* Render tests: drop the always-true assert in degenerate_input
`assert!(... || true)` trips clippy's overly_complex_bool_expr on current
stable (1.96), which fails `cargo xtask ci` at the clippy step. The check
only proves that NaN input does not panic, so call the function and
discard the result instead.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Menu commands open their dialogs; JSON calls never open pickers (#65, #101)
Tools > Drafting Settings... (DSETTINGS, DS, SE) ran the JSON form with no
parameters, which changed nothing, so the menu item looked dead; only the
status-bar gear opened the dialog. Typed or menu-invoked DSETTINGS now
opens the Drafting Settings dialog, and PROPERTIES (PR, CH) shows the
Properties palette.
The engine's OPEN, SAVEAS and QSAVE opened a native file picker whenever
their parameters had no path, so `engine.execute` and the bridged MCP
`execute` tool could block on a desktop dialog. They now ask for a file
only when typed or chosen from a menu; JSON calls get the command's own
missing-path error. The ui.open / ui.saveas commands (Cmd+O,
Cmd+Shift+S) still open the pickers.
Typing the `cmd {json}` form at the command line (e.g. `layer {}`) no
longer opens the command's dialog or picker; the engine runs it with
those parameters.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
File > Close, a file tab's x and File > Close All ran the engine's
close/closeall directly, which drop the drawing. Nothing checked for
unsaved changes on any platform; the web report just made it visible.
- New ui-egui `closing` module: close requests from the UI go through a
queue of drawings (by DocState uid). A dirty drawing is shown and asked
about in an in-app egui modal ("Save changes to X?" Save / Don't Save /
Cancel, Escape = Cancel), the same on desktop and web.
- Save routes to the existing qsave path (picker for untitled drawings)
and closes only if the drawing is really clean afterwards; otherwise it
stays open with its edits and a status message.
- Close All asks per dirty drawing in tab order; Cancel stops it.
- Menu, typed CLOSE/CLOSEALL (null params) and the tab x use the prompt;
JSON calls (engine.execute, control, MCP, scripts) keep discarding
without asking.
- Shortcuts are paused while the prompt is open; ui.inspect reports
`closePrompt`.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Linux: present without a vsync queue so the crosshair follows the mouse (#25)
The canvas hides the OS pointer and draws its own crosshair, so every frame
the swapchain holds back shows as cursor lag. Under vsync (FIFO) wgpu asks for
one queued frame, but Vulkan clamps it to the driver's minimum swapchain size:
three images on X11 (Linux Mint's Cinnamon session), i.e. two frames queued
behind the one on screen. Windows DX12 honours one. The canvas also requested
a repaint on every frame while hovered, which kept that queue full.
- gpu::surface_config: AutoNoVsync (Immediate, then Mailbox, then FIFO; valid
on every surface) on Linux/BSD, frame latency 1. CADCRAFT_VSYNC=1 restores
vsync, CADCRAFT_VSYNC=0 selects low latency elsewhere.
- gpu::FrameCap (240 fps) when presenting without vsync: nothing else paces
frames on X11, and a 1000 Hz input stream rendered ~900 frames per second.
- No continuous repaint while hovering the canvas; input repaints by itself.
The status-bar message schedules its own repaint to clear on time.
- Hover/selection highlight looks up primitives in a per-list sorted index
instead of scanning the whole display list every frame (~2.5 ms per frame
on a 270k-entity drawing).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move frame-cap and status-repaint hunks away from other PRs' to avoid merge conflicts
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Clicks on the command line's history, the UCS icon and the ViewCube ring fell
through to the drawing (picking objects or starting a selection window, or
entering a point during a command), the chevron did nothing, and the wheel over
the overlays zoomed the drawing. The ViewCube's Top face was a click-only
widget, so the canvas still counted as hovered underneath it (crosshair, wheel).
- Every overlay now registers a widget that senses clicks and drags, so egui
hands the pointer to the overlay instead of the canvas.
- Command line: a click on the history lines or the chevron (now pointing up)
opens a scrollable history window with the whole log; F2 and TEXTSCR open it
too (`ui.toggle.history`, Window > Command History). Dragging the history's
top edge sets how many lines show (`ui.cmdline.lines`, 0-12).
- ViewCube (new viewcube.rs): Top face and a Home button run `zoom.extents`;
hover highlights; right-click menu (Home, Hide ViewCube). The compass ring
says that the 2D plan view can't rotate (no view twist yet) instead of
doing nothing.
- UCS icon and the WCS pill open the coordinate-system menu (World, Hide UCS
Icon).
- Control channel: `ui.drag` (real press, move, release) and
`cmdline.state` -> `historyExpanded`.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Light and System interface themes (System follows the OS appearance live)
CADCraft had one hard-coded dark interface: Tokens::get() always returned
Tokens::DARK and Visuals::dark() was installed once at startup.
- theme.rs: Tokens::LIGHT (light grey chrome, dark text) beside the unchanged
Tokens::DARK; Tokens::get() returns the palette of the theme in use (a
UI-thread thread-local set once per frame). Both egui styles are built from
the tokens and egui is switched to the shown theme, so standard widgets,
menus and popups follow too. The drawing area (canvas, grid, axes,
selection, grips, ViewCube, UCS icon) keeps its model-space colours in both
themes: drawing colours are not part of the interface theme.
- ThemePref { System, Light, Dark } in UiState (`theme`), resolved every
frame against ctx.system_theme(), so System follows the OS (winit, or the
browser's prefers-color-scheme) live without touching the saved choice.
No reported appearance: System shows Dark (SYSTEM_FALLBACK). Default stays
Dark (today's look).
- View > Interface Theme > Use System Setting / Light / Dark (ui.theme.system,
ui.theme.light, ui.theme.dark) and ui.theme {theme} for scripts and the
control channel; ui.inspect reports the theme shown.
- Hard-coded colours on the command line (history, border, keywords,
AutoComplete list), the status-bar message and the layer status icon go
through tokens; overlays drawn on the canvas use a new canvas_ink token.
- Desktop app: the choice persists via eframe storage (`persistence`
feature, key cadcraft.prefs); window geometry and egui memory are not
persisted, as before.
Fixes#59.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Document the theme field away from other PRs' edits to avoid merge conflicts
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The MSI gets a setup wizard (WixUI dialogs, no licence or folder page) with a
Shortcuts page: the Start Menu shortcut as before, plus an opt-in "Create a
desktop shortcut" checkbox bound to DESKTOPSHORTCUT. Silent installs take
DESKTOPSHORTCUT=1. The shortcut goes on the all-users desktop (per-machine
install) in its own component with an HKCU key path (ICE38/43/57), so
uninstall removes it; upgrades read that value back and keep the previous
choice unless DESKTOPSHORTCUT is given.
The build now needs WixToolset.UI.wixext 5.0.2 (release workflow and
package.ps1 updated).
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The scripted TEXT ended with a single newline, so TEXT was still asking for
the next line when the test ran undo. Undo then reverted to before the centre
line (taking the uncommitted text with it), and the only post-undo check,
"no Text", passed anyway.
The script now ends TEXT with a blank line (Enter), as an AutoCAD script
would, and the test asserts no command is running afterwards. After undo it
checks that exactly the text was undone and that the four circles, the
outline and the Center-layer line are still there.
Fixes#51.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
`convert` and `run --export` always fitted the model-space extents, so
`zoom`, `pan` or `view.set` in a run had no effect on the image and there
was no way to export a region.
- io: `write_framed(d, name, window)` frames PNG, SVG and PDF on a
model-space rectangle (fitted without margin, centred, no distortion);
`write` is unchanged and still fits the extents. `check_window` rejects
non-finite, inverted, zero-size and out-of-range (±1e12) windows.
- CLI: `run --export` shows the current view when the commands changed
it, otherwise the extents as before. `--window x0,y0,x1,y1` and
`--view extents|current` on `convert` and `run` choose the framing
explicitly. The usage text and README say how images are framed.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* cadcraft-cli mcp: --connect without HOST:PORT is an error, not a headless session (#103)
`mcp --connect` with no address (or an empty or flag-like one) fell through
to the headless backend, so a client meant for the app got successful edits
in a separate drawing. The mcp arguments now go through parse_mcp_args:
--connect must be followed by a non-empty value that doesn't start with `-`,
unknown arguments and a repeated --connect are rejected, and every error
exits non-zero before any MCP request is served. Plain `mcp` is still
headless; a real address still connects, and an unreachable one still fails.
- Unit tests for the parser; integration tests (tests/mcp_connect.rs) run the
binary: missing/empty/flag-like address, headless control, loopback
endpoint routing, closed port.
- docs/mcp.md says what happens without an address.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move tests away from other PRs' hunks to avoid merge conflicts
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Menu items open their missing dialogs; COUNT shows its result; New Layout asks for a name (#126)
Eight menu items ran their JSON form with no parameters and changed nothing.
Typed or chosen from a menu they now do what the menu promises; JSON calls
still run the command and never open a dialog.
- New crates/ui-egui/src/managers.rs: Linetype Manager (load from the
library, set current, global scale, pattern preview), Point Style
(PDMODE tiles drawn in code, size relative/absolute), Table Style and
Multileader Style (list, set current, new, edit properties), Page Setup
Manager (per layout: paper, orientation, plot area, fit/scale, center,
lineweights) and Constraint Settings (infer, bars, transparency, dynamic
constraints, AutoConstrain types and tolerances). All changes go through
their commands (app.run), so they are undoable and scriptable.
- menus::run_ui_command routes LINETYPE/LT, DDPTYPE, TABLESTYLE/TS,
MLEADERSTYLE/MLS, PAGESETUP and CONSTRAINTSETTINGS/CSETTINGS without
parameters to the dialogs; dialogs::show dispatches them.
- COUNT returns a `message` (per-block counts and total) that the command
line shows; Session::start echoes multi-line messages line by line.
- LAYOUT.NEW from the menu or command line prompts
"Enter name of new layout <LayoutN>:"; Enter takes the default, a taken
or invalid name is reported and asked again.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move COUNT test away from other PRs' hunks to avoid merge conflicts
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Properties palette: every control acts or is visibly unavailable (#6)
Several Properties palette controls looked clickable but did nothing: most
current-property rows with nothing selected (linetype, lineweight, thickness,
text/dimension/multileader/table style), the selection-type drop-down, a
circle's circumference and area, a line's length and angle, a dimension's
style, and the All/My switch.
- Nothing selected: Linetype runs `linetype {current}`, Lineweight `lweight`,
Thickness `setvar THICKNESS`, the style rows `style.current`,
`dimstyle.current`, `mleaderstyle` and `tablestyle`. Transparency,
Annotation scale and Plot style(s) have no command yet: greyed with a
"Not available yet" tooltip.
- Header: the type drop-down narrows the selection to one object type
(`select {handles}`), plus Toggle PICKADD (`setvar PICKADD`), Select Objects
(`select`) and Quick Select buttons, with icons drawn in code.
- Geometry: circle circumference/area, line length/angle and dimension style
are editable through `properties.set` (new keys `circumference`, `area`,
`dimStyle`); computed values (line deltas, polyline length/area, block
name) are greyed read-only.
- PICKADD now changes clicking with no command running: off, each pick
replaces the selection and Shift adds.
- Palette text fields keep typed text while the window itself is unfocused
(they read focus from egui memory, not `Response::has_focus`).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move palette tests away from other PRs' hunks to avoid merge conflicts
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Edit > Cut / Copy / Copy with Base Point, View > Zoom > Object, Dimension >
Update, Tools > Draw Order > Bring to Front / Send to Back and the Format >
Layer Tools LAYCUR, LAYFRZ, LAYOFF, LAYLCK and LAYULK did nothing when started
from a menu, toolbar or the command line with nothing selected.
- machines::SelectRun: "Select objects:", then run the command's JSON body on
the picked handles. With a pickfirst selection it runs at once, as before.
It replaces the DIMREASSOCIATE/DIMDISASSOCIATE SelectThen in annotate.rs and
calls the command body directly, so the command records one undo step.
- LAYFRZ/LAYOFF/LAYLCK/LAYULK pick one object ("Select an object on the layer
to be frozen" etc.).
- COPYBASE asks for the base point first, then for objects (skipped with a
pickfirst selection).
- Objects picked at the prompt are deselected afterwards.
- JSON/programmatic calls are unchanged and never prompt.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
TAN and PER snaps need the other end of the line, so the first point of a
LINE (no base point yet) could never snap tangent: a belt line tangent to
two circles was impossible to draw.
Now, as in AutoCAD, a point prompt can accept *deferred* snaps:
- snap: with no base and the new `deferred` flag, osnap offers "Deferred
Tangent" (circles, arcs) and "Deferred Perpendicular" (circles, arcs,
lines, xlines, rays) where no other snap wins. The hit carries the curve
(`SnapHit::deferred`, `Deferred`, `SnapCurve`).
- `resolve_deferred` solves the line once the other end is known: tangent
or perpendicular from a point, tangent-to-tangent (common tangents of two
circles, new `Circle::common_tangents` in geom) and the perpendicular
mixes. Among several solutions the one nearest the picks wins. No
solution (point inside the circle, concentric or nested circles,
non-finite input) is `None`, never a panic.
- prompt: `Prompt::deferred()` opts a prompt in; `Input::Deferred` reaches
only such prompts, everywhere else the session turns it into the point it
was picked at, so other commands snap as before.
- LINE opts in for its first point (and the next one while the first is
still deferred). No solution reports and re-prompts; Undo drops the
deferred point. The rubber band shows the resolved line, also the
tangent-to-tangent one while hovering the second curve
(`Session::cursor_deferred`, set by the canvas).
- canvas: passes the prompt's flag to osnap and sends a deferred pick as
`Input::Deferred`.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Point picks found the outer boundary by tracing the planar graph of all
curves, but islands only came from single closed objects (circles, closed
polylines, full ellipses, closed splines). A square drawn as four LINEs
inside a rectangle was ignored: HATCH drew through it, GRADIENT filled it
and BOUNDARY created only the outer polyline.
- geom::region::boundary_at returns the outer loop plus islands: every
connected group of curves inside the face that doesn't touch its
boundary and doesn't surround the pick point. Each island is the
outline of the group's outside, traced with the same sharpest-turn
walk from its leftmost vertex after dangling ends are pruned. Islands
report which input curves they came from. Traces are bounded by the
group's edge count and islands are capped at MAX_ISLANDS.
- loops_at (shared by HATCH, GRADIENT and BOUNDARY) adds those islands
when every curve in the group is an open LINE, ARC, elliptical arc,
open polyline or open spline. Groups containing a closed object keep
the existing per-object islands, so closed-polyline and circle islands
are unchanged (bulges kept).
- Tests: geom region tests for separate-line islands, dangling ends,
open and attached curves, nesting and hostile input; engine tests for
HATCH (rendered pattern stays out of the hole), GRADIENT (375 filled,
hole clear) and BOUNDARY (two polylines) with four-line and
closed-polyline islands.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* DWG open: stop runaway conversions early, name the limit (#57)
"file too large" was the DXF reader's 50,000,000 group-code cap (MAX_TAGS),
not a file-size limit. For a DWG it was only checked at the very end: after
acadrust had read the whole file, written all of its DXF text into one Vec,
the text had been copied again for decoding, and 50 million tags had been
parsed. A real 10 MB DWG is nowhere near the cap (a synthetic 10 MB AC1024
file is 3.7 million group codes and opens in about 2 s). A damaged file, or
objects the DWG reader misreads, can be: acadrust caps every count at 100,000
and reads zeros past the end of the data, so one desynchronised object becomes
a polyline with 100,000 vertices. A fuzzed 0.25 MB AC1024 file made it write
11 GB of DXF.
- cadcraft-dwg: Limits { max_dwg_bytes: 512 MB, max_dxf_bytes: 1 GB }. The DWG
size is checked before the reader runs; the DXF is streamed into a writer
that fails as soon as it passes the limit, so the conversion stops there
instead of filling memory. Both errors give actual and allowed sizes. The
input is no longer copied, and the DXF writer also runs under catch_unwind.
- cadcraft-dxf: TooLarge { tags, limit } says "drawing too large: N DXF group
codes, more than the 50000000 CADCraft reads". For ASCII the count comes
from the line count before any decoding or tag allocation. Valid UTF-8 text
is borrowed instead of copied. MAX_TAGS is public and unchanged.
- cadcraft-io: errors from the DXF converted from a DWG say so.
On the fuzzed file: before, 76 s and 13.4 GB, then the process aborted copying
the 11 GB text (under a 16 GB memory limit). After, 15 s and 5.3 GB, with
"DWG: converting this 0.2 MB file produced more than 1024 MB of drawing data
(the limit); ...". What's left is acadrust's reader itself (8 s, 5.3 GB on that
file), which this bridge can't bound without patching the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move tests away from other PRs' hunks to avoid merge conflicts
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Programmatic undo/redo ends a running command first (#177)
`undo` or `redo` run through engine.execute, MCP, the control channel or a
script while an interactive command was still running (e.g. TEXT waiting for
another line) swapped the drawing under that command. Interactive commands
push their undo step only when they end, so `undo` popped the previous
command's step instead: it removed the earlier geometry, dropped the running
command's entities, and left the command running from a stale state.
Both commands now first end the running command the way Esc does, exactly as
Session::start already does for Cmd+Z, the menu and the command line. The
command's work so far becomes one undo step, and `undo` then removes that step.
Regression test: programmatic_undo_during_a_command_ends_it_first.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move test away from other PRs' hunks to avoid merge conflicts
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* DXF: write non-ASCII text as \U+ escapes so AC1015 files stay valid (#164)
The writer declares $ACADVER AC1015 (R2000) and $DWGCODEPAGE ANSI_1252, where
text is in the code page, but wrote strings as raw UTF-8. Every other reader
decoded those bytes as CP1252 and turned non-ASCII text into mojibake; only
CADCraft (which tries UTF-8 first) read its own files back correctly.
- cadcraft_dxf::write_ascii now writes pure ASCII: every non-ASCII character
in a string group becomes the pre-R2007 \U+XXXX escape (UTF-16 code units,
so characters above U+FFFF are a surrogate pair of escapes). This covers
TEXT, MTEXT, ATTRIB/ATTDEF, table, block and layer names, xdata and
XRECORD strings alike. The header and the R2000 structure are unchanged,
so the DWG bridge (acadrust, which decodes the escapes) behaves as before.
- The reader combines surrogate-pair escapes, requires four hex digits, keeps
malformed escapes and unpaired surrogates as literal text, and now also
unescapes strings in binary DXF.
- Tests: R2000 save/reopen of CJK, Latin-1 and astral text in TEXT, MTEXT, a
layer name, a block name and ATTRIB/ATTDEF; pure-ASCII output with the
AC1015/ANSI_1252 header; UTF-8 (AC1032) and code-page (AC1015) input still
read; DWG round trip; hostile escapes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move code and tests away from other PRs' hunks to avoid merge conflicts
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
properties.set documented `transparency` but never read it, nothing reported it,
and DXF dropped group 440. Fixes#132.
- properties.set takes "ByLayer", "ByBlock" or a percentage 0..90 (a number or
"50"/"50%"); anything else (out of range, NaN, junk) is an error and leaves
the selection unchanged.
- properties and drawing.inspect / drawing.entities report it in the same form.
- Properties palette: the Transparency row is a menu (ByLayer, ByBlock, a
0..90 percentage) that acts through properties.set.
- DXF: entities write 440 unless ByLayer (ByBlock 0x01000000, a percentage its
alpha | 0x02000000, the alpha mapping layers use); the reader parses it back,
any other value reads as ByLayer.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Render tests: drop the always-true assert in degenerate_input
`assert!(... || true)` trips clippy's overly_complex_bool_expr on current
stable (1.96), which fails `cargo xtask ci` at the clippy step. The check
only proves that NaN input does not panic, so call the function and
discard the result instead.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* DXF: keep layer transparency and description through save and reopen (#96, #97)
The LAYER writer omitted both properties and the reader left them at
their defaults, so a drawing reopened with every layer opaque and every
description empty. Both are xdata on the LAYER record, as AutoCAD writes
them:
- transparency: AcCmTransparency, 1071 = 0x02000000 | alpha, with
alpha = (100 - percent) * 255 / 100 (40 % is alpha 153); values that
aren't "by alpha" read as opaque
- description: AcAecLayerStandard, two 1000 strings (the layer
standard's name, unused, then the description), cut to the 255 bytes
a 1000 group may hold
Both applications are registered in the APPID table.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
`assert!(... || true)` trips clippy's overly_complex_bool_expr on current
stable (1.96), which fails `cargo xtask ci` at the clippy step. The check
only proves that NaN input does not panic, so call the function and
discard the result instead.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The DMG opened as a plain Finder window. It now shows the app icon's figure on
the CADCraft colour, with CADCraft.app and the Applications link side by side.
Same approach as PhotoCraft's and VectorCraft's (storytold/photocraft#1201,
storytold/vectorcraft#493).
- packaging/macos/dmg/: background.svg (source; text outlined, no paper grain,
the figure an <image> of assets/app-icon), background.tiff (1x + 2x HiDPI,
16-colour palette, Deflate, sRGB; ~21 KB)
and dmg-layout.DS_Store (Finder view settings: window, icon size 128,
positions, background).
- generate.py writes the TIFF and the layout on any OS, byte-for-byte
reproducible, with nothing from the machine that ran it (no paths, users,
disk names or volume UUIDs).
- package.sh copies them into the staging folder; the image is still built with
hdiutil makehybrid. Nothing extra runs in the signing job.
- The volume name drops the version: the layout finds the background through
an alias that includes the volume name. The DMG file name keeps the version.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Release: Flatpak x86_64/aarch64, AppImage zsync updates, README Downloads
- AppImages embed gh-releases-zsync update information and ship a .zsync beside them.
- New flatpak job repackages each Linux tarball into cadcraft-<ver>-linux-<arch>.flatpak
(bundle manifest + flatpak-bundle.sh); the from-source Flathub manifest moves to runtime
26.08 with the same commented, minimal finish-args (packaging-lint checks they agree).
- Jobs that sign nothing (version, linux, flatpak, freebsd, web) leave the release
environment, so a branch dispatch is a dry run; the draft-release job runs only on release.
- README: Downloads section listing every build.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Release: check AppImage update info without extract-and-run; fix the deb/rpm description
APPIMAGE_EXTRACT_AND_RUN makes the runtime pass --appimage-updateinformation to the app.
The package description was PhotoCraft's.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Brandon Thomas <shared2@storyteller.ai>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Custom macOS larger runners in a runner group do not pick up jobs; the
built-in M2 Pro label does. Only release.yml uses it, and the
actions_macos_xl budget caps the monthly spend.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The standard macOS runners are shared with all CI and queue for an hour or
more. The macos-release runner (GitHub-hosted M2 Pro) is in a runner group
usable only by release.yml on the release branch, with a monthly spend cap
on the actions_macos_xl SKU.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* About: contributors compiled in from contributors.json (grab bag + table, sortable, name toggle)
Replaces the TSV files and `cargo xtask contributors` with the craftrules standard
(standards/contributors.md): contributors/contributors.json is generated by
craftrules/scripts/contributors.py and baked into the binary by build.rs as static tables.
Each contributor line: @username, opt-in display/real name, PRs, commits, lines +/−/Δ,
binary assets +/−, first/last commit. Sortable by every column; alphabetical ignores
case and the leading @.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Contributors docs: craftrules is two levels up (../../craftrules)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Contributors: self-submitted names in contributors/people.toml, read by build.rs
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Contributors: regenerate (people_entry verified)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* AGENTS.md: contributor credits section (agents offer to credit their own human only); docs
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
- version job guarded to refs/heads/release, so a dispatch from any other ref builds nothing
- FreeBSD x86_64 tarball built in a FreeBSD VM as part of the release run
- remove the FreeBSD build that ran on pushes to main
- fix two empty env keys that made the workflow invalid
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
