Review follow-ups to #390, which fixed PDF/print text for model space but
left the layout-plot path — the one issue #385 was filed from — broken.
viewport_content_wires rewrote points/snap_pts/key_vertices/aabb and then
cloned text_verts through untouched, so paper-space viewport text stayed at
model (UTM) coordinates: a dimension at model (25000, 12000) plotted its
lines onto the sheet and its glyphs kilometres off the page. A text-only
wire fared worse — TEXT/MTEXT carry no stroke points since text went
SDF-only, so it was dropped outright by the empty-clip and AABB rejections.
Project the glyphs through the same proj_abs the snap points use (via the
existing map_text_verts helper), cull per glyph on the quad centroid, count
them towards the paper AABB, and dim them like the wire colour.
Also in emit_text, which diverged from the wire pass it mirrors:
- divide the pen by `scale`, or a Fit/windowed plot prints text as
near-invisible hairlines while its own lines stay correct;
- apply the CTB plot style, or a monochrome.ctb plot renders the lines
black and leaves the text on its screen colour;
- reset the dash pattern, which is persistent PDF state the wire pass
leaves set — a HIDDEN-linetype last wire printed glyph outlines dashed.
export_table's doc claimed it honoured TEXTFILL and widened the bold pen
itself; it did neither. Gate fill_tris on textfill() so hollow-on-screen
text exports hollow, carry the bold flag and widen the pen by the 1.7x the
bake uses, and resolve Face once per family instead of once per glyph (the
atlas mutex is held for that whole walk).
uv_key identifies a glyph by its tile's uv_min, but grow_height rescales
every entry's V and reset rewinds the packer, leaving already-built quads
addressing the wrong tile — garbage on screen and a silent miss in the
export table, i.e. #385 again under #347's grow-the-atlas conditions. Bump
an atlas generation on both and fold it into the tessellation memo guard so
the text re-lays-out instead. Quads built earlier in the same pass that
grows are still stale for that pass; the guard heals them on the next one.
Tests: the integration test's "some wire has text_verts" assert passed on
pre-fix code too (the exporter ignored them; the scene always carried
them), and its dimension half was satisfied by the TEXT alone — assert per
entity handle instead, and stop colliding on fixed names in the shared temp
root. The unit test's bbox came from the same quads the mapping reads, so
it held by construction; pin the run's own world box instead. That still
cannot catch a mirrored corner assignment — verified by mirroring it — so
say so rather than claim coverage the assert doesn't have.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Render SDF text as vectors in PDF/print export (#385).
Text became SDF-only, so all text — standalone TEXT/MTEXT and dimension
text — rode WireModel::text_verts as GPU glyph quads that the CPU PDF
exporter never sampled. Reconstruct the vector geometry at export time via
a new GlyphAtlas::export_table() keyed by the quads' atlas uv_min.
The annotative feature (1b037a6) added a 15th `anno_scale: f32` parameter to `scene::cache::block_cache::expand_insert` and updated the production call sites, but the text_font_rendering integration test was left at 14 arguments, so the test binary fails to compile with E0061 once rebuilt.
Pass `1.0` (no annotative scaling), matching the `BlockCache::build(&doc, 1.0, ...)` call the test already uses.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Since text became SDF-only (the legacy stroke-text path was removed), all
text — standalone TEXT/MTEXT and dimension text — rendered on-screen only as
GPU glyph quads carried on `WireModel::text_verts`. The CPU PDF/print exporter
draws only wire stroke `points` and hatch fills, so it silently dropped every
glyph: dimensions and text were missing from exported PDFs and prints
(present in v0.7.6).
Reconstruct vector text in the exporter instead of resurrecting the stroke
path, so the SDF render path and its caching are untouched:
- sdf_atlas: `GlyphAtlas::export_table()` snapshots each baked glyph's vector
geometry (outline strokes + fill triangles, glyph space) keyed by its tile
`uv_min`, re-resolved through `Face` so it honours the current TEXTFILL.
- pdf_export: a new `emit_text` pass walks each wire's `text_verts` (one
6-vertex quad per glyph), recovers the glyph geometry by `uv_min`, and
affine-maps it into the quad — `DrawLine` polylines for stroke (LFF/SHX)
fonts, filled triangles for TrueType, plus solid decoration bars. Runs under
the same rotation/scale/clip CTM as the wires, so it covers model space,
paper space and windowed plots, and every SDF-text kind (text, dimension,
multileader, table, tolerance).
Text embedded in complex linetypes is SDF-only with no stroke fallback and
stays out of scope.
Tests: a unit test asserts a stroke glyph emits vector ops within the glyph's
world bbox; an integration test drives a real TEXT + linear DIMENSION through
scene.entity_wires() -> export_pdf and checks text reaches the file. Verified
bug-first: with emit_text disabled the end-to-end test fails with the reported
symptom (text adds no content).
Co-Authored-By: Claude Opus 4.8 <[email protected]>
A symbol font's glyphs straddle the cap box by design — the diameter sign's
stroke overhangs it top and bottom — so anchoring a top-aligned line by the
nominal cap height alone hung it above where the drawing puts it, by the size
of the overhang.
Measure how far the first line's glyphs reach above the baseline and anchor by
that when it exceeds the cap. The fold is one-directional, so ordinary letters
(which stop at the cap) keep their exact previous offset.
The test pins the benchmark's own case: without this, the glyph draws from
0.7372 instead of the insertion point — matching the offset measured against
the file's own LINE endpoints for handle 1087.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
`solid_hatch_model` cast each corner from absolute WCS straight to f32 and
left `world_origin` at [0, 0]. At UTM magnitudes (~1e6) f32's 24-bit mantissa
resolves to ~0.06 units, so the quad snapped to a grid and the fill drifted
off its own outline.
Hold the corners in f64 until the AABB centre is known, then store each as a
small f32 offset from it — the same precision-preserving anchor
`hatch_model_from_dxf` already uses for real HATCH entities. All nine callers
feed through this one function, and every reader of `boundary` already adds
`world_origin` back, so they go from adding zero to adding the real anchor.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
A tolerance drew one row of ASCII soup — "RUN | ANGtol(M) | ..." — about
a tenth of its proper size. Six defects, each independent.
Rows were split on the two-character string "^J". A DWG stores a raw
newline, and the text-DXF reader rewrites the on-disk ^J to one before we
see it, so that split never fired on either mainstream path and every row
collapsed into the first. It now normalises all four spellings the break
arrives in — including the \P our own DXF writer emits, which made
open-DWG → save-DXF → reopen asymmetric.
Symbols were replaced with ASCII labels from a table that was itself
wrong: it read r as circular runout, n as angularity, p as position. The
font's own encoding says concentricity, diameter, projected tolerance
zone, and the benchmark's pre-exploded reference agrees. That table was
an inversion of acadrust's, which is self-contradictory — it binds n to
two different meanings. Do not re-derive from it.
The frame is a proportion of the character height: a compartment spans
-h..+h about its row's centreline, so it is 2h tall and the margin falls
out of that. It was built as height + 2*DIMGAP instead, which is a
different entity's geometry — DIMGAP sizes the dimension line's break and
the rectangle around a basic dimension, where it genuinely applies twice
per axis. Even fed a self-consistent style that can only ever produce
1.5h, so it was wrong in kind rather than mis-fed.
Height came from the entity's own field, which a DWG never sets — the
reader carries only the style's handle, leaving a small non-zero default
that a "> 0" test happily used. It now resolves the style by handle first
and name second: a DWG records the handle and defaults the name, a DXF
the reverse, so matching on the name alone would have picked the wrong
style for every DWG-read tolerance. An entity may also override single
style variables on itself, and this one halves its character height that
way; ignoring that drew the frame several times too large.
Rows stacked upward, every row inherited the widest row's columns, and
one full-width border boxed them all — none of which showed while the
parse only ever produced a single row. The insertion point sits on the
first row's centreline; anchoring the frame's bottom-left there offset it
by its own height, and by an amount that changed with the row count.
Cell text ran together because the glyph run asked for no letter spacing
while the width measurement assumed the font's own, so the characters
crowded inside correctly-sized boxes. Compartments are also measured by
what they draw now: the pen advance stops one letter-space past the last
glyph, and counting that made every one of them too wide.
Symbols come from a bundled stroke font, restored from this repository's
own history: it shipped QCAD's gdt.cxf until the text engine moved from
CXF to LFF, which removed every .cxf and left nothing in their place —
the ASCII labels date from exactly there. Converted to LFF, keyed by the
letter its \Fgdt;X escape carries, and registered under that name like
any other font, so a switch to it resolves through the ordinary lookup:
an MTEXT carrying the escape, or a text style named gdt.shx, reaches it
too, with no code of their own. The geometry is public domain by explicit
waiver; provenance is in the file's header.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
A benchmark drawing carrying a circle and an ellipse 800 km below its own
plane showed neither, and ZOOM Extents zoomed out to 800 km to find them.
Both come from sizing the view off the wrong axis.
Zoom and clipping answer to different extents and are now measured that
way: the zoom frames what the view actually shows, the extent across the
screen plane, while near/far span what lies along the eye direction.
Sizing both from the 3-D diagonal let one far-off Z drag the horizontal
zoom out with it. The two agree on a flat drawing, which is why it went
unnoticed.
A camera restored from a file's saved view keeps its pose, but nothing
had sized its near/far, so the projection fell back to a guess scaled
from the zoom distance — far too narrow when geometry reaches off the
plane, and the geometry was silently clipped away. It now takes the
drawing's own recorded extent.
$EXTMIN/$EXTMAX is untrustworthy for positioning — stale after an edit,
or the sentinel when the writer never computed it — which is why fit_all
scans entities instead. It is the right input here: too wide only costs a
little depth precision, too narrow deletes geometry, so the drawing's own
claim errs in the safe direction. It also costs nothing on open.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Installing a newer version of a plugin whose previous version is currently
loaded failed on Windows. The marketplace writes the native library to the
same filename, but a resident cdylib is memory-mapped by its runner process,
so the overwrite is rejected with "being used by another process" (os error 32).
Users hit this upgrading e.g. Land Survey 0.4.0 -> 0.4.4 and assumed a reboot
was required.
A loaded DLL can't be truncated or removed, but it *can* be renamed (the loader
opens it with FILE_SHARE_DELETE). install() now moves every resident library in
the package aside to a `.old` stash and writes the new one under its own name;
the stash keeps serving the running session and the loader picks up the fresh
library on the next start. Stale stashes are swept on the following upgrade, and
this subsumes the previous differently-named-library cleanup.
Adds regression tests, including one that reproduces the exact lock on Windows
(FILE_SHARE_READ | FILE_SHARE_DELETE, no write share) and confirms an in-place
overwrite fails while replace_library upgrades successfully.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Donate, Send Feedback, OCS Web and Manage > About all answered "No
drawing open — use New or Open first." and did nothing.
Two doors guarded the welcome tab and the outer one shadowed the inner.
`dispatch_command` already knew which commands stand alone — About,
Donate, Report and the links were all on its list — but
`on_ribbon_tool_click` refused every ribbon event before dispatch was
ever reached.
The welcome page's own buttons emit RibbonToolClick, so they are only
ever clickable while `is_start` holds: the refusal covered exactly the
case they exist in, and they had never worked. New Drawing / Open File /
Plugins send their own messages, which is why the page still looked
alive.
So the ribbon door now defers commands to dispatch and keeps refusing
only what touches the scene. The policy stays in one place instead of
being half-remembered in two. A drawing tool is still turned away on the
welcome tab (#299) — the wording just comes from dispatch now.
The same gap was swallowing tools the issues never mentioned: CUI and
ALIASEDIT configure the application, not a drawing, and were missing
from the list, so fixing only the ribbon door would have moved their
refusal one door down rather than removing it. HELP and CUILOAD /
CUIIMPORT join them.
The new test drives the real message path and fails on the old code with
the reported wording. It reads the allowlist from source rather than
dispatching DONATE / REPORT / WEBVERSION, which shell out to a browser —
that also fails the day someone adds a welcome-page button and forgets
the list.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Double-clicking a drawing started a whole second copy of the app. Now it
lands as a tab in the editor already open, and that window comes forward.
The election is the bind itself: every GUI launch tries a deterministic
per-user loopback port, and the OS grants it to exactly one process and
releases it on any death, SIGKILL included. Nothing to reap, no PID to
probe, no stale lock after a crash. The winner serves the port from an
iced subscription; the loser checks it is talking to a matching editor,
hands the paths over and exits. Every surprise — no answer, a stranger on
the port, a timeout — boots a normal window, i.e. the old behaviour: the
feature can degrade, but it cannot lose the file.
The gate in main is positional, after every headless mode has returned.
The plugin runner is this same binary re-spawning itself, so a flag list
would rot the first time a mode is added; a position cannot.
Handing a drawing over that is already open switches to its tab instead
of loading a second copy, comparing resolved paths so a symlink or a
`..` hop is still recognised as the same file.
`opening` is a single slot that a second open would overwrite, and
on_file_opened drops any result arriving once it is clear — so drawings
handed over at once would silently vanish. They queue in pending_opens,
drained from every path that clears the slot, error and cancel included.
Exec becomes %F so one launch takes the whole selection; the CLI takes a
list and boot fans it out through the same door. --new-instance forces a
separate process; --read-only / --script / --new always get their own.
Not reused: the automation server. It dispatches arbitrary commands and
writes arbitrary paths, and its open replaces the active tab's document
in place — publishing that on a port every local process can reach would
be unforced privilege escalation. Only its JSON-lines framing is shared.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Frequent entity-only edits (MOVE/ROTATE/SCALE/MIRROR/ALIGN, COPY/ARRAY,
draw-add, ERASE) captured a full ~800k-entity CadDocument clone (~46ms) on
every edit — the last big per-edit cost after the incremental-render work.
They now record only the touched entities' before/after images:
- Scene self-records through the five mutation primitives when a recording is
open (UndoRecording + is_recording_undo guard, zero cost off-path).
- apply_entity_delta installs one side per handle (in-place / re-insert with
the original handle / remove) plus a per-handle cache reseed, then a single
bump_entities so the incremental caches patch instead of a full re-tess —
undo/redo of a small edit is fast too, not just the edit.
- HistorySnapshot is now Full | Delta; a Delta is self-inverting and moves
between the undo/redo stacks without re-capturing. Commands that touch
non-entity state (new layer, group, *D block, viewport) fall back to the
full snapshot via per-command predicates, and a poison flag debug-asserts
any mis-scope.
remove_entity leaves the handle dangling in its owner block record's
entity_handles (harmless on its own, a lookup miss is skipped), so a
re-insert strips stale entries first — otherwise add_entity would list the
handle twice and the DXF/DWG writer would emit the entity twice.
Round-trip tests cover transform/erase/add/copy (do->undo->redo equality,
handle preservation, no duplication).
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Three cuts to the per-edit cost, on top of the incremental wire arena:
- Enable OCS_WIRE_GPU_PATCH by default (kill-switch OCS_WIRE_GPU_PATCH=0), so the
in-place / append-tombstone arena — validated on the reference drawing (wire
block ~300ms + 265MB upload -> ~2ms) — is on for everyone; a bail always falls
back to the identical batched path.
- Resident patch: an all-Modified edit never reorders, so reassemble by walking
the recorded assembled order (~10k handles) instead of scanning the whole
document (~800k) to re-derive draw order. Add/Remove still scan (positions
change). ~14ms -> <1ms on a move.
- SDF text: a new content id misses the glyph cache every edit, but if no
text-bearing entity (Text/MText/Dimension/MLeader/Leader/Table/attribute/INSERT)
changed since the last build the glyphs are identical, so reuse them. ~23ms -> 0
on a plain-geometry edit.
A move/edit on this 800k-entity drawing now spends ~50ms (dominated by the undo
document clone) instead of ~85ms, and the wire GPU upload is gone.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
draw_depth_map rebuilt on every geometry_epoch bump — an O(document) walk (all
objects for SortEntitiesTable, all entities grouped-by-block and ranked), ~45ms
per edit on a 788k-entity drawing, and it runs on every viewport snapshot.
But draw order depends only on each block`s entity COUNT (the rank denominator),
the entities` handles, and the SortEntitiesTable — none of which a Modify (move /
rotate / scale / colour / hide) changes (owner and type are preserved, and the map
ignores visibility). Gate the cache on the journal: an all-Modified edit keeps the
map; only Add / Remove (count change) or a table change (a `full` delta) rebuilds
it. So the common move/edit now skips the walk entirely.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
The single-batch arena required a set with no mesh/solid edges, so any drawing
with a 3D solid or mesh (this one: 1693 mesh-edge wires of 12200) disqualified the
WHOLE set — every edit fell back to build_wire_buffers re-uploading ~265 MB
(4.4M instances), stalling ~1s per edit on a shared-memory GPU.
Split the resident set into a regular-2D arena and a mesh-edge arena (the wires
that need the draw loops mesh-edge skip/black treatment), one WireArena each,
passed as &[&WireModel] subsets (no clone). Both patch incrementally; the mesh
arena carries is_3d_mesh_edge so the existing draw loop handles it. A changed
handle absent from an arenas subset is a no-op for it, so each edit touches only
the batch it belongs to. Result: the wire block drops from ~300ms + a 265 MB GPU
upload to ~2ms.
Two correctness points folded in from adversarial review:
- mesh-edge wires are stamped draw_depth 0 (matching WireGpu::from_run) so their
hidden back edges dont peek through the shaded fill; the mesh arena also always
bails on append (all its wires share depth 0, so relocation would reorder
coincident edges).
- buffers are capped at wgpus 256 MB limit — a batch larger than one buffer
returns None so the caller chunks it via the batched path (fixes a create_buffer
panic on the 355 MB regular batch).
Still behind OCS_WIRE_GPU_PATCH (default off). OCS_PERF prints the per-edit wire
timing + outcome.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
The in-place-only arena still fell back to a full batched rebuild on add/remove —
re-uploading the entire wire instance buffer, which on a drawing whose wires
expand to millions of segments (~265 MB) stalls ~1s per add/remove on a
shared-memory GPU. Only move/edit avoided it.
Add append/tombstone so add/remove patch the instance buffer too: a new or
segment-count-changed entity bump-allocates a slab at the tail (buffers carry
1.5x headroom); an erased entity is tombstoned (blanked at const slot 0). The
huge instance buffer only grows/blanks by the changed entity.
Three correctness points make add/remove safe (each verified by an adversarial
review, findings folded in):
- draw_depth_map normalises each entity`s z-bias by the block entity count, so a
structural change re-scales every entity`s bias. Keep a CPU mirror of the const
array; on any structural change refresh every slab`s draw_depth and re-upload
the small (~1 MB) const buffer, leaving the instance buffer untouched.
- a tail-appended entity draws last, which only mis-orders wires whose overlap
resolves by submission order rather than the z-bias: transparency, and 3D-solid
fallback edges (Solid3D/Region/Body/Surface are excluded from draw_depth_map, so
their opaque edges share depth 0). `append_unsafe` bails to a full rebuild when
the set contains either.
- an in-place edit also flushes its consts to the GPU (a colour change is not
structural, so the whole-buffer refresh would otherwise skip it).
Still behind OCS_WIRE_GPU_PATCH (default off); every uncertain case returns false
and the caller does the identical full rebuild, so correctness never rides on the
fast path.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
hide_selected, end_isolation and register_solid_model called bump_geometry() —
a full-drawing invalidation (clears the tessellation memos, bumps block_epoch) —
even though each touches a known, bounded set of handles. On a large drawing
hiding a few objects then re-tessellated everything still visible.
Report the exact changed handles via bump_entities instead: hide/reveal mark
just the affected entities Modified so the resident set drops or re-adds only
those, and register_solid_model reports its one solid (also making a bulk
register loop O(n) rather than O(n^2)). No block-definition changes are
involved, so skipping block_epoch is correct.
Deliberately left on the full path: layer freeze/off (block sub-entity layer
visibility legitimately needs block_epoch), viewport-scale (annotation-key
sensitive), isolate (already cheap — keeps few), and undo/redo/style/file-open.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Point the acadrust patch at the OpenAEC-Foundation/acadifc fork (still the
`acadrust` crate). Its main carries the Arc copy-on-write entity storage that
makes the undo document clone O(entities) refcount bumps instead of a full deep
copy — cutting the per-edit snapshot on a 788k-entity drawing from ~1.7s to
~70ms. No source changes here: acadrust`s public API is unchanged.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
On a geometry edit the resident set gets a fresh content id, so the batched wire
path (build_wire_buffers) re-emits and re-uploads EVERY wire instance — the whole
buffer even when one line moved. Behind the OCS_WIRE_GPU_PATCH flag (default off,
native only), add a persistent per-entity instance arena: one long-lived
VERTEX|COPY_DST instance buffer + its WireConst storage, with a per-entity slab
table. An in-place edit (move / rotate / scale / colour — same segment count)
becomes a queue.write_buffer of just that entity`s slab.
Scope is deliberately IN-PLACE ONLY. draw_depth_map normalises each entity`s
draw-order z-bias by the block`s entity count, so an Add or Remove re-scales EVERY
entity`s z-bias — which a per-slab patch can`t honour — and a segment-count change
relocates the entity. patch() returns false for any of those (and for mesh-edge /
scissored sets), and the caller falls back to the identical full batched rebuild,
so correctness never rides on the fast path. An adversarial review of an earlier
append/tombstone design surfaced exactly these draw-order hazards; restricting to
in-place removes them while keeping the common interactive-edit win.
The Scene hands the changed handles to the arena via model_wire_gpu_patch,
recorded by try_resident_patch for the Model block and matched by content id in
prepare(); scissored paper viewports and mixed 2D/3D sets keep the batched path.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
MATCHPROP only copied the generic properties (layer, color, linetype,
linetype scale, lineweight), so matching text to text left the
destination's style and height unchanged. Copy the source's text style
and height to TEXT and MTEXT destinations, in both directions, like
AutoCAD's "Text" special property.
Adds an in-process regression test that drives the real interactive
command path (source pick -> destination selection) and fails on the
old code with the reported symptom.
Fixes#361
Co-Authored-By: Claude Fable 5 <[email protected]>
On a geometry-only edit the resident wire set was rebuilt from scratch: every
memoized entity deep-cloned into a fresh Vec (mod.rs clone-all) and the whole set
re-sorted by draw order — O(total wires) on each single-entity change even though
the per-entity tessellation memo already reused everything.
Add try_resident_patch: when the cached set is only a journal-replayable gap
behind and uniquely held, move its wires out (Arc::try_unwrap), group them into
per-entity runs, re-tessellate only the changed entities, and reassemble in
document order with the same stable draw-order sort. Unchanged entities are MOVED,
not deep-cloned; only the changed handful are re-tessellated (into the memo, so a
later full build still reuses them). It bails to the identical full build whenever
an assumption does not hold — no cache entry, journal not replayable (full delta /
ring overflow), the Arc still shared, a wire not named with its handle, or an
entity`s wires not contiguous — so correctness never rides on the fast path.
A changed entity is classified through the shared resident_entity_visible (also
now used by the full build) so it is included/excluded exactly as a rebuild would.
Adds a differential-oracle test: the patched resident set equals a from-scratch
rebuild (same wires, same draw order) after add / move / erase, and asserts the
patch path actually runs for every edit (a resident_patch_hits counter) rather
than silently always falling back.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
The pick/snap spatial index (entity_index) rebuilt its whole quadtree on every
geometry_epoch bump, and annotation_affects_wires re-scanned every entity to
decide whether any is annotative — both O(document) on each single-entity edit.
Drive both off the delta journal. entity_index now replays the changes since its
cached epoch and applies them to the existing quadtree (which already supports
insert/remove/update, out-of-root items falling into overflow): Removed drops the
handle, Added/Modified update its AABB or move it to the unbounded list, and only
a full/overflow delta triggers the old full rebuild. annotation_affects_wires
stays warm unless an annotative entity changed (PSLTSCALE toggles arrive as a
full delta and still recompute).
Adds a differential-oracle test asserting the incrementally-patched index equals
a from-scratch rebuild after add / move / erase.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Every geometry mutation previously bumped a monotonic geometry_epoch and each
derived cache compared its stamp and, on a miss, re-walked the whole document —
so drawing or moving one line rebuilt every hatch / image / mesh / wipeout model
even though none of them changed.
Add a per-mutation delta journal: bump_entities(&[(Handle, ChangeKind)]) records
exactly which handles changed (Added / Removed / Modified), folding the memo drop
and the epoch bump into one call a site cannot forget. A bounded ring keeps the
recent deltas; replay_since(last_epoch) coalesces the changes past a consumer or
returns None (full rebuild) when a spanned delta is `full` or the consumer fell
off the ring. bump_geometry / bump_geometry_no_blocks push a `full` delta so the
journal is never missing a step.
Route the common single-entity ops through it: add (Added), update (Modified),
erase (Removed), move/rotate/scale/mirror (Modified), copy (Added). Gate the four
per-category model caches on category relevance — an edit whose handles are not in
a category leaves that category cache warm; a removal is treated conservatively
(the erased handle can no longer be classified), and the mesh cache also
invalidates on any INSERT edit since it carries the instanced block meshes.
Net: a plain-entity edit no longer rebuilds the hatch/image/mesh/wipeout model
sets. The resident wire clone-all and the GPU re-upload are still whole-set — the
journal is the groundwork for making those per-entity next. Adds 6 journal unit
tests (replay, coalescing, full-fallback, ring overflow, category gate).
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Export to PDF could end in complete silence: the flow hands off to an
async rfd save dialog, and when that future resolves to None the app
mapped it straight to Task::none(). None does not only mean the user
cancelled — on Linux the XDG portal SaveFile call can fail outright
(rfd then tries a zenity fallback, which may not be installed) and the
future still just resolves to None. rfd reports those failures only
through the `log` crate, and logging is opt-in (--log / RUST_LOG), so
the reporter saw exactly what #369 describes: no dialog, no error, no
console output, no file.
- EXPORT / EXPORTPDF now accept an optional path argument that exports
straight to the file with no dialog — both the workaround for systems
where the native save dialog cannot open and a headless-testable
entry point. A relative path lands next to the drawing; a missing
extension gets .pdf.
- The dialog-None arms (PlotExportPath / PlotWindowExportPath) now push
a command-line message pointing at EXPORTPDF <path> instead of doing
nothing.
- The success message prints the full output path, not just the file
name, so users can find the PDF.
Verified end to end on the real binary via --serve: LINE + CIRCLE then
EXPORTPDF <path> writes a valid PDF with the expected geometry. Both
new tests fail on the previous code.
Fixes#369
Co-Authored-By: Claude Fable 5 <[email protected]>
Architectural / fractional dimension text split feet / inches / fraction
*before* rounding to the fraction grid, so a value like 35.9999" (float noise
on an exact 3ft object) rounded its fraction up to 16/16, reduced to 1/1, and
leaked out as a bare "1" with no carry — printing 2'-11 1" instead of 3'-0"
(#380). The fix rounds to integer 1/denom ticks first, then decomposes, so the
carry propagates through inches into feet. Fixes#380.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Each pipeline slot owned its own gpu_wires: every paper viewport and every
Model tile rebuilt an identical resident wire set into its own GPU vertex
buffers and its own handle index. With N content viewports drawing one shared
tessellation that is N copies of the (potentially huge) wire buffers in VRAM
plus N batch+index builds on every content change — the issue-131 memory
exposure multiplied by the viewport count.
A wgpu::Buffer is internally ref-counted, so sharing a WireGpu by Arc shares
its GPU allocation. Make Pipeline.gpu_wires an Arc<Vec<WireGpu>> (and
wire_handle_index an Arc<map>), turn upload_wires into build_wire_buffers(&self)
that returns those Arcs, and add MultiPipeline.wire_buffer_cache keyed by
wire_content_id. prepare() builds an entry once on a cache miss and hands Arc
clones to every slot with that id — so N viewports (and, since one MultiPipeline
backs all panes, N panes) upload the wire vertices exactly once between them.
Camera uniforms and scissor stay per-slot, so the draw still issues one call per
viewport with the right view; only the geometry is deduplicated. The shared Vec
is never mutated in place — only reassigned on a content change — so the sharing
is sound.
The cache drops entries no slot still references (Arc::strong_count == 1) once it
passes a small bound, so a live buffer (held by any pane) is never evicted.
Model multi-tile gets the same dedup for free, and two panes showing the same
model now share buffers.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Model space held a full, camera-independent, un-culled wire set with a
stable content id, so pan/zoom re-drew without re-tessellating or
re-uploading and idle frames collapsed to a single blit. Paper did none of
that: the sheet was camera-keyed (full re-tess every pan/zoom), content
viewports re-baked per zoom step with their own frustum cull + LOD, and a
per-frame wire_force_nonce forced a GPU re-upload every frame AND poisoned
render_signature so paper never hit the single-blit scene cache.
Route every space through one Scene::resident_wires_for(block, anno, frozen):
the full, un-culled, LOD-free set per (block, ambient bg, anno, frozen) key,
held resident and stamped with a stable WIRE_CONTENT_GEN id. Model tiles,
BEDIT, the paper sheet (a thin border-drop + printable-area dressing over the
resident base) and paper content viewports all share it. When annotation is
inert (PSLTSCALE off and no annotative entity, cached per epoch) the per-
viewport scale is normalized away so every viewport reuses one tessellation.
Result: paper pan/zoom does zero re-tessellation and zero wire re-upload, and
paper now hits the single-blit scene cache exactly like model. Deletes
wire_force_nonce, the per-viewport height/view wire cache, and view_cull_aabb.
Tradeoff (deliberate, matches model): no cull/LOD anywhere.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
A 3ft object measuring 35.9999" (float noise) printed "2'-11 1"" because the
architectural and fractional formatters split feet/inches/fraction *before*
rounding. A fraction that rounded up to a full unit (16/16 -> 1/1) reduced to a
bare "1" and leaked out with no carry into the inches or feet.
Round on the fraction grid first using integer ticks, then decompose, so carries
propagate: 35.99" now reads 3'-0" instead of 2'-11 1". Adds regression tests.
Fixes#380
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Picks up the DXF fix that reads DIMENSION group code 2, so dimensions
imported from DXF keep their baked *D block and render (their text no
longer drifts vs the DWG of the same drawing).
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Picks up the DXF over-annotative MULTILEADER fix (companion to the MTEXT
one): enable_annotation_scale now defaults false and DXF reads code 293.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Picks up the DXF over-annotative-MTEXT fix: acadrust now defaults
MTEXT.is_annotative to false, so imported DXF MTEXT is no longer flagged
annotative and over-scaled in annotation-scaled viewports.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
On native the button opens the OS save dialog (folder + name + overwrite)
and on web it starts the download, so it does not itself save — "Continue"
reads truer than "Save".
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Selecting a block and running BEDIT (command, ribbon button, or a
double-click on the block) opens a block-editor "space" tab, named after
the block, alongside Model/Layout in the status bar. The tab draws ONLY
the block record's own entities in block-local coordinates; edits are
live on the definition, so every INSERT updates. Save Block / Discard
buttons (side toolbar) finish the session; Discard restores the on-entry
snapshot. Space switching is locked while a block is open, and BEDIT and
REFEDIT are mutually exclusive.
Implementation: a Scene.block_edit_block override at the top of
current_layout_block_handle scopes render/pick/new-entity ownership to the
edited block while current_layout stays "Model" (so the pervasive
Model-vs-paper code keeps model behaviour). The model render/pick paths
that hardcoded model_space_block_handle (wire tile set, hatch render +
pick, block-instance meshes) are made binding-aware so the block isolation
actually takes effect and model content does not leak in. Entering fits
the camera to the block (fit_all already scopes via the binding). Renaming
the block tab renames the block itself — new Scene::rename_block re-keys
the record, updates the Block marker and repoints every INSERT reference.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
WebGL2 has no vertex/fragment storage buffers, so the batched storage-buffer
hatch renderer (hatch_gpu) is disabled on wasm and real hatch fills never
rendered on the web build — the pick highlighted the boundary and the command
then appeared to cancel (issue #204).
Add a wasm-only per-hatch renderer (hatch_web_gpu + hatch_web.wgsl) that packs
the variable-length boundary / family / dash arrays into one RGBA32F data
texture read with textureLoad, instead of the fixed-size uniforms of the
wipeout path. This removes the MAX_FAMILIES (16) / MAX_HATCH_BOUNDARY_VERTS
(1024) / MAX_DASHES (128) caps, so every hatch — solid, gradient, and
arbitrarily complex line patterns — renders on the web. The fill algorithm
(in_polygon, per-family line/dash/dot, gradient, LOD, double-single vertex
transform) is identical to the desktop path; only the array ACCESS changes.
Pattern phase is anchored at world_origin with the boundary stored raw, and
scale / gradient-range clamped, to match the desktop batched renderer exactly.
Native (hatch_gpu + hatch.wgsl) and the wipeout mask renderer are untouched;
the new module and pipeline are gated on target_arch = "wasm32".
Also silence the resulting (and pre-existing) wasm-only dead-code warnings for
native-only paths so both the native and web builds compile clean.
Closes#204.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Dashed/dash-dot lines could end in a gap ("empty space at the end"),
especially inside paperspace viewports. The dash pattern was only centred
on each wire, which makes the two ends symmetric but does not force them
onto a solid dash.
Implement AutoCAD "A" (aligned) linetypes: the line begins AND ends on a
solid dash, interior dashes stay nominal, and the two end dashes stretch
symmetrically to absorb the remainder (so parallel lines share an
identical interior). wire_distances computes the end-dash length D and
total length L for a dash-first pattern (D = (L - k*P + a) / 2); the GPU
in_dash walk force-lights the two end regions and phases the interior to
resume right after the start dash. Lines shorter than one pattern period
draw solid.
Only alternating dash-first patterns are aligned; gap/dot-first,
consecutive-dash, NaN-break (plinegen=false) and solid lines keep the
prior behaviour. Native and web (wasm) shaders kept in sync.
Refs #13.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Make objects genuinely annotative — carrying real per-scale
representations that interoperate with other CAD apps — rather than just
a native flag. `create_annotation_context` synthesizes the extension-dict
chain (AcDbContextDataManager -> ACDB_ANNOTATIONSCALES -> per-scale leaf)
from the entity's placement; the editable Annotative toggle now covers
Text and block references as well as MText/MLeader; and a new
"Annotation Object Scale" dialog (OBJECTSCALE) adds/removes an object's
per-scale memberships. Requires the acadrust ObjectContextData encoder
(Cargo.lock bump).
Also fixes a render bug where objects carrying an *empty*
ACDB_ANNOTATIONSCALES (a single-representation marker with no per-scale
reps) were treated as annotative and (mis)scaled by the annotation factor
in scaled paper viewports, ballooning the text: an object is now
annotative-by-context only when its scale collection is non-empty, so
such objects render at their base geometry.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Pipeline slots are addressed by list index, but the renderer drops
off-canvas viewports from the list, so a slot can be reused by a
different viewport across frames — e.g. when the first paper viewport
scrolls off the canvas the second slides into its slot and inherits its
stale, differently frustum-culled GPU buffers. The text upload is gated
on a geometry-epoch change (unaffected by the index shift), so the
survivor rendered the gone viewport's text and its own vanished.
Give each viewport a stable identity, remember which one last used a
slot, and reset all cache keys when the occupant changes so wires, text,
hatches, meshes and Face3D re-upload for the new viewport. No cost in the
steady state — the reset fires only when a viewport enters/leaves view.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Follow-up to #374. resolve_hatch_rings flagged every outline nested in
the fill ring as a hole, so with 3+ nesting levels a click in an outer
band added grandchild islands too — even-odd fill then re-filled the
innermost shape. Skip a candidate when another outline sits strictly
between it and the fill ring, so only the immediate next level holes out.
Closes#377.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
HATCH/GRADIENT/BOUNDARY pick-inside now resolves the click against every
enclosing outline, fills the smallest-area one, and emits inner outlines as
NaN-separated holes. DXF export writes one boundary path per ring.
Co-authored-by: KarimJerbi <[email protected]>
The "Dim line color" row on Leader and Dimension entities was
read-only: an earlier attempt wired it to Leader.override_color, which
acadrust never serialises, so the pick was lost on save. Store it
instead as a standard ACAD_DSTYLE per-object dimension-style override
(DXF code 176, an ACI index) through the existing dim_override codec,
so it round-trips through both DWG and DXF like the other dim
overrides. RGB picks collapse to the nearest ACI, matching the rest of
the dim-colour stack (dimension styles are index-only through the file
layer). The renderer prefers the override over the style's DIMCLRD.
The write branch is guarded to leaders and dimensions so a mixed
selection cannot stamp the override onto other entity types.
Bumps acadrust to c9fe982, which carries parsed XDATA onto dimensions
on DXF read (it was dropping common.extended_data), so the dimension
override persists on DXF save as well as DWG.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Some DWGs written by other programs leave the root named-objects
dictionary unrecoverable: the header pointer names a non-dictionary and
the real NOD child dicts are owned by a table-control handle, so nothing
has owner==NULL. Navigating the root then silently no-ops, and
registering a new named-object entry (page setup, annotation scale, CTAB)
against the missing root orphans it — it never round-trips.
Add shared annotative::root_named_dict_handle(): resolve the header
pointer, else adopt the richest owner==NULL dict, else synthesise a fresh
root and write it back. Route ensure_plotsettings_dict, add_scale's
scalelist-create branch, and set_saved_active_layout through it. Resolve
the plot-settings dict via the PlotSettings owner when the header pointer
is stale, mirroring scalelist_dict_handle.
Verified page setups survive a DWG save/reload on the repro file with
layouts intact.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Double-click rename did nothing and ratio edits were lost when switching
scales — and on drawings saved by other programs nothing could be edited
at all.
- Fallback scales were virtual (no Scale objects), so edit/rename no-oped;
materialize the standard set into real staged objects when the manager
opens, reverted on close unless applied.
- On third-party DWGs the mutators walked the root named-objects dict ->
ACAD_SCALELIST, which can be unreachable, and matched dictionary keys
that are internal ids (*A1, A0) rather than scale names. Locate the
scale-list dict robustly and match scales by object name + handle.
- Rename rows inline via double-click; repair a dangling current-scale
header pointer on staged rollback.
- Stage text/mleader style-manager edits on switch so they persist like
the scale manager.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
The scale manager now runs as a transaction, matching the style managers'
staging: New starts a blank editor draft (nothing is added to the drawing
until Apply), Delete removes immediately but staged, and closing the window
without Apply restores the snapshot taken on open — so a scale created or
deleted mid-session is discarded on cancel. Apply commits the staged list
changes as a single undo entry. This replaces the earlier behaviour where
New copied the selected scale and edits landed immediately.
A `changed` flag guards the commit so a no-op / failed Apply (blank or
invalid fields, or an edit that didn't apply) no longer dirties the drawing
or pushes a phantom undo entry. Set Current is deliberately outside the
transaction — it takes effect immediately like the scale pill (and syncs
CANNOSCALEVALUE), so it is never rolled back on close. Renaming the current
scale follows the pointer; editing one of the built-in fallback scales (shown
when the drawing has no scale list) materialises it as a real scale; the
snapshot restore touches only the ACAD_SCALELIST root entry.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
Replace the scale popup's inline delete/add affordances with a single
"Manage scales..." row that opens a proper manager, matching the style
managers: it reuses the shared style-manager palette and chrome (toolbar +
list + editor frame), so it looks and behaves like the Text / Dim / MLeader
managers.
The manager lists the drawing's annotation scales (name + ratio, the current
one checked) and edits them through a Name / Paper units / Drawing units
editor: New adds a scale from the fields, Apply renames / re-ratios the
selected one, Delete removes it (never the current scale), and Set Current
makes the selected scale the model-space annotation scale. Backed by
Scene::add_scale / remove_scale / edit_scale on ACAD_SCALELIST.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
The drawing's annotation-scale list (ACAD_SCALELIST) is now editable, so the
set of scales offered by the status-bar scale pill can be grown or trimmed:
- Scene::add_scale / remove_scale create or drop a Scale object and its
SCALELIST dictionary entry (creating the SCALELIST dictionary, and
registering it in the root, for drawings that carry none).
- SCALELISTEDIT gains ADD <ratio> / DELETE <name> alongside the existing
listing (e.g. `SCALELISTEDIT ADD 1:50`).
- The scale popup (model space) grows a delete affordance on each non-active
scale and a "+ Add scale..." row that primes the command line. The current
annotation scale is protected from deletion.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
The Annotative row becomes an editable toggle for the two entity types that
carry a native per-object annotative flag — MTEXT (is_annotative) and
MULTILEADER (enable_annotation_scale). Turning it on flags the object
annotative (it then scales with the current annotation scale in model space);
turning it off clears the flag AND removes the per-object annotation context
(the AcDbContextDataManager -> ACDB_ANNOTATIONSCALES leaf subtree) plus the
legacy annotative XDATA markers, via scene::annotative::set_entity_annotative /
clear_annotation_context, so the object stops resolving annotative instead of
staying on through a stale context. The shared SCALE objects are left intact.
Style-derived types (text/dimension/leader/table/insert) keep a read-only
Annotative row — their state comes from the assigned style, and making a single
object annotative there needs a per-object context object (a dedicated
acadrust ObjectContextData writer) which is out of scope here. Verified an
existing annotative DWG round-trips its 134 per-object contexts through a
save/reload unchanged, so the far more common open-and-save path is unaffected.
Co-Authored-By: Claude Opus 4.8 <[email protected]>
