0 Fork 0

mirror of https://github.com/gittrahan/support-fins.git synced 2026-10-23 04:12:33 -05:00

Files

T

matthewtrahan 5cc1da01ac plugins: entry takes material, padStyle and cutout -- plugins can print PETG, the site's pad styles and wall cutouts

The entry sends the full clearance set on every call, as the site's worker does:
buildFins keeps it as module state and a plugin's V8 context lives across calls,
so a partial set would leak PETG into the next PLA run. Unknown values throw
instead of silently building PLA. Sway braces take the material's gap and bite.

2026-09-30 11:19:03 -05:00

engine

plugins: entry takes material, padStyle and cutout -- plugins can print PETG, the site's pad styles and wall cutouts

2026-09-30 11:19:03 -05:00

plugins: review nits -- CI step name, slicer docstring, Orca build size

2026-09-30 10:10:58 -05:00

tests

plugins: entry takes material, padStyle and cutout -- plugins can print PETG, the site's pad styles and wall cutouts

2026-09-30 11:19:03 -05:00

bundle.py

plugins: move the engine bridge and bundle step into plugins/shared (step 1 of one engine for every plugin)

2026-09-25 16:18:40 -05:00

ENGINE-SENSITIVITY.md

plugins: move the engine bridge and bundle step into plugins/shared (step 1 of one engine for every plugin)

2026-09-25 16:18:40 -05:00

README.md

plugins: document the shared slicer and what goes in shared vs an adapter

2026-09-30 10:10:58 -05:00

README.md

Support Fins — shared plugin code

Code every plugin that can run JavaScript shares, so they all run the website's engine (web/*.js), unmodified, instead of keeping their own copy of the geometry. A fix on the site reaches a plugin the next time it is built.

engine/fins_entry.js    posed triangle soup in (mm, z up) -> fin + bed-pad triangles out
engine/bridge.js        base64 in/out for Python hosts running the bundle in V8 (mini-racer)
bundle.py               esbuild: bridge.js + web/*.js -> one IIFE, global SupportFinsEngine
py/supportfins_host.py  Python side for plugins that run the bundle in mini-racer: start V8
                        (host_engine), run the engine on a soup and map the fins back to
                        the caller's frame (host_compute). Orca inlines it at build time
py/supportfins_slice.py mesh -> per-layer polygons (split_shells, slice_soup, group_loops),
                        for hosts whose API takes layer polygons, not a mesh. Pure numpy.
                        Orca inlines it at build time
tests/                  Deno tests: same fins as the website, anywhere on the plate;
                        the base64 bridge round-trips exactly
ENGINE-SENSITIVITY.md   engine note: tine placement moves under 1e-13 mm of noise, and
                        how fins_entry.js neutralises it

python3 plugins/shared/bundle.py out.js     # needs esbuild (npx fetches it on demand)
deno test --allow-read plugins/shared/tests/
python3 -m pytest -q plugins/shared/py/tests/   # needs numpy, trimesh, mini-racer==0.14.1

What goes where. Anything two plugins could use lives here: engine host, mesh slicing, and next the options every dialog is built from. plugins/<host>/ only calls into that host's API (read the part, show a dialog, add the result). A plugin that needs a new piece of geometry adds it here, with its own tests, instead of in its adapter. Fusion is the exception: its fins_core/ is pure Python because Fusion has no numpy, and moves here only when another plugin wants per-fin objects.

Used by: Orca (inlines the bundle into its single-file plugin) and Cura (ships the bundle and the Python host next to the plugin). Onshape (FeatureScript) and Prusa (Lua) can't run JavaScript, so they don't use this.

CI: .github/workflows/plugins.yml rebuilds and tests the plugins on every PR and push that touches web/ or plugins/, and on main publishes the builds to the plugins-latest pre-release.