Skip to content

Arc walls of extrusions, outline sweeps and their shells mesh without needle triangles (OPE-101) - #34

Merged
aka-blackboots merged 2 commits into
mainfrom
fix/arc-wall-needle-triangles
Oct 31, 2026
Merged

aka-blackboots merged 2 commits into
mainfrom
fix/arc-wall-needle-triangles

Conversation

@aka-blackboots

@aka-blackboots aka-blackboots commented Oct 31, 2026 •

Copy link
Copy Markdown
Contributor

Fix needle triangles

…ithout needles

An extruded or one-segment swept profile with an arc edge gives zero-area
triangles on its curved wall, and a weld at 1e-8 m leaves edges used by
more than two triangles. These tests mesh the display buffers at the static
bucket and at the 2 mm drawing bucket and require no zero-area triangle, no
two corners closer than 1 micrometre and every welded edge used exactly
twice.

Signed-off-by: Vishwajeet Mane <manevishwajeet1@gmail.com>
Builders pad a face's uv box by the geometric tolerance. A padded box
fails the exact rectangle test, so the face goes to the trimmed path,
which laid its even grid over the padded box. Each grid line then sat
1e-8 to 5e-8 away from the edge sample it should meet, and the merge
kept both: one zero-width strip per arc sample on every extruded or
swept arc wall.

The even grid now spans the outer trim loop, taken before holes are
shifted, with its count scaled from the box so the density stays the
one the count was made for (a loose two-turn box keeps one turn's
count). The B-rep does not change.

Expected diff: 69 golden keys, all on faces with padded boxes (arc
extrusions, sweeps, shells, curved booleans): mesh and mesh-measured
volume only; triangles in the changed sections 191,050 to 163,346, none
more. Wall chord deviation is unchanged (D wall at deflection 0.73 mm:
3.649e-4 m before and after). The stored meshes arc-edged-extrusion
and arc-edged-extrusion-with-holes held 44 needle triangles each from
the old kernel and are re-recorded from this one (144 to 100 and 164
to 120 triangles); their B-rep files do not change.

Not in this change: an arc edge sampled more coarsely than the face
grid still leaves T-junctions on the wall rows; that has a separate
cause in how edge sample counts are chosen.

Signed-off-by: Vishwajeet Mane <manevishwajeet1@gmail.com>
@aka-blackboots
aka-blackboots merged commit c3a9eea into main Oct 31, 2026
0 of 8 checks passed
@aka-blackboots
aka-blackboots deleted the fix/arc-wall-needle-triangles branch October 31, 2026 13:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant