Community poll: what should I build next in dxf-render / dxf-vuer?

[arbaev]

🗳️ Community poll: what should we build next in dxf-render / dxf-vuer?

Hey everyone! dxf-render and dxf-vuer have a long roadmap, and I'd love to hear from you which features matter most for your use cases.

1. 💬 Leave a comment with your top 3 features (just paste the IDs, e.g. R-1, I-1, UI-3).
2. 🎯 Briefly describe your use case — it helps us prioritize features that solve real problems.

Viewing only for now — DXF editing is explicitly out of scope.

---

🎨 Rendering — more DXF coverage

New entity types

• R-1 · IMAGE / WIPEOUT — raster images embedded in DXF (parser ready, needs render)
• R-2 · 3DSOLID — ACIS solids (at least wireframe / bounding box)
• R-3 · VIEWPORT — render model-space content inside paper-space viewports
• R-4 · TABLE (ACAD_TABLE) — tables with cells and styles
• R-5 · TOLERANCE — GD&T feature control frames
• R-6 · REGION — 2D regions

Existing entity improvements

• R-7 · SHX fonts — AutoCAD SHX support (conversion or custom parser)
• R-8 · Vertical text — TEXT/MTEXT with vertical flow direction
• R-9 · Bidirectional text — correct RTL ordering (Arabic, Hebrew)
• R-10 · Full DIMSTYLE coverage — all DIMSTYLE variables
• R-11 · Full MLINESTYLE — fills, end caps, joints
• R-12 · Gradient HATCH — gradient fills (currently solid + patterns only)

Paper Space / Layouts

• R-13 · Layout switching — toggle between Model Space and Paper Space
• R-14 · Title blocks & sheets — borders, title blocks, sheet scales
• R-15 · Layout selector prop/component

DWG support

• R-16 · DWG parsing — DWG → DXF via WASM lib (LibreDWG / ODA File Converter) or a native binary parser
• R-17 · Format auto-detection — detect DWG/DXF by file signature and handle transparently
• R-18 · DWG version coverage — R14 / 2000 / 2004 / 2007 / 2010 / 2013 / 2018

---

⚡ Performance & scalability

• P-1 · Streaming parser — incremental parsing for large DXFs (>50 MB)
• P-2 · Level of Detail (LOD) — geometry simplification at low zoom
• P-3 · Frustum culling — skip off-screen objects
• P-4 · GPU instancing — InstancedMesh for mass INSERT blocks
• P-5 · Lazy block loading — defer rendering blocks outside viewport
• P-6 · Memory optimization — BufferGeometry reuse, proper dispose on file switch
• P-7 · Progressive rendering — show partial scene during load

---

🧭 Navigation & interaction

Camera

• N-1 · Minimap — overview navigator in the corner
• N-2 · Named views navigation — jump to VPORT / Named Views from DXF
• N-3 · Zoom to selection — zoom to selected layer or region
• N-4 · Zoom to layer extents
• N-5 · Navigation history — undo/redo camera positions
• N-6 · Animated transitions — smooth camera tweens on fit/zoom

Measurements & info

• I-1 · Distance measurement tool — ruler between two points
• I-2 · Area measurement — closed region
• I-3 · Angle measurement
• I-4 · Snap to geometry — vertices, midpoints, centers
• I-5 · Entity inspector — click → properties (type, layer, color, coords)
• I-6 · Attribute search — text search across INSERT ATTRIB values

Selection

• S-1 · Click to select — highlight entity on click
• S-2 · Rectangle selection — marquee select
• S-3 · Layer hover highlight — emphasize all entities on layer hover
• S-4 · Spatial association heuristic — heuristic association of leader/dimension to the nearest geometry pointed to by the arrow (optional flag, experimental — DXF does not store this association)

---

🖼️ UI & UX

Panels

• UI-1 · Properties panel — details of the selected entity
• UI-2 · Blocks panel — list all INSERT blocks with navigation
• UI-3 · Text/dim style panel — overview of styles in use
• UI-4 · Extended statistics — charts of entity/layer distribution
• UI-5 · Breadcrumbs — Layout > Block > Entity path
• UI-6 · Context menu — right-click → Zoom to / Properties / Copy coords

Layer panel

✔️ L-1 · Layer search filter — text filter by layer name (done in #9)

• L-2 · Layer grouping — auto-group by prefix (A-WALL, A-DOOR → group A)
• L-3 · Drag-and-drop layer order
• L-4 · Persist visibility — remember layer state across sessions (localStorage)
• L-5 · Visibility presets — quick configs ("architecture only", "electrical only")

Toolbar

• T-1 · Button groups — visual separators between groups

---

📤 Export & integration

Export

• E-1 · SVG export — vector export of visible layers
• E-2 · PDF export — preserving scale and layers
• E-3 · PNG export options — resolution, DPI, transparent background
• E-4 · Copy to clipboard — screenshot current view

Print

• E-5 · Print layout — print prep with border and scale
• E-6 · Print-friendly CSS — @media print styles

Loading

✔️ E-7 · Load from ArrayBuffer / Blob — file-storage integration (done in #9)

• E-8 · Authenticated URL loading — custom headers on URL fetch
• E-9 · .dxf.gz support — auto gzip decompression

---

♿ Accessibility

• A-1 · ARIA attributes — aria-label on all controls
• A-2 · Keyboard navigation — Tab/Enter/Space across UI
• A-3 · Keyboard pan/zoom — arrows to pan, +/- to zoom
• A-4 · Screen reader announcements — loading, errors, layer changes
• A-5 · Focus management — visible focus, correct tab order
  ✔️ A-6 · prefers-reduced-motion — disable TAA and transitions (done in release: dxf-render@1.3.0, dxf-vuer@2.4.0 #9)
• A-7 · High contrast mode — Windows High Contrast support

---

🧩 API & extensibility

DXFViewer

• API-1 · v-model for layer visibility — two-way binding
• API-2 · Events: entity-click / entity-hover — with entity data
• API-3 · Programmatic API — viewer.zoomToEntity(handle), viewer.highlight(entities)
• API-4 · Plugin system — register custom entity renderers

Other

• API-5 · Headless mode — server-side rendering (Node + headless GL) for thumbnails
• API-6 · React wrapper — dxf-react on top of dxf-render
• API-7 · Web Component — <dxf-viewer> custom element, framework-free

---

✅ Quality

• Q-1 · Visual snapshot tests — reference DXF vs. rendered image diff
• Q-2 · Parser fuzz testing — resilience to malformed DXF
• Q-3 · Performance benchmarks — 1 MB / 10 MB / 50 MB files
• Q-4 · E2E tests — Playwright for core scenarios
• Q-5 · Storybook / Histoire — interactive component catalog
• Q-6 · API Reference — auto-generated from TSDoc
• Q-7 · Reference DXF suite — compatibility across AutoCAD / BricsCAD / LibreCAD / QCAD

---

💡 Missing something?

If your must-have isn't listed, just comment below!

Thanks for helping shape the roadmap!

[snakefood3232]

I’m working with large-format CAD exports that often exceed 50MB, so P-1 (streaming parser) would unblock our pipeline on DXFs that currently time out or crash. I’d implement an incremental SAX-style tokenizer that yields entities as chunks arrive, keeping the existing API surface intact. Can have a PR up within 48 hours. I’ve built streaming parsers for SVG and STEP files in the past, so the pattern is familiar.

[arbaev]

Hi @snakefood3232! Thanks a ton for jumping in — SVG/STEP streaming experience is exactly the right background for this.
A few things before you cut a PR, so we're aligned:

1. Sample file, please. Could you share a drawing that currently times out or crashes? Even trimmed-down, a real 50 MB+ file would be gold — I'd reproduce locally and add it to the benchmark suite (P-3 on the roadmap). Synthetic is fine too.
2. Keep the public API intact. Entry points are parseDxf(text): DxfData and parseDxfAsync(text): Promise. I'd rather route internally (small files → current path, large files → your streaming tokenizer, both resolving to the same DxfData) than add a new public signature. Existing consumers get the speedup for free.
   One caveat: parseDxfAsync takes a string today, so the whole file is already in memory before we start. True streaming wants ReadableStream or similar — worth thinking about whether the input type should grow. Curious how you'd approach it.
   Progressive consumption (onEntity callback, async iterable) is interesting but out of scope for v1 — let's prove the streaming path works behind the existing API first.
3. Scope check. The parser has ~25 entity handlers, cross-referenced sections (HEADER/TABLES/BLOCKS/ENTITIES), and 854 tests in packages/dxf-render/src/parser/tests.
   Tokenizing is maybe 20% of the work — the rest is not breaking existing behavior. If you want to land this incrementally (streaming scanner first, then section-by-section), that's very welcome and much easier to review than one big PR.
4. What's the actual symptom — timeout or crash? They point to different fixes:

• Crash → OOM, streaming parser is the right call.
• Timeout → slow time-to-first-frame, which is more about progressive rendering (P-7) than parsing. A streaming parser won't help much if we still wait for full DxfData before rendering.

Want to make sure we're fixing the thing you're actually hitting.
No rush on 48h — quality > speed. Looking forward to the sample!