`HatchModel.boundary` stores multi-region hatches as a single Vec
with NaN-NaN sentinels separating each disconnected sub-path
(documented on the field). The paper-canvas hatch fill ignored the
sentinels and forwarded every vertex straight to the iced canvas
path builder. When a layout contained a multi-region hatch, lyon
received a polygon containing NaN coordinates and panicked with
`Tessellate path.: Internal(IncorrectActiveEdgeOrder(4))` from
inside `iced_wgpu::geometry::Frame::fill`, taking the whole app down
the first time Layout 1 painted.
Walk the boundary instead, opening a fresh sub-path on each sentinel
and closing whichever was active. Each region is now filled as its
own closed polygon. Reproduces with #22.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Both the on-screen paper canvas and the PDF / print pipeline only knew how
to draw the boundary outline of pattern hatches (ANSI31, ANSI37, dotted
brick, etc.) — the actual hatch lines were a GPU-only feature, so any
paper-space view or PDF export showed a hollow polygon where there should
be ruled fill.
Add a CPU rasteriser as `HatchModel::pattern_segments()`:
- builds an edge list from the boundary, splitting at NaN sentinels and
closing each sub-path so even-odd inside tests handle islands / holes;
- for every PAT family computes the world-space step vector
(`(dx,dy)` rotated into the family's frame, scaled by `model.scale`)
and the inclusive k-range bounded by the boundary AABB plus a 4096-
line safety cap;
- intersects each line with every boundary edge, sorts the hit
parameters, walks them in pairs (even-odd) and emits one segment per
in/out interval. Dash patterns are walked along each clipped span with
phase preserved via `t mod period`, matching the GPU shader's
absolute-t dash evaluation;
- pattern origin (`family.x0 / y0`) is rotated by `angle_offset` and
scaled the same way as `hatch_batched.wgsl`, so PAT patterns with
non-zero origins (brick / offset herringbone) line up with the GPU
render.
`paper_canvas::draw_hatch` and `pdf_export::emit_hatch` now route
pattern hatches through this rasteriser; solid / gradient hatches keep
their existing fast paths. The previous outline-only behaviour was a
visible regression on every layout containing a ruled fill.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two precision problems were producing visibly wrong hatches on UTM-scale
(and similarly-offset) drawings even after world_offset was correct:
1. **Boundary tessellation** — `hatch_model_from_dxf`'s polyline-bulge
arc tessellation ran the whole arc-center computation in f32. With
vertices at WCS magnitude ~1e5 the f32 ULP is ~1 cm, which propagated
into the sampled arc points and showed up as visibly wavy / jagged
hatch boundaries on edges that crossed the long axis of the drawing.
Re-run the full bulge → arc math in f64; only cast at the to_xy
call that feeds the (also f64) boundary buffer.
2. **Pattern phase** — the per-fragment `perp / perp_step` modulo blew
precision when `xz` was at offset-rel WCS magnitude and perp_step
was sub-cm. Move the precision-critical bits out of f32 by keeping
each hatch's anchor in f64 on the model side: `HatchModel` now
carries a `world_origin: [f64; 2]` and the `boundary: [f32; 2]`
vertices are stored as small offsets from it. The GPU vertex shader
adds `origin` back inside the view_proj multiply for clip position
(sub-pixel-lossy at large hatches but invisible); the `xz` varying
that the fragment shader uses for in_polygon + pattern math stays
in hatch-local space at full f32 precision.
For pattern alignment across adjacent hatches with the same family,
the GPU pipeline snaps the anchor to the family's perp/along grid
in f64 before sending it down. Snap uses the same QCAD-PAT
convention (`fam.dy = perp_step`, `fam.dx = along_step`) the
shader's check_family uses — an earlier attempt used the world-frame
`(-dx*sin + dy*cos)` formula, which produced a meaningless grid and
no alignment.
Trade-off: pattern phase is "global modulo per-family-grid" rather
than tied to WCS (0,0). Same hatches still align with each other;
the absolute phase relative to the WCS origin may differ from the
pre-fix behaviour on small drawings.
Consumers that read `model.boundary` directly (hit_test, paper_canvas,
add_hatch round-trip) reconstruct WCS via `boundary[i] + world_origin`.
The hatch wgsl bind-group entry for binding 0 was bumped to
`VERTEX | FRAGMENT` because the vertex shader now reads `h.origin`.
The fragment-shader hatch LOD substitute + sub-pixel skip from e7675be
and the scissor plumbing from fbbac42 still work — those touched
different fields of the same struct.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
paper_canvas_wires() now returns Arc<Vec<WireModel>> backed by
paper_canvas_cache (epoch-keyed). Pan/zoom frames hit the cache and
return a pointer copy; only geometry changes rebuild the wire list.
Completes the paper-canvas caching chain (L+M): sheet cache →
projected-per-viewport cache → full canvas Arc cache.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Option I: add per-viewport Arc wire cache so paper-space viewport rendering
no longer re-tessellates model content every frame.
Option J: hit_test_wires() returns Arc<Vec<WireModel>> — O(1) pointer copy
in model space instead of a full Vec clone on every mouse move during commands.
paper_canvas_hatches/wipeouts() now route through the epoch caches added in F.
Option K: add world_snap_r + wire_in_range() pre-check to Snapper::snap().
All Endpoint/Midpoint/Nearest/Perpendicular/Intersection loops skip wires
whose chord sphere is outside the snap circle using O(1) scalar comparisons,
eliminating O(entities × vertices) matrix multiply cost when zoomed in.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
White paper fill was calculated from entity bounding boxes (paper_entity_extents),
so adding a viewport caused the paper to visually shrink to the viewport's bounds.
Now paper_limits() (layout min/max limits) is used directly, and the unused
paper_entity_extents() function is removed.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- paper_entity_extents() computes the bounding box of actual DXF entities
so the white fill always matches the drawn title-block / frame borders,
regardless of whether Layout min/max_limits match entity positions
- Falls back to paper_limits() (→ A4 297×210) when layout has no entities
- Removed paper_boundary_wire: the white fill now acts as the visual paper
edge so the near-white wire (invisible on white) was redundant
- LayoutCreate sets limits to A4 landscape immediately after add_layout()
to override acadrust's imperial default (12×9)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- paper_limits() falls back to A4 landscape (297×210mm) even when no
Layout object exists in the document
- paper_bg_color default changed to white [1,1,1,1] (paper surface color)
- PaperCanvas::draw() fills canvas with dark desk color first, then draws
the paper area rectangle in paper_bg_color so the sheet is clearly
visible against the desk background
- view.rs container background uses the fixed desk color instead of
paper_bg_color so it no longer conflicts with the paper-surface color
- BACKGROUND RESET for paper space now resets to white instead of dark gray
- Removed unused PaperSheet variant, paper_sheet() constructor,
build_paper_sheet_primitive() and associated match arms (dead code
after 2D canvas replacement)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Paper space is now rendered as a 2D vector canvas (iced::widget::canvas)
instead of a 3D shader widget. This allows direct interaction with
paper-space entities (title blocks, viewport borders, annotations)
without needing to enter MSPACE first.
Changes:
- src/scene/paper_canvas.rs: new PaperCanvas<'a> canvas::Program that
renders paper wires (with linetype dashes), solid/gradient hatch fills,
and wipeout backgrounds using the paper camera's orthographic transform.
- Scene: three new public helpers (paper_canvas_wires, paper_canvas_hatches,
paper_canvas_wipeouts) feed the canvas with the same data the old shader
used, including inactive viewport projections and interim/preview wires.
- view.rs: paper_canvas_view() now wraps PaperCanvas instead of
shader(ViewportPane::paper_sheet(…)). The PaperViewportPane 3D shader
overlay for the active MSPACE viewport is unchanged.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>