RichardPotthoff has quit [Ping timeout: 240 seconds]
RichardPotthoff has joined #openscad
fling has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
fling has joined #openscad
<dTal>
wait, "difference()" works as invert when called with 1 argument? That's hugely useful and undocumented?
<dTal>
I just accidentally ran a homegrown minkowski fillet/round module without defining the usual "big" and "bigger" giant squares/cubes to do the invert, and while the console complained about the undefined calls, the routine *still worked*
<dTal>
...okay, a simple test case in a new window doesn't work.
<dTal>
on the latest nightly, you get spurious circles from the minkowski round, presumably caused by some leftover fragment of geometry due to precision loss. They move around, or vanish, if you play with the value of round2d() or $fn.
<dTal>
It doesn't do it with OpenSCAD-2020.11.03.ai6054-59abb96-x86_64.AppImage
<InPhase>
I get warnings about missing elements and no output. Not sure what you're talking about seeing or expecting to happen.
<InPhase>
There's no logically correct output for geometry attempted on things that don't exist.
<dTal>
forget the first thing I was talking about
<dTal>
I figured that out - the bpaste is a completely separate issue
<dTal>
sorry for the confusion
<InPhase>
I think there was a behavioral change for minkowski with empty objects.
<InPhase>
I remember discussions with... was it peepsalot? struggling over what the right thing was to do in this scenario.
<InPhase>
I think that's what you're hitting here.
<dTal>
But I'm not doing a minkowski with an empty object.
<InPhase>
Perhaps this shape should end up in the testing suite as well.
<InPhase>
Clearly we don't have any 2D shapes producing internal vertices being massaged in this way, or the issue would not have gotten merged without fixing.
<InPhase>
dTal: I award you 5 bonus points for good attention to detail. :)
<InPhase>
Hmm. I wonder if it's a side-effect of the refactoring work that permitted fast-csg. This feels like some sort of floating point artifact. The spots appear to be on the scale of approximately 0.001 in thickness.
<J22>
i think it is because minkowski is not rotation a circle around the edges .. so the problem is not seen with offset.. - so with minkowski there are "unhappy" constellations of edges ( like you can see with roof)
<InPhase>
Hmm, yes, it's definitely dependent upon the $fn of the 0.1 circle.
<J22>
i think the sqrt(2) also play a role .. (just crashed scad)
<J22>
nah the sqrt is not
<InPhase>
Hmph. It ONLY seems to produce the artifact when the 0.1 circle is $fn=128.
<InPhase>
I can't hit it for anything else?
<InPhase>
Okay, but if I change the top level $fn to something else, the artifacts still exist but are just different.
<InPhase>
It depends on a matching $fn between the two circles.
<J22>
i am pretty sure it is like with roof .. i only get that issue if both circles use the same $fn
<J22>
you can solve this by rotation one circle rotate(1)
<J22>
when two tri faces meet with an edge that is not reaching the perimeter of the geometry
<dTal>
Yeah I'm also seeing geometry errors of a magnitude of about 0.0006
<dTal>
I've calculated the center point of the round precisely, but if I put a circle of r=0.1 exactly there, I can see it doesn't quite match up - it matches the fillet if I translate it up by 0.0006
<dTal>
look for yourself: translate([sqrt(2)/2+0.1,0.39821934127147884]) circle(0.1)
<dTal>
s/up/down
<dTal>
InPhase: I have to have a GitHub and all that stuff to create an issue, right?
<InPhase>
A basic account, sure.
<InPhase>
You just register, hit New Issue, and it's basically like posting on a forum or on reddit.
<InPhase>
It's helpful for the person who notices a thing to post it, because sometimes more details are needed.
<dTal>
I didn't need attention to detail - those tiny slivers turned into huge circles with the final step of my minkowski rounding :)
<dTal>
interesting that offset() doesn't do it
GNUmoon has quit [Ping timeout: 240 seconds]
myosotis has quit [Quit: myosotis]
fling has quit [Ping timeout: 240 seconds]
fling has joined #openscad
lastrodamo has quit [Quit: Leaving]
GNUmoon has joined #openscad
poisond has quit [Ping timeout: 256 seconds]
poisond has joined #openscad
GNUmoon has quit [Remote host closed the connection]