was ever drawn. The commands to type stay
as written.
Recovery copies were written with plain `save_bytes`, so a file
recovered after a crash came back without its version history and
branches, even though `file.recover` reads both. Copies now go through
`save_bytes_with_history` like a normal save, with each open branch
tab's latest edits (`branches::current`), since `sync` only puts those
in before the next command.
A copy was also only redone when the main file's revision changed, and
a file with edits only in a branch tab didn't count as changed. Both
checks now include the file's branch tabs.
A minimised or covered main window draws no frames, and control requests
were only answered inside a frame, so every control-channel and attached
MCP call (save_file among them) timed out after 60 s while the app was out
of sight. eframe still calls the app's logic then whenever a repaint is
requested, which the control server already does for each request.
The desktop app's logic now answers, while the window is hidden, the
requests that need no frame: engine.execute, engine.commands,
document.inspect, node.get, window.list and ui.render. Screenshots, input
and the interface's own state still wait for the window, and their
timeout now says that the window has to be on screen.
Closes#269
Release builds on Windows have no console, so an eframe::run_native error
(no usable graphics adapter, a device the driver refuses) went to an unseen
standard error and the app quit silently. The error is now logged and shown
in a message box with a hint: unset WGPU_BACKEND when it is set, otherwise
update the driver or try another graphics API (OpenGL on Windows), and
where to report it. With --control the box is skipped so agents aren't
blocked.
Windows' preferred-language list is also the input-language list: a language
added for a keyboard sits in it, and taking the first entry with a catalog
turned the interface German on a Polish system (a German keyboard) while the
system and the other apps stayed Polish. Only the first preferred language now
decides; when it has no catalog the interface reads English. macOS'
AppleLanguages and gettext's `LANGUAGE` still take the first language with a
catalog. Fixes#291.
Generated with Codebuff 🤖
Co-Authored-By: Codebuff <noreply@codebuff.com>
The Recents section shows recent files as cards: the picture each file
carries (thumbnail.png in .figcraft and .fig, an image file itself) fitted
on its first page's colour, a divider, and a footer with the file's icon,
name and "Edited ... ago" (or "File not found"); the full path on hover.
Pictures are read off the frame's thread, only for cards in view, cached
by path and modification time, capped in bytes and pixels; missing,
damaged or oversized ones give a placeholder. They fade in, and cards
show a skeleton until their file has been looked at.
A sort menu (Alphabetical, Date created, Last modified; either order) and
a grid/list toggle sit above them. The list is a Name, Last modified and
Created table whose headers sort. View (per section) and sort are kept in
the preferences and set with ui.set {homeRecentsView, homeRecentsSort,
homeRecentsOrder}. A click selects, a double-click or Enter opens, arrow
keys move the selection; a right-click offers Open, Show in folder and
Remove from recents (file.openRecent {remove}).
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bflatastic <bombayapirat@gmail.com>
state.json holds online owner tokens. The temporary file is now created
0600 (and narrowed if a crash left one behind) before anything is written,
and the rename carries that mode to state.json, so an older
world-readable file is replaced on the next save.
Co-authored-by: Justin John <babyjjbabyjj549@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
On Windows, eframe offered wgpu every backend and wgpu ranked Vulkan
first. On a laptop with two AMD GPUs whose drivers are years apart, the
Vulkan driver faults (0xc0000005 inside amdvlk64.dll) right after the
first frame, so the app died as its window opened. An access violation
in a driver thread is not something the app can catch.
Offer wgpu Direct3D 12 on Windows, with OpenGL behind it for hardware
without a Direct3D 12 driver. WGPU_BACKEND still names the backend on
every system, and other systems keep eframe's defaults.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
SVG and PDF exports check the visible subtree's depth before bounds or drawing and refuse, with an actionable message, a document nested past 256 levels instead of silently leaving the deep layers out; nothing partial is written, and an inner layer can still be exported. The CLI and MCP report the same error. Fixes#208.
Also: on Windows, figcraft and figcraft-cli link with an 8 MB main-thread stack (Windows gives 1 MB where Linux and macOS give 8), so an export at the depth limit no longer overflows it.
Co-authored-by: bflatastic <bflatastic@users.noreply.github.com>
MCP and control-channel lines are bounded while they are read (32 MiB; an oversized frame gets an error and the stream closes without being drained). The loopback control listener caps connections at 8, applies 60 s socket read and write timeouts, caps both request queues at 8 and discards requests whose client stopped waiting. Fixes#201.
The recovery copy of one of your online files is no longer discarded when
the file reopens from its server. It is kept until the outcome is known:
- the server has everything in it: the copy goes;
- the copy has changes that never reached the server and nothing else
changed there: they are applied on top as one edit and sent to everyone;
- anything else (others changed the file, the server forgot the client):
the copy opens beside the file as "<name> (recovered copy)", with a toast;
- the server can't be reached: the copy opens as an unsaved document.
The note beside a copy is written after it and records who the app was in
the file and the server's last change it had. The app joins again as that
client (the first hello resumes; a new sync client numbers its batches
after the ones the server processed), so the layers it made are its own to
send. The old copy is removed only once this run has copied what it became.
collab.recoverFile; a tooltip on the tab mark; docs.
A file put on a server without the server's key stays its owner's after its
tab closes or the app quits. The app records it with the preferences (server,
id, name, owner token, the owner's links, what others join with; at most 200;
validated when read back) and never shows the token: not in command results,
ui.inspect, logs or debug output.
- Home > Online lists your online files: open one as its owner (a new tab, or
its tab comes to the front), start and stop sharing, copy the code or link,
rename on the server, remove from the server (confirmed), forget here. One
request per server when the section opens; offline and gone states.
- collab.files, collab.openFile, collab.forgetFile; collab.manageFile takes
its server and token from the record and gains share / unshare.
- Server: Admin owned (a compatible addition within protocol version 2) says
what became of the files a list of owner tokens owns. Older servers leave
the list as recorded.
- Tabs mark files in a live session (shared, or open only to you); toasts
about a file that isn't in front name it; after a crash a recorded file
reopens from its server as its owner.
- Draft catalogs (de, fr, ja, ko, pt-BR, zh-Hans) translate the 71 strings
and 5 command labels they lacked and the new ones; es and es-419 follow.
- Docs: collab-protocol, server, mcp, control-protocol, parity pages, gaps,
ROADMAP.
The hicolor theme, the .ico and the window icon now show the whole tile with its rounded
corners, like the other Crafting Apps, instead of a crop that cut most of the corners off.
The owner's engraving-style chameleon drawing, traced to three colours (ink, paper,
FigCraft green #02da96) on the shared 512-unit tile. assets/app-icon/figcraft.svg is the
master; cargo xtask icon now renders the PNGs (macOS grid), the hicolor theme and the .ico
(tight crop, 16-256 px) from it with resvg instead of drawing the old geometric mark.
Off macOS the window icon uses the tight 256 px render.
- **Installed plugins survive a crash.** Importing, installing, updating or removing a plugin, and a
plugin changing its settings, now sets `Runtime::unsaved`; at the end of each frame the app hands
the preferences to a new `Services::save_prefs`, and the desktop app writes `state.json` at once
(temporary file + rename, the same atomic write the quit path uses, now shared). Nothing waits for
quit any more.
- **The UI never waits on the network.** New `Session::start_background(id, params)`: the command
runs as usual until it makes a server request; the request goes out on the link's own thread
(`LiveLink::background`, implemented by the desktop `NetLink` with one short-lived thread per
request) and the command stops; when the answer arrives (`Event::Net`, taken by `live_poll`, which
the network's waker already triggers) the command runs again from the start on the file it
started on, answered from the answers so far, until it needs no new request. All server requests
go through one funnel (`Session::net`), so `plugin.directory.*`, `library.publishToServer`,
`library.add` (links), `library.usage` and library fetches work this way without being split.
A re-run of `library.publishToServer` caches the library it actually sent (a fresh copy differs
in its time and, on a first publish, its id). One job per command; `cancel_background` drops a
late answer.
- **UI:** `FigApp::run_background(ctx, id, params, done)` with a boxed completion. Manage plugins >
Browse shows "Fetching from the server..." with Cancel instead of Browse; the install confirmation
shows "Installing..." with Cancel and stays open until the download is in (an error leaves it open
to retry); row buttons wait during an install; closing the window cancels. Assets > Libraries:
Publish to server shows "Publishing..." + Cancel, Show usage and pasting a team library link show
"Fetching from the server..." + Cancel. No polling repaints: the window draws again when the
answer's waker fires; pending state is a scan of a short Vec, no per-frame allocation.
- **Programmatic callers are unchanged:** MCP, CLI, control channel and scripts still run the same
commands synchronously (and the browser link, which has no background requests, too).
- New string "Installing..." (en/es); docs: plugins.md (immediate saves, Browse/Install never wait),
server.md (team library requests never wait).
The part of gap 15 that doesn't need video, from the public documentation of prototype triggers, plugins/widgets and prototype expressions.
- **Controller buttons on key/gamepad triggers.** A key trigger's key can be a controller button: `Gamepad A` ... `Gamepad Home`, the standard gamepad layout's 17 buttons (face buttons by position, shoulders, triggers, View/Menu, stick presses, d-pad, Home). The engine keeps its input-agnostic model (a key trigger with a key name; `figcraft_doc::proto::GAMEPAD_BUTTONS`, `gamepad_button`, `is_gamepad_button`, `uses_gamepad`). The interaction editor gets a Controller menu beside the key field; the trigger list shows the button's own name.
- **Reading controllers.** A new `Services::gamepad_held()` (bitmask of held buttons; default none). The player polls it only while the presented prototype has a controller trigger and the window has the focus (about 30 times a second), plays each press once (held buttons don't repeat), and fires the first matching key/gamepad trigger on the screen like a key.
- Desktop, Windows: Windows.Gaming.Input (Xbox-compatible controllers) through the `windows` crate the app already depends on (one more feature flag, `Gaming_Input`; no new crate).
- Browser: the Gamepad API through `web-sys` (features `Gamepad`, `GamepadButton`, `GamepadMappingType`; standard-mapped controllers only).
- macOS, Linux, FreeBSD: no controllers yet (documented gap).
- Control channel: `ui.input {events: [{gamepad: "A" | "Gamepad A" | 0}]}` presses a button in the prototype playing (presenter, else preview pane); an error when nothing plays.
- **REST import** maps controller triggers (`device` other than `KEYBOARD`) to the button at the code's standard-layout index instead of skipping them; two buttons at once or an unknown code is skipped with a warning (`.fig` goes through the same path).
- **Widgets react to clicks in a presented prototype.** A click on a widget's clickable part in the presenter or the preview pane (when no interaction on the layer takes it) calls the plugin back through the canvas's own callback with the same sandbox limits, on the player's copy of the file (`widgets::click_on`, a scratch session holding the copy and the installed plugins): the widget changes in the prototype only, the file and its undo history stay untouched, restarting starts the widget over. Failures (plugin missing or failing) toast; toasts now also show over the full-window presenter.
- **Variable mode lookups.** Lookups (`accent[Dark]`, `mode(Theme)`), set-variable-mode affecting later expressions in the same interaction, and every operator the documentation lists already held; they are now behaviour cases. New `proto.evaluate {expression | expr, id?}` (MCP and agents) computes an expression where a layer's interaction would run, without changing anything. No other cheap expression gap found: every documented operator (numbers, text join, booleans, comparisons, negation, mode lookup) already existed.
- Docs: prototyping-parity (new rule rows, features, revision), target-app-parity area 13 and totals, ROADMAP, gaps.md gap 15, roadmap.md M5, mcp.md, plugins.md, control-protocol.md.
Dependencies: none added. Windows uses the existing `windows` 0.62 dependency with the `Gaming_Input` feature; the browser uses the existing `web-sys` with three more features. The usual Rust controller library (gilrs) was not taken: its Linux backend links the system's libudev (a C library, needing headers on CI and the FreeBSD job), which breaks the pure-Rust rule.
A self-hosted plugin directory, with no third-party service: `figcraft-server` keeps plugin bundles,
the app browses, installs and updates from it, and `figcraft-cli` publishes to it.
- **Bundles** (`figcraft_plugin::bundle`): a plugin's manifest, module and up to 32 plain-named
assets in one uncompressed, length-prefixed binary form (nothing can expand beyond the bytes
sent). Its SHA-256 names it. `Bundle::decode` never allocates beyond its input and refuses
trailing bytes; `Bundle::check` parses the manifest (refusing a plugin API this build doesn't
run), requires a version of 1-4 numbers, compiles the module without running it (only the
sandbox's imports allowed) and checks asset names (no paths, no dot files, no Windows device
names, unique ignoring case). Manifests gain an optional `author`.
- **Protocol**: a new connection kind next to managing and team-library connections:
`OpenPlugins {key}` then `Plugins {list | fetch | publish | unpublish}`. Browsing needs no
credential; publishing and unpublishing need the server key (a wrong key is refused). Older
servers refuse the connection and the app says the server has no plugin directory. The protocol
version is unchanged; old clients are unaffected.
- **Server** (`apps/figcraft-server/src/directory.rs`): `<data>/plugins/index.json` plus
`<data>/plugins/<sha256>.bin`, so no publisher-chosen name reaches the file system. Every publish
is checked in full and must be a newer version; caps `max_plugins` (versions, 0 turns it off),
`max_plugin_mb` (<= 32), `max_plugins_mb`. Fetches re-hash the file before sending (a damaged
bundle is never sent) and share a download budget (~32 MiB/s) so a flood is refused, not queued.
Damaged index entries are skipped and reported, never deleted; an unreadable index lists nothing
and refuses writes until the operator fixes it. Operator moderation:
`figcraft-server plugins [list]` and `plugins remove ID [--version V]` (offline or through the
running server). Logs name ids and versions, never keys.
- **App**: Plugins > Manage plugins gets Installed / Browse tabs. Browse lists a server's plugins
(version, author as typed, size, access) with a search; nothing is fetched until Browse is
pressed. Install / Update open a confirmation listing the permissions asked for, the source and
the SHA-256; only its button downloads. The download must match the listed SHA-256 (and the one
shown), then is checked like an imported plugin before anything is written. Installed plugins
live per user under `<settings>/FigCraft/Plugins/<hash of id>/` and run in the same sandbox;
Remove deletes the folder.
- **Commands**: `plugin.directory.list {server?, query?}`, `plugin.directory.install {id, version?,
confirm, server?, sha256?}` (refused without `confirm: true`; never opens a dialog),
`plugin.directory.publish {path | plugin, server}`, `plugin.directory.unpublish {id, version?,
server}`. `ui.set {pluginsTab, pluginConfirm}` on the control channel.
- **CLI**: `figcraft-cli plugin publish FOLDER --server ws://host:port/` with the key from
`FIGCRAFT_SERVER_KEY` (or `?key=`), never printed.
- Docs: plugins.md (directories, publishing, commands, manifest), server.md (operating the
directory: access, checks, caps, moderation, backups, security), collab-protocol.md, mcp.md,
cli.md, control-protocol.md, target-app-parity area 18 and Ecosystem, ROADMAP, roadmap.md M11,
gaps.md gap 6. Plugins & widgets 55%/32% -> 60%/35%, 10-30 h -> 8-26 h.
- New UI catalogs `crates/ui-egui/locales/fr.toml`, `de.toml` and `pt-BR.toml`: every UI string (1,351 keys, 36 with plural forms), every command label (519) and every main-menu wording (29) translated, 1,899 entries each. Machine-drafted from our English, marked `status = "draft - needs native review"` like Spanish, so Preferences shows the draft note. Placeholders are identical to English; `|context` keys are translated per context (sizing "Fill", stroke "Join", glass "Light", open comments).
- Language conventions: French uses no-break spaces (` `) before `: ; ? !`, inside guillemets and before `%`; German uses formal address and German quotes; Brazilian Portuguese uses Brazilian vocabulary (arquivo, salvar, tela, voce).
- Not added to `COMPLETE`: that list forces every later PR to translate new strings into each complete language, so the three drafts are registered as incomplete and fall back to English per key; Spanish stays the only enforced language. `docs/translating.md` says so and how to list what a draft lacks.
- Plurals: `plural_one` now takes 0 as singular for French and Brazilian Portuguese (CLDR integer rules); English, Spanish and German stay `n == 1`.
- Language choice: Preferences > Language lists Francais, Deutsch and Portugues (Brasil) by their own names (from the catalogs). System now walks the OS's preferred-language list and takes the first language with a catalog: Windows' full `GlobalizationPreferences.Languages`, all of macOS's `AppleLanguages`, gettext's `LANGUAGE` then the locale variables on Linux/BSD; the browser still uses `navigator.language`. A tag with only a regional catalog reads it (`pt-PT`, `pt` -> `pt-BR`). `pref.language` docs list `fr|de|pt-BR`.
- Overflow (German is the longest): `widgets::dropdown` now keeps its label on one line, ending it with an ellipsis and showing it in full on hover (it used to wrap mid-word, e.g. the auto layout sizing and stacking dropdowns); `elided_dropdown` merged into it. Two-column sublabels in the right panel and the home screen's cards elide the same way. English is unchanged.
- Docs: localization-parity (languages, string counts, status, hours), translating, target-app-parity, ROADMAP, gaps (gap 5), roadmap (current focus 4), mcp. Localization 20% -> 30%, 45-75 h -> 40-65 h; ready for real work stays ~52%; full parity ~289-564 h -> ~284-554 h; beta ~170-305 h -> ~162-292 h.
Found and fixed in a Windows pass of everyday workflows: unsaved documents are now copied in the background (at most every 10 s) and offered back after a crash or forced quit (file.recover); Ctrl+arrow / Ctrl+Backspace / Ctrl+Delete text motion follows Windows and Linux conventions (keymap::text_motion); number-field scrubbing keeps its value; typing in a Tab-focused field works; default linear gradients and their handles behave; Escape closes dialogs without deselecting; clipped panel hints wrap; the unsaved-changes dialog is translated.
File > New window, Open in new window on a tab, and dragging a tab out open files in separate OS windows (egui deferred viewports, redrawn only when needed), sharing one session, preferences, libraries and clipboard. Closing a window asks about unsaved files (the main window now asks too); closing the last window quits. window.new / window.list / window.focus and per-window screenshots for the control channel and attached MCP. The browser keeps one window.
UI strings come from per-language TOML catalogs compiled in (English text as the key and fallback, named placeholders, one/other plurals, context suffixes); command labels are translated at display time while ids, CLI, MCP and control output stay English; accessibility names follow. Preferences > Language (System / English / Español, live; the system language from the OS or the browser), pref.language for scripts. A key check test guards the catalogs. The Spanish catalog is a machine draft awaiting native review; docs/translating.md explains adding a language.
The last three catalogued menu items got commands: File > Open recent (recent files kept in preferences, capped, missing files greyed and dropped when picked, Clear recent), outlines for the selected layers only, and mask outlines. Menu breadth 100%.