All NURBS and mesh bugs and their test models in one place

[ruevs]

This is a list of all SolveSpace issues related to NURBS or mesh bugs.
Links to the .slvs models reproducing the problems and the fixes may be useful for regression testing.
If the "Fixed on" column is empty the issue is probably still open (or I did not update this list :-)

Source	Model	Author	Fixed on	Fixed in
#32 #540 #683 #683 (comment) #683 (comment) #732	failed.lathe.slvs.zip adapter-plate.tar.gz wrong.zip wrong2.zip wrong3.zip coincident4.zip	@jockbert @baryluk @starsolaris @starsolaris @starsolaris @phkahler	Not really 2020-10-09	#737 194945f ab10e38 #786 8a3e5b4
#35 (comment) #35 (comment) #1410 (comment) #1452	NURBS_issues.zip NURBSTests.zip IntersectionReallyReallyReally...Should.zip bug-report-naked-edges-with-only-cuboids-extrude-and-unions.zip	@phkahler @ruevs @ruevs @kristianlm	Improved 2023-09-22	#1410 https://github.com/solvespace/solvespace/commit,/1963a836ef8e3c1f5411d2b301c86259cbdf6fa8
#171 #1124 #1268	bug.zip tube-turns.zip Geometry Error 1.zip	@whitequark @Symbian9 @blinkenlight	#652 (comment)
m #175	aaa-minus-weirdly-broken.slvs	@wpwrak
m #177	an-top.slvs	@wpwrak
m #259	case.zip	@jonas-schievink	2017-05-22
m #295	16 sided small circles			#626
#296 #693	table.zip polygon_failures.zip	@tgirod @phkahler	2020-09-08	#694 0413c1b d8f5a8d
#297 (comment)	bug-repro.zip	@rpavlik	Improved 2023-09-22	#1410 https://github.com/solvespace/solvespace/commit,/1963a836ef8e3c1f5411d2b301c86259cbdf6fa8
#303 #303 (comment)	tangent-circles-break-contour-fill.zip Tangent_Circle_Test.zip	@natronium @phkahler	2020-09-13	#697 e74185b
#315	clip.zip	@wavexx	2020-08-13	#673 04b332d
#317	Archive.zip	@AndyPhil	ca. 2020-08-20?
m #320 (comment)	320_death_star.zip	@phkahler
#408	screwdriver-sheath-with-taper.zip	@rpavlik	2020-06-12	#626 e83e483
#470	revolve-extrude-union-error.slvs	@Symbian9
#471 #472	no model no model	@phkahler @phkahler	2020-03-27	#473 f7b6f69
#473 (comment)	fillet.zip	@maz3max	Try?
#489 (comment) #550	model.zip helix.zip gear90.zip	@kalle @andres-erbsen	2020-07-01	#643 9802b5d
#497 (comment)	issue_497.zip	@phkahler
#537 #537 (comment) #537 (comment) #594 (comment)	jaw-.zip example2.zip Prisms_EarFail.zip 537_2D.zip CouldntFindAnEar_NURBSTests.zip	@rpavlik @ruevs @phkahler @ruevs	Relevant code
#577	wishes
m #632	texttest.zip	@ tjader0	mesh
m #647	no model	@phkahler	2020-07-12	#648 c75b2e4
#652 #652 (comment) #652 (comment)	spherecut.zip triangle.slvs.zip PrismConeNURBSNormalsTangents.zip	@xzcvczx @Christoph-D @ruevs	2020-10-12 2020-10-13 2020-10-16 2020-10-18	#734 5d29f65 #736 ruevs@d900578 ruevs@13c4563 #746 408128a
#652 (comment)	lathe-intersection.slvs.zip	@Christoph-D
#666 (comment)	A_wheel.zip	@phkahler	2020-08-03	#667 3c2f82b
#680 #439 (comment) #680 (comment)	second_revolve_test.slvs.zip path.zip	@zzj1213 @zzj1213	2020-10-15	#737 phkahler@194945f #786 8a3e5b4
#696	no model	@phkahler	2020-09-13	e74185b
#728	gusseted-truss.zip	@vcaputo
#486	A math improvement	@phkahler	2019-09-20	162897e
#623	A math improvement	@phkahler	2020-06-02	94a3cfd
#759	perf_test.zip Performance degradation	@phkahler	2020-10-23	caused by #737 ab10e38 fixed in #765 0f1ece2
m #1091 m #1091 (comment)	turc-nurbs-1.slvs.zip turc-nurbs-3.slvs.zip	@tomsci @tomsci
#1091 (comment) #1091 (comment) #1091 (comment) #1091 (comment) #1291 #1291 (comment) #1298 (comment) #1305 (comment) #1108	turc-nurbs-2.slvs.zip turc-nurbs-2_simplified.zip NURBS_ThreePrisms.zip corner-nurbs_slvs3x.zip cube_cut.zip CylinderTangentFromTheInside_AKA_Fillet_NURBSTests.zip FilletTests.zip Cluster_set.zip FT232BL_Case.zip	@tomsci @ruevs @ruevs @Symbian9 @phkahler @ruevs @ruevs @BKLronin @nanoant
#1058 #1058 (comment) #1059 #1279	model_and_stderr_output.zip DoubleWalledSphereZeroLengthEdgeInstersectionCrash.zip crash.zip keycap.slvs.zip	@kekscode @ruevs @nowakf @suhr	"Unexpected zero-length edge." crash
#1281 #1281 #1281	Archiv.zip ShovelStiel_Ultralighr.slvs.zip Archiv_morecutouts.zip	@BKLronin @BKLronin @BKLronin
#1208	slvs3x-cone_fillet.zip	@Symbian9
#1204 #1205	slvs3x-nurbs_cut.zip slvs3x-hex-hole_asm.zip	@Symbian9 @Symbian9
#1281	Many interesting models Archiv.zip ShovelStiel_Ultralighr.slvs.zip Archiv_morecutouts.zip	@BKLronin
#1390	wedge-diff-nurbs.slvs.gz	@vthriller

[maz3max]

My error persists :/ (Just tried the most recent master)

[ruevs]

It does :-)

[phkahler]

My error persists :/ (Just tried the most recent master)

@maz3max It looks like my old comment was wrong - that was not the same issue as 315. One thing I noticed just now is that your final Lathe operation creates a cylinder coincident with the existing hole. That will cause problems. I tried changing the sketch so it was just the 2 arcs and the short segements needed to cut the fillets but that didn't work either. Also tried making the segments so they are not coincident with the hole and plane and that didn't work either. After those changes I'm not sure why it still fails. That is a difficult problem. As a work around I tried leaving out the hole extrusion and making a similar cross-section to cut the entire hole at once. I will probably open a new issue including that and another sketch I have - both work for intersection but not difference.

I really hate to pollute this long-term tracking issue with this, but here's a super cool pro-tip on sketching inside another object:

After the original extrusion I dropped a datum point on the line highlighted in yellow. Then selected a pair of lines on the end of the extrusion and that point to create a new workplane. You can slide the point off the end of the line - it is the one at the bottom of the picture - and the workplane will follow. You can constrain against the entities from the other groups even though they are behind the workplane- it will constrain projected to workplane. You can even create the lathe group out there in space and then move that original datum point back to somewhere along the edge to move the hole into position. Go back to a previous group to constrain the position of the point along that edge and you're done. Like I said earlier, in this case it does not work - the hole works as intersection but not difference. Just wanted to point out the construction method.

Also, lastly. I had one more idea. Use that technique but don't create a cutting surface coincident with the top or bottom. This is how to work around the issue without going to triangle mesh:
437_work_around.zip
The hole can still need to be constrained in the first extrusion group.

[maz3max]

Thanks for looking into that! :)

[ghost]

Another case, inspired by "nucleus" OpenClipArt

• SLVS: nucleus_slvs.zip

NURBS operation says that all is ok, but real mesh is failed:

[giannissc]

Another one causing issues
dove-table.zip

[JiangXL]

Another NURBS Fail case: NURBS_Fail.zip. When two holes with same diameter, NURBS surface will fail.

[phkahler]

@JiangXL That is a common failure. There are some known work-arounds. Try creating the sketch plane for the second hole by using lines that are not orthogonal to the original box - make them at some arbitrary angle. Another option that may not be OK (depends on your design requirements) is to make one hole 0.001 larger or smaller than the other.

[JiangXL]

Thank for your kind reply! I decide to make one hole larger than other. I'm curious whether it is a NURBS limitation. If not, it will be so great that solvespace can solve it in the future.

[phkahler]

I'm curious whether it is a NURBS limitation. If not, it will be so great that solvespace can solve it in the future.

It is a problem in the NURBS code. A cylinder is composed of 4 surfaces of 90 degrees each. The problem seems to be failure to find the intersection of trim curves with the edge of those surfaces. I'd like to add code to split trim curves against trim curves which I suspect will fix this issue. Either that or figure out why the curve-surface intersection is failing on those edges and find a fix for that.

[ghost]

It is a problem in the NURBS code. A cylinder is composed of 4 surfaces of 90 degrees each.

Could be this value increased to 8 (45° stepping) for "Lathe" operations?

[phkahler]

@Symbian9 this issue is not meant to be a dumping ground for NURBS issues. @ruevs created an index to other issues.

[ghost]

For #738 (comment)

That is a common failure. There are some known work-arounds. Try creating the sketch plane for the second hole by using lines that are not orthogonal to the original box - make them at some arbitrary angle.

Here is a sample of such solution (not a bug case).

• SLVS: cube-orthogonal-holes_slvs.zip
  

[ConductiveInsulation]

I was told here to post and upload it here too. The "red hole" moved by a bit when i saved the file at another angle but its still there. Dont fully understand why it's there and how to remove it

hook.slvs.tar.gz

[phkahler]

@ConductiveInsulation Did you add the threads as a helix around an existing cylinder? If so, try having some (more) overlap so the threads cut into the cylinder an let the boolean union clean it up. That's just a suggestion to try, not a sure thing. I have a bolt model that does not use a boolean and has some red in the threads - that one is a a real mystery to me.

[ghost]

I have a bolt model that does not use a boolean and has some red in the threads

@phkahler, May I take a look on this SLVS?

[ghost]

Another one NURBS fail on intersection (not sure which issue is related here)

• NURBS intersected solid fails #1204

[phkahler]

@Symbian9 Please open new issues for new NURBS problems so they get unique issue numbers. We can add them to the table at the top of this tracker.

Also, the second version seems to work fine for me (on Windows: Version 3.x commit 5cb299b)

[ghost]

Please open new issues for new NURBS problems so they get unique issue numbers. We can add them to the table at the top of this tracker.

OK.

[MayeulC]

Here is my humble contribution to this list:

Screenshots and file

nurbs-fail.slvs.tar.gz