MVT icon client preview — what was built, and what is withdrawn

kerfmaster pin list · Done

recordStates what the code does today. Auditable against the code — this is the only type that can go stale.

Content last changed 2026-08-06 — computed from the item itself, not typed.

Contract — Report stage

This contract describes the code as it runs today.

takes healed-geometrymaterial-table

makes delivery

fails if A published figure that outlived the geometry it was measured on.

Contract last changed 2026-08-06 — computed, not typed. Dated separately from the text above, so neither date can speak for the other.

1 open question blocks this pin

Q14 Does a strip roughly 1.6 by 0.20 inches, cut on both sides in 1/8 steel, stay flat or bow?
Jordan — A customer sheet claim about whether a joined option holds its shape.

The first end-to-end run of art → measured options → two customer-facing pages. Written for the MVT icon; the pieces are general, the numbers are not.

What the code does

svg_io.py — reads paths out of an Illustrator SVG.
mvt_prep.py — turns the client art into cuttable geometry: the slot along the ring's outer circle, the ligaments that break it, and the bridges.
build_mvt_preview.py — measures three options, writes data.json, and renders the customer page and the shop page from it (one dataset, two renderers), with the revision archive of job-reference. Measurement helpers: narrowest() (erosion scan for the thinnest metal anywhere), inradius(), neck_at(), Zsec() (section modulus, for what load bends a bridge).

The three options: A one piece, ring held to the cloud by the ligaments and the triangle by bridges; B the triangle lifted onto spacers; C three separate plates on spacers, no bridges, no ligaments, no slot.

The slot-break rule

Where two filled shapes overlap, the one on top is shown by cutting a slot along the buried edge. The rule the code implements:

The slot exists only where the edge is buried in the other shape's metal — mvt_prep.buried_runs() walks the circle and keeps the angular runs that lie inside the cloud.
It is broken at the run ends that face an interior void, never at the ends where the edge emerges into open air, because the emergence points are where “this shape is on top” is actually read (interior_ends(), trim_run(); setback LIG = 3.0 pt, slot width SLOT_W = 0.75 pt).
The ligament width is not a free parameter — it is set by how the covering shape's edge crosses the circle, so it cannot be widened by moving the slot.

A measurement that was corrected before it reached the customer

The hand-off note claimed the cloud boundary and the ring's outer circle ran 0.0038 in apart over ~3.5 in. Re-measured at 30 in scale it does not hold — it conflated the single closest approach with a much looser run. Measured: gap under 0.0040 in for 0.05 in of arc; under 0.0625 in for 0.79 in; under 0.1250 in (one material thickness) for 1.51 in; under 0.25 in for 2.99 in. Closest approach 0.00020 in at 144.8°. The customer copy states the 0.125 figure, because that is the one that matters for cutting. Kept here because the difference between “closest approach” and “how much of it runs that close” is a whole class of wrong number, and this one was caught on the way to a client document rather than after.

⚑ Status: the machinery stands, the figures do not

Every published figure is withdrawn — revision Q0QB. Defects D15 and D16 meant the cloud was cut short by its whole top-left lobe and again by the ring's inner disc, so option weights, bend loads and the risk ranking were all computed from geometry that was missing metal. Option C's cloud was stated at 8.84 lb and measures 11.82 lb.
The slot and ligament figures are unverified, not wrong-by-proof. The buried runs (57–87° and 115–144°) and the ~0.21 in ligament cap were computed from the same truncated cloud. The rule above does not depend on the missing lobe; those particular numbers do, and are not re-measured until the rebuild.
“Moving the ring buys nothing structural” is superseded. The ligament measurement behind it was sound and the conclusion drawn from it was out of scope — see evidence discipline.

The reference geometry is now Aristide's modified file

The client's original SVG is history. incoming/1785975510_3070743_MVT-ICONtest_Modified.ai is the reference, measured in cadmaster/MVT_MODIFIED_MEASURED.md (e6ee4d4): at 30 in wide in 1/8 mild steel the piece is 30.00 × 22.29 in, 15.30 lb — cloud 333.39 in² / 11.819 lb, ring band 68.68 in² / 2.435 lb, triangle 29.57 in² / 1.048 lb, scale 0.168729 in/pt. His cloud is one closed outline in one piece; narrowest metal anywhere in it 1.029 in; the triangle is 0.850 in clear of the ring's inner edge, so it is still a separate island.

src — built 2026-08-05 (5fa69c8, 9d7486f, dd3bf55); reference geometry measured 2026-08-06 (e6ee4d4); withdrawal and defects recorded 2026-08-06. Written into the record 2026-08-06 settling that debt.