Gap 10: a file counts the library instances it places, creates, pastes or duplicates and reports them like its detaches. The server counts what a file's totals grow past their highest in the current week and keeps a year of weeks (actions.json). library.usage {days} and Show usage give insertions and detachments over the past 30 days, 90 days or year. New variants are appended; an older server refuses them and Show usage says so.
A request line can carry timeoutMs beside id and method, capped at 600,000 ms; without it the wait stays 60 s and older apps ignore the field. figcraft-cli mcp --attach PORT --timeout SECONDS sends it with every request and reads the answers that much longer. A timeout's message says how long the request waited.
* M11: a widget opens its plugin's panel (plugin API 5)
Gap 6: a widget callback's ui_show opens the plugin's panel and keeps the module loaded; the panel's callbacks get the widget's id and state and redraw it in the same undo step. Older plugins' ui_show from a widget still opens nothing. The Counter example sets its count from a panel.
* M11: open_panel takes its opener as one value
Gap 14: variable.extendCollection and Extend collection on a collection's row in the Variables editor. Extensions follow edits of the collection they extend at once (normalising keeps them in step whenever variables change), extensions extend again, and a loop of extensions from a hostile file comes apart into plain collections.
* M4: the eyedropper with Cmd/Ctrl+Shift makes a colour style or variable of what it picks
* M4: the Create from color window's collection picker is its own function
An odd-length pattern is drawn repeated ([4, 2, 1] as [4, 2, 1, 4, 2, 1]),
so editing its Dash as three values changed the second gap. The pattern is
written out repeated first and only the edited entry changes.
library.check refused nothing while automation roots confined it: it
counted libraries it never asked about and remembered them as yours. It
now refuses before remembering any. library.updates and library.list
answered a refused check with the team library cached earlier, as if the
server had nothing newer; the refusal now reaches them instead of
falling back like an offline server. The MCP and control protocol docs
say that a root on a network share still reaches that file server.
The UI starts a request through its gate (a new Running state) just before running it and never runs one the transport already cancelled. At the timeout the reply says it was dropped and not applied, or that it is still running and will take effect; a reply that just made it is delivered. Only screenshots and input blame the window.
* M12: an ended vote's results say who voted for each object
* M12: a board keeps earlier votes' results, listed in Results and deleted one at a time
* M12: the vote Results popover is drawn by small helpers, and its stepper, rows and Clear are tested
Fills every key the Finnish catalog was missing against the current
English catalog and command set (missing_translations_report: fi 0 / 0 / 0),
following the catalog's existing terms (sisämarginaali, täyttö,
reunaviiva, tehoste, ilmentymä, yhdistin, taulu) and its space before %.
No existing row is changed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The stroke settings show the first dash and gap only, and editing either
wrote back [dash, gap], so an imported pattern like [4, 2, 1, 2] collapsed
on the first edit. Each field now changes its own value and keeps the
rest of the pattern.
The roots confined files only: a confined agent could still publish the
open document to any server, host it on the network or join a session.
The engine reaches the network only through the live link, which now
refuses a call made while the roots confine it (hosting, joining,
publishing, sharing, team libraries, plugin directories), and skips the
background team library checks. The MCP and control protocol docs say so.
The timing assertion could fail under load. finite_check_candidates is now its own function, and the test checks that a rename among 200,000 layers looks at only a handful of them.
check_finite_geometry went over every layer of the document after every
command, so each edit cost time in proportion to the file, not the edit.
It now walks Document::changed_ids against the version the command
started from; the imbl diff skips what the two versions share.
100 renames among 200,000 layers: ~0.60 s before, ~0.02 s after.
* M12: stamps stick to the objects they land on
A stamp sticks to the topmost object under its centre (BoardItem::Stamp gains an optional on, left out when empty so older files open with stamps on nothing) and the board's normalize pass carries it when that object moves or turns, once when both move together. A stamp dragged onto another object sticks to that one, off everything to nothing; one whose object is deleted stays where it is. Gap 24.
* M12: the object a stamp will stick to is highlighted
While the stamp tool is over an object, or a stamp is dragged over one, the object it will stick to is outlined with the canvas's hover outline (Overlay::stamp_target, found by the same stamp_target as board.stamp and settle_stamps). A drag no longer drops what it moves into the auto layout of a sticky, shape with text or table: board objects take no drops; a stamp already inside its sticky moves with it once. Builds on #1671. Gap 24.
* M3: team library analytics count detached instances
Detaching an instance of a library component counts it in the file (libraries.detached, saved with it, capped). The background library check reports the counts after its fetch (ReportDetached); the server keeps them per file beside the file's usage report in usage.json (same caps) and answers Detached with them. Show usage and library.usage list detaches per component and per file (detached, detachesKnown). The new request and reply variants are appended, so every older library message keeps its bytes; an older server refuses the new requests, which the app ignores. Gap 10.
* M3: the detach test waits for the instance count too (the two arrive in separate reports)
Main merged #1652 (the Bosnian catalog) and added more strings; ROADMAP keeps main's rows with the count updated, and bs.toml translates the two new strings (a pasted table's size limit, connector labels without a snapshot), so it covers all 1,906 UI strings.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013Tayw8nmxoLLmKv13jK93z
proto.linkNotice from the Actions palette, the control channel or MCP changed the preference in memory only until the app next saved its preferences. It now sets Session::prefs_unsaved, which the app saves at the end of the frame like installed plugins and live-session tokens; the notice's own Don't show this again goes the same way. Gap 15.
The bar over a selected connector adds start and end point menus with every line end, a line style menu (solid or dashed, weights 1 to 6) and a colour wheel opening the colour picker, beside the board colours and line type; every control names itself on hover. UI only, through board.setStyle and the stroke's dashes and weight.
Four strings main added since the last merge (pasted design layers and
their missing images, image fills and shared styles not in the file) get
Bosnian translations, so the catalog covers all 1,905 UI strings again;
the counts in ROADMAP.md and the localization parity doc follow.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013Tayw8nmxoLLmKv13jK93z
downloads_have_a_share_per_address failed once under a full gate: its 40 library fetches could take long enough for the address's download share, measured against Instant::now() and refilling 1 MiB a second, to refill one more. Store::clock() is now what every rate limit the store checks reads: the clock, or in test builds a time the test holds it at (a #[cfg(test)] field). The hardening, download burst and admin budget tests hold it still; the admin test is back to its exact budget (#1564 had allowed for refilling). Found flaking in gates.
BFlat