pca006132 has quit [Remote host closed the connection]
pca006132 has joined #openscad
kintel has joined #openscad
<kintel>
JordanBrown Just leaving some thoughts for future history browsing: Specification on how to convert doubles to 32-bit integers before performing bitwise operations: https://tc39.es/ecma262/#sec-toint32
<kintel>
If you can prove that your number fits in 32 bits, a cast might be the fastest, but trunc + modulo might need some care
<pca006132>
I don't think we are sufficiently fast for this optimization to make a difference :)
<kintel>
hehe, fair, but we still needs to be correct, which we're not today
<kintel>
Thanks to good test cases, that got caught early :)
pca006133 has joined #openscad
<pca006133>
indeed, this is a correctness issue
<kintel>
pca006132 While you're here: How robust do you think Manifold's Project() function is?
<kintel>
I remember early clipper1 versions struggled with that, not sure about early clipper2 - I don't quite trust that library yet as it broke quite early when I started porting OpenSCAD to clipper2
pca006132 has quit [Ping timeout: 268 seconds]
pca006133 is now known as pca006132
<pca006132>
hmmm
<pca006132>
it is not really tested much
<pca006132>
I think we rely on clipper to sort things out correctly
<kintel>
yeah, that scares me a bit ;)
<kintel>
Anyway, I'll try and see what happens
<pca006132>
I think it will be great to find out any issue and report to clipper2
<pca006132>
development there seems very active
<kintel>
We don't really have any projection stress tests either, as we haven't received many bug reports in the past..
<kintel>
pca006132 Do you know if you do any optimizations in the Projection case? e.g. if you know you're going to project, you don't have to perform a union. ..but performing the union may or may not help I guess, in terms of reducing the # of vertices..
<pca006132>
currently we don't
<pca006132>
but we can
<pca006132>
there are also possibilities of parallelization
<kintel>
I'll run some tests and ask less theoretical questions :)
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
L29Ah has left #openscad [#openscad]
<pca006132>
theoretical questions are fine, I am a pretty theoretical guy (not with manifold though) :)
<pca006132>
but for performance optimization we typically want to start from cases where it is slow
<pca006132>
just a lot easier that way, especially since we have many tests for free
L29Ah has joined #openscad
mmu_man has quit [Ping timeout: 252 seconds]
califax has quit [Remote host closed the connection]
califax has joined #openscad
califax has quit [Remote host closed the connection]
califax has joined #openscad
mmu_man has joined #openscad
kintel has joined #openscad
<kintel>
pca006132 Somewhat surprisingly, Manifold's Project() was faster than the existing OpenSCAD implementation. Could be that the 32-bit Clipper2 is a bit faster than the 64-bit Clipper1..
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
teepee_ has joined #openscad
teepee has quit [Ping timeout: 260 seconds]
teepee_ is now known as teepee
mmu_man has quit [Ping timeout: 255 seconds]
ali1234 has quit [Remote host closed the connection]