Direct distance entry (#1429) turned every bare number at a point prompt
into a distance, so ROTATE's angle and the values automation types were
read as coordinates. The step reads the token first now — `feed_command`
reports whether it claimed it — and only an unclaimed one becomes a
distance. That also puts the token back on the unified step driver, which
owns the P / L selection keywords and the dimension-snap bookkeeping, and
stops `on_text_input` running twice for the same token. A tracking ray
carries its own base, so it no longer demands an anchor that the first
point of a command does not have yet.
Plugin command execution (#1425) re-implemented the command line inside
the plugin host: it dispatched families directly, skipping the teardown
that starts a command (grip edits, selection gestures, `last_point`, the
accepted snaps), and made every `CmdResult::Dispatch` in the application
skip plugin dispatch to avoid its own re-entrancy. A scripted line now
runs through `run_command_line`, with the trailing Enter left off while
the plugin streams the command, and the re-entrancy is handled by a flag
set for the length of the plugin's call. A pause parks the line's Enter
with the rest of its tokens instead of dropping it.
A font reference (#1432) tried the family lookup before the face-file
index, and that lookup ends in a prefix match a longer file name loses to
(`ISOCPEUI.TTF` would take `ISOCP`). The exact file answers first.
Co-Authored-By: Claude Opus 5 <[email protected]>
Reads and sets the form new dimensional constraints take (0 dynamic,
1 annotational), the same state DCFORM sets.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
Dynamic dimensions carry DIMEXO/DIMEXE/DIMASZ overrides in text-height
ratios, a two-line angular dimension draws extension lines only beyond
its sides, a selected dynamic angle shows triangle grips at the arc ends
with squares at the arc point and text, and the constraint dimension's
Properties follow the reference: kind titles, General before Constraint,
angular-precision Value and Measurement, drawn text position, reference
row order for Fit, Primary Units and Tolerances, and text/dimension line
alignment rows reading On as the aligned state.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
Eliminate major performance bottlenecks in the per-frame view render path:
- Memoize document render environment resolution using (geometry_epoch, len) keying, replacing per-viewport O(N) linear scans over document.objects.
- Index render_environments and render_settings handles in ObjectDataCache during the single build_cache pass.
- Cache resolved background image lookups in memory to eliminate repeated path.is_file() filesystem syscalls per frame.
- Add has_face3d fast-path to bypass wire splitting when no Face3D entities exist, and use FxHashSet<u64> handle indexing when present.
- Invalidate all cached structures across bump_geometry, bump_geometry_no_blocks, and bump_entities.
- Add comprehensive correctness unit tests and before/after benchmarks.
Performance impact (10k objects):
- P95 latency reduced from 341.00 µs to 173.00 µs (-49.3%)
- Throughput increased from 4,700 FPS to 6,000 FPS (+27.7%)
- Direct render environment evaluation: ~85 µs down to <50 ns (>1,700x)
Three-point angular dimensions take their sweep from the two rays and the
arc point (extension lines follow their own rays, straight and reflex
angles draw), dynamic angular dimensions store no text point so DIMTAD
places the text beside an unbroken arc, and dynamic dimensions carry a
DIMGAP override of a quarter text height.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
DCANGULAR picks two lines, an arc or three points, takes the dimension
line location to choose the sector and the value or name, and creates the
dynamic two-line or three-point angular dimension driven by an angN
parameter. Reference messages for invalid, repeated and parallel sides,
existing names and invalid names; angle values display within one turn at
the angular precision; the first side (or vertex and first point) stays
when the value changes; the ribbon Dimensional panel follows the reference
order.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
cosmic-text's fallback search can land on a face that claims the character and
outlines to nothing: macOS's system font does that for every glyph, because
ttf-parser reads no contours out of it. The code took the first non-.notdef
glyph id and stopped, so the character drew nothing while believing it drew
something — invisible text, and no warning either.
An empty outline is now treated as no answer: the search moves on to the next
face, and when cosmic-text's own list has nothing that draws the character, the
installed families are walked for one that does. That list is the platform's own
handful of names (on macOS `.SF NS`, `Menlo`, `Apple Color Emoji`, `Geneva`,
`Arial Unicode MS`) and covers neither Cyrillic nor Greek, however many fonts the
machine has that do.
`cargo test --lib a_fallback_glyph_is_never_empty`, and the Turkish letters in
`lff::tests::fonts_parse_and_resolve` render on macOS now instead of failing.
A DWG text style stores its TrueType font as a *file* name — `GOST2304_TypeA_italic.ttf`,
sometimes with a directory — which is what AutoCAD resolves. The family lives
inside the file and is usually a different string (`GOST2304 Type A`), so looking
the reference up as a family missed an installed font and fell through to the
embedded stroke fonts. Those carry no Cyrillic at all (`simplex`, `standard`,
`romans`, `iso3098` hold no U+04xx glyph), so a Russian drawing came out with
every word missing: the letters went on to the system-font fallback, which
returns nothing on macOS.
Installed faces are now indexed by file name, and a reference — a style's font
file, or an inline `\f` code, which carries either form — resolves by family
first and by file name second. An embedded stroke font still wins, so
`romans.shx` keeps rendering as it did.
`cargo test --lib face_files_resolve_to_their_family`
Infinite construction lines tessellated to a single unit stub at the base
point, which can never close a face (and overwrote the real wire segments),
so XLINEs were silently ignored by BOUNDARY. Clip unbounded curves to the
finite extents via Liang-Barsky; lines missing the geometry contribute
nothing. Same shared helper fixes RAYs.
Tests:
- xline_splits_rectangle_into_two_faces: 10x10 rect split in two
- ray_splits_rectangle_into_two_faces: upward ray splits in two
- xline_missing_the_geometry_contributes_no_face: distant xline is a no-op
Placements (entity lookups, intersections, String labels, projections)
recomputed per view() and per mouse-move hit-test, with overlay re-clones
and 2-3x offset recomputes. New constraints_epoch bumped on every set
mutation/visibility/undo/DWG-load (plus new set_constraint_enabled API);
single-slot Arc<[GlyphEntry]> memo on an 11-field key (hover/selection
excluded); overlay de-cloned/de-duped to the single geometry source;
hit paths reuse the cache.
Tests:
- cached_glyph_placements_hit_returns_same_arc: same-key Arc identity
- cached_glyph_placements_invalidated_by_metadata_edit: edit + selection change yield new Arc
- glyph_hit_test_entries_matches_precomputed_boxes: shared hit-test helper
Bench (cargo bench glyph, 60 constraints): 8.18us -> 0.05-0.15us (~55x).
A/B reference deleted.
The pinch is read from AppKit rather than through iced, which is the kind of
decision that reads as arbitrary without the reason next to it. The file also
records what the pixel/notch split costs: a smooth-scrolling mouse pans with a
trackpad.
A trackpad could only zoom: every scroll delta, wheel notch or not, went
into the same zoom step, and a pinch never arrived at all. iced's winit
layer converts only the window events it has an iced equivalent for, so
winit's `PinchGesture` was dropped before the app saw it.
Scroll deltas now split by what produced them: a wheel reports notches and
still zooms, a precise-scrolling device reports pixels and pans. The same
pixel split keeps working everywhere else — the browser build reports every
wheel notch as pixels, so there they keep zooming.
The pinch is read straight from AppKit with a local `NSEvent` monitor, since
the app cannot see the gesture through iced at all, and handed to the app
through a subscription shaped like the SpaceMouse bridge. It maps 1:1 onto
the camera: fingers 5% further apart leave the view 5% closer.
The pan and zoom bodies are now shared by the mouse and the gestures, so the
middle-button drag, the wheel, two fingers and a pinch all move the view
through one path — in model space, in a tiled pane and inside a floating
viewport alike.
#1418 added both as direct dependencies although each was already in the
tree: printpdf's to_lopdf_doc returns the lopdf document whose compress
and save_to are plain methods, and the codec's encoding_from_code_page
hands back the Big5 / GBK / Shift_JIS / EUC-KR encodings by their DXF
code-page names. Cargo.toml keeps its dependency list.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
#1415 made a dimensional Distance keep a negative value's sign in the
parameter while the geometry uses its magnitude (and a zero fold the
points together), but left the value-domain test asserting the old
refusal. Assert the new reading, and keep the radius refusal covered.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
Symmetric held the connecting direction perpendicular to the axis with
the kernel's angle-based Perpendicular, which locks its turn side at
construction: a point starting far from its mirror image had to swing
the pair through the axis direction, the residual wrapped, and the
solver stalled with the point unmoved. The perpendicularity is now the
projection of the pair onto the axis direction being zero, a polynomial
residual the solver follows from anywhere. Circle centers and ellipse
centers use the same equation. tests/parametric_constraints_solve.rs
symmetric_constraint_mirrors_one_circles_center_across_the_axis_line
passes again.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
- .github/workflows/ci.yml runs `cargo test --workspace --locked` on
pushes to main and on pull requests, with the Linux libraries the
release build installs. No formatter check: the tree is not
rustfmt-clean and stays that way.
- flake.nix, flake.lock and the direnv hook are removed. Nothing used
them, and their pinned dependency hashes were already stale after the
codec and kernel bumps.
- Under cfg(test) the "system" language resolves to en-US. The unit
tests compare English text through one process-wide loader, and every
apply_config of a default config asked for the system language, so a
Turkish locale failed 22 tests and parallel tests raced each other
through a language switch (the thickness-row test found its section
under one language and looked it up under another).
- The arc grip test follows #1352: the start, end and midpoint grips
drive the whole arc.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
printpdf 0.9's `PdfSaveOptions::optimize` is a no-op — its `doc.compress()`
call is commented out — so every page's content stream (megabytes of vector
operators for a CAD sheet) was written uncompressed: a 16-sheet drawing set
exported as 280 MB. Serialise through `printpdf::to_lopdf_doc` and run
lopdf's `compress()` before writing; the same export is now 49 MB and
readers see ordinary FlateDecode streams.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Two layout defects that together broke every numbered note in a Chinese
drawing:
- Words were split on whitespace only, so a Chinese sentence — which has no
spaces — was one unbreakable word. When it did not fit after "7.<tab>" the
whole sentence dropped to the next line and still ran past the column.
Ideographic text may break between any two characters except before a
closing mark or after an opening one (the basic kinsoku rule); the atom
splitter now emits those break opportunities.
- `\pxi…,l…,r…,t…;` paragraph indents and tab stops are multiples of the
text height, and the first-line indent is relative to the left indent
(a hanging list is `i-1.5,l1.5`). They were applied as raw drawing units,
i.e. a fraction of a unit, so wrapped lines snapped back to the margin
and the number/tab column collapsed. Scale them by the entity height and
resolve the first line from the left indent.
Adds a layout test: a 40-ideograph paragraph in a 30-unit column wraps into
several lines, none wider than the column.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Only the first line's ends are pinned while the second line is made
parallel, the way the Parallel constraint itself moves it; the distance
is then measured on the solved geometry.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
Point & line and 2Lines take a text baseline or an ellipse axis as the
line; a zero value folds the points onto the first one; 2Lines turns the
second line about its start. The Properties Constraint section edits
Constraint Form (Dynamic on the *ADSK_CONSTRAINTS layer, Annotational on
the current layer), Reference (driven constraint in parentheses whose
parameter follows the measurement) and Description (stored with the
parameter and the native variable). Dynamic dimensions keep a 12 px text
and 8 px arrows at any zoom, drop trailing zeros and show formulas as
`fx: len=d1*2`; the hidden constraints layer stays out of the layer lists.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
Glyph quads bake the atlas UV of their tile. `grow_height` rescales every
cached entry's V and bumps the atlas generation, and the tessellation memo
folds that generation in — but the caches that hold finished wires do not:
the resident wire set is keyed by geometry epoch, block definitions by
block epoch, and a full build can grow the atlas part-way through (a CJK
sheet set bakes thousands of tiles on first sight). Text tessellated before
the growth then samples the wrong tiles for as long as those caches live,
which on a big drawing is forever: whole tables of scrambled glyphs, and
the block contents stayed scrambled even after a REGEN.
- `resident_wires_for` rebuilds (dropping the block definition cache) until
the atlas generation is stable across the pass.
- `Scene::update` compares the live generation with the one the scene's
caches were built against and bumps geometry once when it moved, so a
later growth (opening a layout, editing text) re-lays-out everything.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
`shx.rs` only accepted the classic `AutoCAD-86 shapes 1.x` container, so
every modern Autodesk font (`txt.shx`, `simplex.shx`, `romans.shx`, … are
`unifont 1.0`) and every Asian big font (`chineset.shx`, `hztxt.shx`,
`bigfont.shx` — `bigfont 1.0`) was rejected and text fell back to the LFF
substitutes and TrueType fallback glyphs. Drawings authored with SHX big
fonts (the norm for Chinese / Japanese / Korean sheets) never showed their
real glyphs.
- Parse all three containers. Unifont glyphs are keyed by Unicode; big-font
glyphs by their double-byte code from a sparse hash table of
(code, length, offset) entries. Big fonts reference subshapes with a
two-byte `7, 0, hi, lo`.
- `bigfont_glyph` encodes a character with encoding_rs (Big5 / GBK /
Shift_JIS / EUC-KR, most likely first for the file name) and takes the
first code the font has.
- `Face::WithBig` pairs a style's big font with its primary face: full-width
characters come from the big font, everything else from the primary —
how a STYLE with a "Big Font" behaves in AutoCAD. `resolve_text_style`
resolves `big_font_file` like the primary (next to the drawing, then the
user fonts folder) and rides it along in the font name.
- The missing-fonts dialog lists unresolvable big fonts too.
Unit tests cover a synthetic big font (Big5 lookup, cap-height scaling, no
aliasing of unknown codes), a synthetic unifont, and the pairing.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
DCFORM sets the annotational or dynamic form and continues into the
DIMCONSTRAINT option prompt, whose wording and last-used default match the
reference. -PARAMETERS manages parameters from the command line; a
double-clicked constraint dimension opens its value prompt. Dynamic
dimensions rescale on every camera change and their display overrides stay
out of the saved file. A driving dimension between an entity's own ends no
longer has its length retained, so a new value moves the second point along
the dimension axis; parameter edits hold the first constraint point.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
DCALIGNED opens with `[Object/Point & line/2Lines] <Object>`. Point & line
asks `Specify constraint point or [Line] <Line>:` then `Select line:` (or the
line first and `Select constraint point:`), measures perpendicular to the
line from the constraint point to the line end nearest to it, and stores a
directed distance with the line as direction reference. 2Lines asks
`Select first line:` and `Select second line to make parallel:`, adds a
Parallel constraint solved with the first line pinned, then measures on the
solved geometry between the nearest ends. Unsupported line picks report the
reference's message; the same measurement again reports `The constraint
already exists on the selected objects.` and ends the command; an invalid
expression reports both reference lines and asks again.
The offered default and seeded expression carry twelve significant digits;
a typed value moves the second point (or the other object) while the first
point and its measured line stay, negative values keep their sign in the
parameter and use their magnitude, zero folds the points together, and the
dynamic dimension's origins follow every solve. The lock mark sits after the
text, a selected dynamic dimension shows triangle grips at its points and the
text square, and its Properties show the Constraint section (Name and
Expression editable) plus Text rotation.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
A plain number shows as typed (d1=100), an expression under the fx: prefix
(fx: d2=d1*2); the text stays horizontal through DIMTIH/DIMTOH overrides
and the lock mark follows the text.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
Every dynamic dimension gets a DIMSCALE override for 12 px text, refreshed
per zoom band, at creation and after load. A SCALE/ROTATE of an entity whose
own end points a driving dimension measures now pins its second constraint
point so the dimension moves the first end, as the reference does.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
When a text style's SHX / stroke font lacks a character, the glyph comes
from a system TrueType font through `ttf_glyph::fallback_glyph`. That
substitute was normalised to the Latin cap height (9 units = text height),
which is right for Latin, Greek or Cyrillic stand-ins but wrong for CJK:
an ideograph fills the whole em square, and 1 em is 1.3–1.5 cap heights in
every common CJK font (DFKai-SB has no OS/2 cap height at all, so it fell
to the 0.7 em guess). Every substituted Chinese character therefore came
out ~40 % taller and wider than the big font (`chineset.shx`, `hztxt.shx`,
…) the drawing was authored with, and whole lines of notes ran past their
frames.
A big font draws an ideograph one text height tall and one text height
wide, so size full-width / ideographic fallback glyphs by their em box
instead. Other scripts keep the cap-height normalisation. Real TrueType
styles (`Face::Ttf`, shaped runs) are unchanged.
Adds a unit test that a fallback `中` advances by about one text height and
its ink fits the text height (skipped on machines without a CJK font).
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
`viewport_content_wires` projects model-space wires into paper coordinates
for plotting: the points and the linetype stations are multiplied by the
viewport scale, but `world_width` (the band width of a wide polyline) was
copied unchanged. The PDF / print exporter strokes that band at
`world_width` in paper millimetres, so a 15-unit-wide bus bar seen through
a 1:100 viewport came out as a ~21 mm black blob instead of 0.15 mm, and
every single-line diagram sheet plotted as solid black shapes.
Scale `world_width` by the viewport scale alongside the points, and keep
the per-point taper widths in step (or drop them when clipping changed the
vertex count, so they cannot misalign). Negative (fixed pixel) and zero
(lineweight) widths are left untouched.
Adds a regression test that plots a 15-unit-wide polyline through a 1:10
viewport and expects a 1.5-unit band in the projected plot wires.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
The dynamic dimension of a dimensional constraint is now linked in the
drawing file (AcDbAssocDependency + AcDbAssocDimDependencyBody on the
constraint node), reread on load, hidden by DCHIDE/DYNCONSTRAINTDISPLAY
and never plotted; it carries a lock mark before its text. New
CONSTRAINTNAMEFORMAT and DYNCONSTRAINTDISPLAY variables and the
DIMCONSTRAINT option front end. A transformed entity whose own end points
a driving dimension measures keeps only the first constraint point, so a
scaled line returns to its dimension value.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
DCLINEAR, DCHORIZONTAL, DCVERTICAL and DCALIGNED follow the reference flow:
`Specify first constraint point or [Object] <Object>:`, `Specify second
constraint point:`, `Specify dimension line location:`, then `Dimension
text = …` and `Enter value or name and value <d1=100>:`. Linear reads
horizontal or vertical from the dimension line location the way a linear
dimension does; Object takes a line's, an arc's or a polyline segment's
ends. A miss reports `No valid constraint point found.` / `No object
found.` and asks again; another object type reports `Invalid selection for
Linear. Select a line, polyline segment or arc.`.
Each constraint creates a drawing parameter (`d1`, `d2`, …, or the typed
`name=expression`), drives the constraint through it, and draws a dynamic
dimension in constraint gray whose text follows CONSTRAINTNAMEFORMAT
(`d1=100.0000`); the dimension is associated with the constrained objects
and its text is refreshed after every solve. A typed value other than the
measured one moves the second point while the first stays, and so does a
later parameter edit. Deleting the constraint removes its dimension and,
when nothing else reads it, its parameter.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
A fresh reference session shows the "=" bar of every object made equal
through Multiple right away; the earlier hidden state was not the rule.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
A second object of the wrong kind (a circle after a line, a line after a
circle) or a pair that already carries Equal now reports and asks for the
second object again instead of ending the command. The mismatch message
names what the first object asks for — `Select a line segment, polyline
segment, text, MText, major or minor axis of ellipse or elliptical arc.`
after a length object, `Select a circle or arc.` after a radius object —
and the same-pick message carries the reference's double space.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
- An object that carries several Equal relations shows a single "="
badge, as the reference does, instead of one stacked badge per relation.
- A pick made in the Multiple flow leaves its new constraint bar unshown
until the object is selected or bars are shown, matching the reference;
the single flow still shows the bar at once.
- A typed coordinate at the Fix `Select object:` prompt and at the
Horizontal/Vertical `Select an object or [2Points]` prompt picks the
object under that point, like a click, instead of being read as a point.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
Karim Jerbi