<GitHub>
[solvespace] Evil-Spirit commented on issue #226: @whitequark, Somehow too many redundant constraints can't be converged. I've figured out what 3xPT_ON_CIRCLE can't converge (just try to make it on empty sketch). After disallowing redundant, too many redundants threated as non-converged for no reason. This is probably bug, but I am not sure whether rank test can work fine or not for such amount of redundants. https://github.com/solvespace/solve
<GitHub>
[solvespace] Evil-Spirit commented on issue #225: @whitequark, Somehow too many redundant constraints can't be converged. I've figured out what 3xPT_ON_CIRCLE can't converge (just try to make it on empty sketch). After disallowing redundant, too many redundants threated as non-converged for no reason. This is probably bug, but I am not sure whether rank test can work fine or not for such amount of redundants. https://github.com/solvespace/solve
<lexszero>
hi. i'm trying to design a dodecahedron (or some other regular polyhedron) built from rectangular beams in solvespace, and it's a huge pain
<lexszero>
I've failed even to draw a simple dodecahedron of line segments. I started with drawing a pentagon in plane, then drew one side facet in a inclined plane, tried to step rotate it, and now I can't constraint it so that row of facets stays in place.
<GitHub>
[solvespace] whitequark commented on issue #229: I don't think this is a bug. We allow all kinds of degenerate constraints, like line segments with length 0, and so on. It seems odd to prohibit those in some cases (like when specified explicitly) but not others (when they're inferred from other constraints). https://github.com/solvespace/solvespace/issues/229#issuecomment-292179306
<popsch>
basically I want to first design the 3D object, then slice it into surfaces and cut them. to be able to build the structure, I need to add joints
<wpwrak>
popsch: you could make an everything.slvs, then "import" it as assembly in part-a.slvs and part-b.slvs, and in each remove what you don't want there.
<popsch>
wpwrak, where can I find everything.slvs?
<popsch>
or you mean first make individual parts before importing :)
<wpwrak>
no, the other way around: make the master geometry in everything.slvs, then cut off what you don't want
<wpwrak>
but it may be easier to start with part-a, then use it in part-b for reference, etc.
<popsch>
I see. the question is whether there's a way to quickly create repeated shapes in a parametric form? openscad allows me to program a zigzag lined shape and use it as a difference on another shape. This could quickly make a joint as shown in the picture. Is there a way in solvespace to specify a function that spits out a shape (or some other way to generate a shape)?
<popsch>
for example, just making a gear is basically a circle with a number of cutouts around it
<wpwrak>
you can repeat a group (translate/rotate), but there's not generalization of this, i.e., no x = f1(t); y = f2(t)
<popsch>
ok
<popsch>
again, I'll play with it and learn
<wpwrak>
you could make a gear that way. but not a thread
<wpwrak>
yeah, the scientific approach - idea, then experiment ;-) byw, the reference manual is very useful: http://solvespace.com/ref.pl
<wpwrak>
especially if you've already figured out bits on your own, and can then read how it's really supposed to work
<lexszero>
popsch: solvespace is hardly useful nor productive for most of real-world mechanical engineering tasks. but you can try.
<popsch>
I'll experiment with it and then see. I like the solver part, which allows me to dynamically resize parts
<lexszero>
yes, that's a cool feature. i'm starving for something combining openscad programmatic approach and solvespace constraint-based approach, really.
<Jonimus>
whitequark: sorry I wasn't near the machine I was building on and since i got it to build with NMake I didn't look to far into it. I can get you exact errors tonight.
<popsch>
it would be nice to have a library online to download and import specific shapes that have already been defined.