A facing-job generator for the Makera Z1. Pick X, Y, depth and material; get
a .nc the controller will preview, boundary-trace and run. Also checks any
.nc (Makera Studio, EasyTrace5000, FlatCAM, hand-edited) against what this
machine has taught.
Live at https://facing.nottseter.no.
bun install
bun run dev # http://localhost:3117
bun test
bun run typecheck-
Facing jobs: general mode (one pass per depth level) or fine finish (rough to an allowance, then a final pass at 90° with about half the stepover). Four patterns: serpentine X, serpentine Y (default), one-way Y, spiral. Serpentine Y is the default because Y is more rigid than X on this machine (tested), so cutting along Y gives the flatter surface.
-
Header and preview: Makera's
;@MKR|header, a toolpath thumbnail, and an MDI corner walk (the "reach" check) so the laser can be used to confirm the stock position before cutting. -
Bits: one bit, the 3.175 × 12mm single flute (Metal series), with Makera's published speeds and feeds. Chamfer and fly-cutter options exist; the checker also knows Makera's table for other bits.
-
Check a file (
/#check): reports Fail / Silent / Warning / Note findings for an uploaded.ncand draws its toolpath. Nothing is stored. It checks that a file is valid Z1 code within the parameters this project has tested, which were mainly PCB milling. "No findings" is not a promise that a job is safe, and a Warning means "differs from what is known to work", not "wrong".A few of its rules (variables and expressions,
G92,G10,G38.x,M498) come from firmware notes in blackveilprecision/z1-macros and have not been tested on this machine; each finding says so.
The machine knowledge lives in src/facing.ts, src/mkr.ts, src/materials.ts
and src/check.ts, with each rule commenting where it was learned.
docker run --rm -p 3117:3117 ghcr.io/nilsan/cnc-facing:latest # http://localhost:3117
docker build -t cnc-facing . # or build locallyThe image has no volumes and stores nothing. Set options with -e, e.g.
-e PORT=8080 (then publish that port too).
Settings are environment variables, read at startup in src/server.ts. Locally,
set them inline (PORT=8080 bun run dev). In Docker, pass -e VAR=value to
docker run, or put them under environment: in the service in
deploy/docker-compose.yml. The image's defaults are set with ENV in the
Dockerfile.
| Variable | Default | |
|---|---|---|
PORT |
3117 |
|
HOST |
0.0.0.0 |
127.0.0.1 for localhost only |
CNC_FACING_ALLOW |
127.0.0.0/8,::1/128,10.0.0.0/8,172.16.0.0/12,192.168.0.0/16 |
comma-separated CIDRs that may connect (loopback and private LAN ranges); 0.0.0.0/0,::/0 turns the filter off |
The image overrides the HOST/CNC_FACING_ALLOW defaults above: it listens on
0.0.0.0 and has the filter off. The allow-list is a blunt "do not answer strangers" filter, not authentication.
The filter is off in the image because behind the reverse proxy every request
comes from the proxy; to use it without a proxy, set CNC_FACING_ALLOW yourself.
Requests are rate-limited per client (a generous bucket for the form, a small one
for /api/check, which parses whole files) and request bodies are size-capped.
Behind a proxy the client is taken from X-Forwarded-For. CNC_FACING_RATE_LIMIT=off
disables the limits, e.g. for load tests. The screenshots in docs/ are made by
scripts/screenshots.py.
Pushes to main run .github/workflows/build.yml (typecheck, tests, then push
ghcr.io/nilsan/cnc-facing). The image runs as the cnc-facing stack in Komodo
on bf.nottseter.no, using deploy/docker-compose.yml, behind Caddy.