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>
They ran on every push to main and on any PR touching Cargo.toml/Cargo.lock,
which is nearly every PR. See craftrules release/playbook.md §2.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
With the Move tool, the filled area under the pointer is hatched with "/" lines in the selection colour (1.4 px), clipped by the area's own fill rule so holes stay clear. The hatch goes when the pointer is 3.5 px from a point or 2 px from a segment (screen pixels at any zoom); between those clearances and the 6 px hit radius the area wins. Bezier handles, the transform box's handles and other edited layers' parts come first. Line spacing follows octave folding: 8 px at every power-of-two zoom, growing with the zoom and stepping back at sqrt(2) * 2^n. The pattern is anchored to the vector's origin, so it stays on the shape when panning, and several selected areas share one pattern. The last area listed wins where areas overlap, and only vectors with a visible fill have areas.
A click selects an area with its points; they move with it (drags, nudges, the Design panel) but keep the unselected look and show no handles. Shift adds an area or takes a selected one out (its own points go, points selected on their own or shared with another selected area stay). A press on a selected area drags everything selected, as one undo step; Escape puts it back. Drags show no hover hatch. Any other press without Shift drops the areas and the points selected only through them, and a click on an empty spot such as a hole clears them and stays in edit mode. Inside the transform box around points selected on their own, the box keeps its drag unless the area is already selected. Area indexes are dropped when the network's topology changes; moving points keeps them.
The network gets region walking, an inside test per region by its own fill rule, the region under a point, a region's points and a topology fingerprint (vector_regions.rs). The engine does hit testing and lays out and clips the hatch in screen space (vectoredit/regions.rs); the canvas only draws the lines. vector.edit.select takes regions, and vector.edit.state reports regions, regionVertices and regionHover. Tests: vector_regions_tests (doc) and regions_tests (engine), covering fill rules, overlap, curves, reversed segments, clearances at 100% and 400%, selection, drags, undo and cancel, and the hatch spacing at seven zoom levels, its direction, anchoring, coverage and clipping.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A `.figcraft` image member that is damaged, too large or missing was skipped
without a word, and saving the file again dropped it for good. The load now
lists each lost image (up to 20, then a count) with its hash and the reason,
file.open puts it in `repairs` and sets a session message, and the app shows
that as a toast.
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
Co-authored-by: bflatastic <bombayapirat@gmail.com>
usvg unpacks SVGZ with no limit and parses SVGs embedded in `<image>` data
URIs with no shared depth or byte budget, and the nesting check ran on the
compressed bytes. FigCraft now unpacks SVGZ itself within 64 MiB, checks the
real document, and resolves embedded SVGs with its own data resolver: one
32 MiB budget of unpacked bytes and at most 4 levels, an embedded document
past a budget is left out. Rasters still resolve; linked files stay unread.
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
Co-authored-by: bflatastic <bombayapirat@gmail.com>
* feat(dev-mode): Leptos code for a layer tree (#223)
inspect.code takes language: leptos, and the Inspect panel lists it: the
HTML markup inside a #[component] fn's view!, with the stylesheet after it
as for React. Text is Rust string literals (line breaks as <br/>) so the
macro parses it; px or rem like the other web languages.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(dev-mode): Leptos components take their properties, instances are calls (#223)
Leptos joins React in component code: a component or set is a #[component]
fn with #[prop] defaults (variant enums with as_str, bool, String, ViewFn
for instance swaps, Option<ViewFn> for slots), layers bound to booleans
show through .then, variants differ by data-* CSS rules, and instances
(also inside layer trees) are calls passing the values they change. Names
are snake_case, Rust keywords get a trailing _. Code Connect takes leptos
too (use <path>::<Name>; view! markup). Markup now takes the language
instead of jsx/leptos flags; React and Leptos share the stylesheet pass.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Justin John <babyjjbabyjj549@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bflatastic <bombayapirat@gmail.com>
View > Full screen (`view.fullscreen {on?}`, F11 on Windows and Linux,
Ctrl+Cmd+F on macOS) makes the window fill the screen, taskbar included,
and leaves it again. FigCraft's own menu item, translated in every catalog.
Tests: the key and the command toggle the window, junk parameters are
ignored; the shortcut table pins the new binding (49 pinned).
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
Co-authored-by: bflatastic <bombayapirat@gmail.com>
add_space(16) only indented the first row of a horizontal_wrapped, so with
many colours on a page the later rows started at the picker's edge and ran
past its right side. The swatch rows now sit in a frame inset 16 points on
both sides.
Co-authored-by: Justin John <babyjjbabyjj549@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Snapshots hold more than the document (recent batches, deletions), so near
the cap compaction could write one over max_file_bytes, which both readers
then refused: the file vanished on restart. Reading now allows twice the
cap, and compaction refuses a snapshot beyond that, keeping the old
snapshot and log.
Co-authored-by: Justin John <babyjjbabyjj549@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.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>
validate() now reports children listed twice, pages with a parent, and
nodes not reachable from a page (cycles, detached subtrees). repair()
drops repeated children, clears page parents and removes every node not
reachable from a page in one linear pass, replacing the orphan loop.
Co-authored-by: Justin John <babyjjbabyjj549@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Docs: roadmap entries for the community fixes landed in #211, #213, #214, #216 and #218
Their ROADMAP.md lines were left out when the five PRs were rebased (all five edited the
same header line); the wording is the contributors' own.
* Docs: the header lists all five community fixes
The AppImage ships figcraft-cli in usr/bin, but AppRun only ever launched the
app, so the CLI (and its MCP server) could not be reached without extracting
the image. `FigCraft.AppImage cli mcp` now execs `figcraft-cli mcp`, skipping
desktop integration because the CLI is headless.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Boolean operands restarted subtree_outline at depth 0, so booleans nested
in booleans bypassed the 64-level limit and could overflow the stack.
Co-authored-by: Justin John <babyjjbabyjj549@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Agent Host changes for agents/fix-command-batch-history-restore
* Implement centralized transactional fix for command batch failures - #1
* The batch rollback tests compile, are formatted and expect the tab that was in front
The new module used a private helper of plugins_tests and an unused import; the plugin test
asserted tab 0 was in front, but it opens a second file first, which is the one in front
when the plugin starts and the one the rollback puts back.
---------
Co-authored-by: bflatastic <bombayapirat@gmail.com>
crates/text/build.rs asked cargo to watch both craft-fonts manifests
whether or not they existed. Cargo takes a watched path that doesn't
exist as changed, so figcraft-text and every crate above it were rebuilt
on every build: with no craft-fonts checkout, and also with a checkout
of craft-fonts' main branch, which has no fonts/latin-manifest.txt.
Watch a manifest only when it exists, and its folder while it is
missing, so a manifest added later is still noticed. With no checkout
at all there is nothing to watch; the build warning and the README now
say how to pick one up after a first build.
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>
Every plugin import now counts as a host call (reply_len, reply_read and set_output were free), and bytes copied out of or into the plugin's memory spend fuel, reserved before the copy or allocation, including when call overhead is set to zero. Fixes#192.
The log byte budget test from #216 runs with the default fuel, since copying a megabyte of log bytes is itself charged now.
With a layer's proportions locked, setting only its width or only its height through node.set (the W and H fields) scales the other side by the layer's own ratio; setting both in one edit leaves them as typed, an unlinked layer is unchanged, and a zero side has no ratio to keep. Fixes#220.
Plugin log lines are admitted before guest memory is copied: a line over 8 KiB, more than 1,000 lines or more than 1 MiB of log bytes in a run fails the run with a limit error, and the same budgets bound what is forwarded to the host logger. Fixes#191.
file.revert is refused while the file is in a live session (hosting, joined, publishing or reconnecting): reverting replaced the document's identity and broke its session. The refusal changes nothing, and reverting works again after leaving. Fixes#198.
The attached MCP backend no longer re-sends a request when its reply is lost: the app may already have applied it, so the error says the operation may have completed and to inspect the document before retrying; the next call reconnects, and a failed reconnect says the request was not sent. Fixes#194.
Also: a connection the app already closed (it drops ones idle past its socket timeout) is noticed before the next request is written and replaced then, so a pause between calls is not an error.
Co-authored-by: bflatastic <bflatastic@users.noreply.github.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.
edit.repeat no longer records itself as the last command, so repeating keeps applying the command it repeats instead of recursing, and it refuses when there is nothing to repeat. Fixes#181.
The sample budget is enforced while a path is flattened, not after: the element count is checked up front, each curve is flattened on its own, and flattening stops as soon as the budget is exceeded. Fixes#204.
The sessions test waited for a new layer to arrive and then asserted on its name, which the host sets in a second batch; it now waits for the name. Test only.
A nested instance opened from a .fig or file JSON document is first linked to its
own component; matching it to the outer main's layer then went by that component's
name instead of the layer's, so nested instances of variants (or under a name of
their own) and their layers stayed unlinked, and a second export lost what they
changed. They go by their own name until they mirror a layer.
A variable width is written only as the same reach on both sides of the path (a
point with unequal sides drew nothing elsewhere). A stroke wider on one side, on a
shape that shows nothing else, is written as a vector filled with the area it
paints, with the stroke itself kept as plugin data that the import puts back.
Bound layer and text fields also name the number of the field they set. Docs: text,
effect and grid styles and symmetric variable widths are checked; non-colour
bindings come across as values.
Variable-width points say each side's reach as a share of the stroke weight (in
pixels they drew far too wide elsewhere). Progress docs: instance sizes, variables
with modes and shared styles in the .fig export, and what was checked by opening
written files in other software.
Instances write what they change from their main as an override addressed by the
main's id (size, name, paints), what follows from that for their layers as derived
data (size, place, outline, colours their variables give), and a scale factor when
they were scaled as a whole; layers of a nested instance swapped for another
component keep their overrides. Variable collections, modes and variables are nodes
on a hidden page; paints, gradient stops, effects, layout grids and layer fields
name the variables bound to them, and frames and pages the modes they set. Shared
styles are nodes there too, linked from the layers that use them. Variable-width
strokes say their reach on each side of the path. The import reads all of it back
(style definitions and unused styles, variable order and scopes, bindings on paint
lists, stops, effects and grids, page modes, text bindings, width profiles).
FIGCRAFT_DEFAULT_SERVER, set when compiling, names another server or, set to nothing, ships a
build with no default: online sharing then stays off until the user enters a server in
Preferences. Normal builds and the release workflow are unchanged.
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.
MarekSpodBiedry