aj/OpenNest
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Add OpenNest.Benchmark: generic head-to-head engine comparison harness

Browse Source 
Loads any .nest file (or folder of them) via NestReader and nests every
drawing with quantity > 0 using each registered NestEngineBase, so it
works sight-unseen against arbitrary real jobs without any hardcoded
geometry. Optionally sweeps a fixed --sheet-sizes list instead of each
file's own plate size.

- BenchmarkJob/JobLoader build immutable job specs; a fresh Plate and
  NestItem list is created per (job, engine) run so state never leaks
  between engines or jobs.
- NestValidator rejects a layout if any part falls outside the work
  area, any two parts are closer than PartSpacing (checked via each
  part's own world-space polygon inflated by the spacing, so it holds
  for arbitrary concave/holed geometry, not just bounding boxes), or a
  drawing gets more parts than requested.
- Scoring matches Plate.Utilization() (placed area / full sheet area);
  ties among fully-placed layouts break on the smaller used bounding
  box (more usable remnant).
- Report prints a per-job ranked breakdown plus a per-engine summary
  (wins, avg utilization, time), and can write a flat CSV.

Verified end-to-end against a synthetic .nest file (not committed)
against the four built-in engines; caught a genuine out-of-work-area
bug in StripNestEngine in the process.
This commit is contained in:
aj
2026-09-15 18:31:36 -04:00
parent 587000f68a
commit 6a0fba0fec
11 changed files with 839 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files
CLAUDE.md
+10 -1
View File
@@ -18,7 +18,7 @@ NuGet dependencies: `ACadSharp` 3.1.32 (DXF/DWG import/export, in OpenNest.IO),
## Architecture
Eight projects form a layered architecture:
Nine projects form a layered architecture:
### OpenNest.Core (class library)
Domain model, geometry, and CNC primitives organized into namespaces:
@@ -69,6 +69,15 @@ GPU-accelerated pair evaluation for best-fit nesting. `GpuPairEvaluator` impleme
### OpenNest.Training (console app, depends on Core + Engine)
Training data collection for ML angle prediction. `TrainingDatabase` stores per-angle nesting results in SQLite via EF Core for offline model training.
### OpenNest.Benchmark (console app, depends on Core + Engine + IO)
Compares registered `NestEngineBase` implementations against each other on real `.nest` files. Fully generic — it never hardcodes drawing geometry, just reads whatever drawings/quantities/plate settings each input file already has.
- `JobLoader` builds `BenchmarkJob`s from a `.nest` file or a folder of them via `NestReader`, using every drawing with `Quantity.Required > 0`. `--sheet-sizes` can sweep a fixed list of plate sizes instead of each file's own.
- `BenchmarkRunner` gives each (job, engine) pair a fresh `Plate`/`NestItem` list (`BenchmarkJob.CreatePlate()`/`CreateItems()`) so engines can't see each other's mutated state, then calls the engine's `Nest()` and times it.
- `NestValidator` checks the returned layout: every part inside `Plate.WorkArea()`, every pair at least `Plate.PartSpacing` apart (checked geometrically via each part's own world-space polygon, inflated by the spacing — works on arbitrary concave/holed shapes, not just bounding boxes), and no drawing over its requested quantity. An invalid or throwing run scores zero for that job.
- Scoring matches `Plate.Utilization()` (placed drawing area / full sheet area, `Plate.Area()`). If an engine placed every requested part, ties are broken by the smaller used-bounding-box (`Report`'s ranking rule) — a more compact layout leaves a bigger usable remnant.
- `--engines Name1,Name2` filters to specific registered engines (default: all); `--csv <path>` writes a flat per-job CSV alongside the console report.
### OpenNest.Mcp (console app, depends on Core + Engine + IO)
MCP server for Claude Code integration. Exposes nesting operations as MCP tools over stdio transport. Published to `~/.claude/mcp/OpenNest.Mcp/`.