Uncaught TypeError in reTesselateCoplanarPolygons

[kirevdokimov]

I use .union() to merge geometries, but when I do so, it causes errors.

Expected Behavior

Geometries are merged with no issues

Actual Behavior

.union() throws an exception

Uncaught (in promise) TypeError: Cannot read properties of undefined (reading 'length')
    at reTesselateCoplanarPolygons (reTesselateCoplanarPolygons.js:176:40)
    at retessellate.js:25:44
    at Array.forEach (<anonymous>)
    at retessellate (retessellate.js:23:14)
    at union8 (unionGeom3.js:21:17)
    at union8 (union.js:42:35)

Steps to Reproduce the Problem

I assume this basic example works for everyone

const c1 = cylinder({ radius: 3, height: 12, center: [0, 0, 5], segments: 12 })
const c2 = cylinder({ radius: 3, height: 18, center: [0, 0, 5], segments: 12 })

const u = union(c1, c2)

Specifications

• Version: "@jscad/core": "^2.6.9", "@jscad/modeling": "^2.12.2",
• Platform: Windows 11, Brave Browser,
• Environment: Node v16.15.1, esbuild, Brave Browser Version 1.69.153 Chromium: 128.0.6613.85 (Official Build) (64-bit)

Same issue for Google Chrome Version 127.0.6533.122 (Official Build) (64-bit)

[kirevdokimov]

polygonindexKey is equal to "clear" or "remove" while we iterate over startingpolygonindexes, what causes vertices2d to be undefined.

If we check for Nan with if (isNaN(polygonindexKey)) continue line, other issue occurs:

Uncaught (in promise) TypeError: Cannot read properties of undefined (reading '1')
    at interpolateBetween2DPointsForY (interpolateBetween2DPointsForY.js:11:16)
    at reTesselateCoplanarPolygons (reTesselateCoplanarPolygons.js:226:15)
    at retessellate.js:25:44
    at Array.forEach (<anonymous>)
    at retessellate (retessellate.js:23:14)
    at union8 (unionGeom3.js:21:17)
    at union8 (union.js:42:35)

[kirevdokimov]

This is because activepolygonKey is equal to "clear" or "remove" sometimes as well.

Checking for NaN here with if (isNaN(activepolygonKey)) continue line solves the issue

[z3dev]

@kirevdokimov thanks for the excellent example, and the in-depth analysis.

there are lots of tests but not for this specific example. Given the similarity of the two cylinders, there's obviously something about the retellsation. Have you taken a look at the results of the Union before the retessellation? (Sorry for the spelling)

I'll look over the fix.

[kirevdokimov]

Responded in a related merge request: #1361

[platypii]

I was not able to reproduce this issue with the example you gave. Maybe I'm missing something?