Fix arc length cost

[adam-urbanczyk]

No description provided.

[codecov]

Codecov Report

Merging #962 (c7d55b6) into master (4fcada6) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master     #962   +/-   ##
=======================================
  Coverage   96.08%   96.08%           
=======================================
  Files          40       40           
  Lines        9257     9257           
  Branches     1112     1112           
=======================================
  Hits         8895     8895           
  Misses        209      209           
  Partials      153      153           

Impacted Files	Coverage Δ
cadquery/occ_impl/sketch_solver.py	91.14% <100.00%> (ø)
tests/test_sketch.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4fcada6...c7d55b6. Read the comment docs.

[jmwright]

@lorenzncode I requested your review of this PR since you have been closely involved. You can decline if you would rather not.

[jmwright]

Thanks @adam-urbanczyk

[adam-urbanczyk]

Yes, please do have a look @lorenzncode . The other issue/PR might be more complicated.

[lorenzncode]

I checked out the branch and ran a few manual tests. I did not find any problem with arc length cost.

There may be an issue related to bounds.

s = (
    cq.Sketch()
    .segment((0.3, 0), (0, 0), "seg1")
    .arc((0.0, 0.0), (-0.25, 0.5), (0.0, 1.0), "arc1")
    .constrain("seg1", "Fixed", None)
    .constrain("seg1", "arc1", "Coincident", None)
    .constrain("arc1", "Length", 60)
    .solve()
)

The above which omits Radius and ArcAngle constraints, and with relatively large Length constraint value returns an "entity" with arc angle > 2*pi.

	[ 5.01675398  6.6817405   8.35544592 -2.49756986  7.18094529] (CIRCLE)

The resulting Edge length is ~45 when expected value is 60.

I tried setting an upper bound of 2*pi to arc angle (where currently only the lower bound for radius is set). The result was an Edge with length 60 and entity:

	[16.70179153 19.90442625 25.9833798  -2.44346111  2.30916842] (CIRCLE)

I think the bounds issue can be handled separately from this PR. Looking forward to see how the sketch solver progresses!

[adam-urbanczyk]

@lorenzncode I tried adding ub and lb but I get failures for some test cases. I think this will be a longer story (different solver, different parametrization, ...). In the meantime I'd propose to merge this to fix the obvious bug. @jmwright if you agree, feel free to go ahead with merging.