hyperair has quit [Read error: Connection reset by peer]
ferdna has quit [Quit: Leaving]
cart_ has joined #openscad
tcurdt has quit [Server closed connection]
tcurdt has joined #openscad
hyperair has joined #openscad
pkriens has joined #openscad
pkriens has quit [Client Quit]
Junxter has joined #openscad
guerd871 has quit [Ping timeout: 268 seconds]
mmu_man has joined #openscad
guso78k has joined #openscad
<guso78k>
teepee you mentioned "Library Manager" for OpenSCAD previously in the IRC. How far this idea already evolve ? is there a place where ideas are written down ?
<guso78k>
yes , "Library Manager" exactly matches the idea, i was focusing.
<guso78k>
I'd like to see a big list database of GITHUB repositories with an URL and a Name each, where ANYBODY can add entries and browse the ENTRIES from within openSCAD
<teepee>
that is pretty much this. it's essentially the flow used by arduino too on the backend side
<guso78k>
yes, i have been using Arduino, too
<guso78k>
"My" Idea would limit the the usage to only XX-Scad Designs and also to Github. Github because it offers great versioning control
<guso78k>
it should be usable without the user knowing about git as a tool.
Junxter has quit [Read error: Connection reset by peer]
Junxter has joined #openscad
teepee_ has joined #openscad
teepee has quit [Ping timeout: 260 seconds]
teepee_ is now known as teepee
guerd87 has quit [Read error: Connection reset by peer]
guerd87 has joined #openscad
guerd87 has quit [Read error: Connection reset by peer]
guerd87 has joined #openscad
splud has quit [Server closed connection]
splud has joined #openscad
mtm has quit [Ping timeout: 256 seconds]
mtm has joined #openscad
J24k79 has quit [Quit: Client closed]
J24k79 has joined #openscad
guso78k24 has joined #openscad
redlizard has quit [Server closed connection]
redlizard has joined #openscad
kintel has joined #openscad
<kintel>
guso78 JS as a language: Are you building that on the same infrastructure as Python? I remember there was some discussion/ideas about making the underlying infrastructure general enough to easily add languages on top.. Which JS engine did you start integrating?
<kintel>
(I think JS will be significantly easier to integrate as there is proper sandboxing support)
<kintel>
guso78 Non-triangulated data: You'd have to go all the way back to the user-defined primitives, as we have to triangulate early, e.g. because Manifold only supports triangles.
<kintel>
We could add face attributes mapping back to original non-triangular faces, but that will get pretty messy I fear
<guso78k24>
kintel, i started to integrate js next to my current python implementation and js directory next to the python directory
<guso78k24>
reason for that is, that i see people have difficulties with the indents of python
<kintel>
sounds good. In any case, it's a good opportunity to refactor the commonalities into some general infrastructure to make any duplication between the two minimal :)
<guso78k>
but yet i know already, that it will be hard to achieve same level as i did in python. look at the curly braces workaround, as javascript does not support named parameters as scad language does.
<kintel>
yeah, positional parameters are also hard to remember : /
<guso78k>
kintel, i have written procedures to undo the triangulation by combining faces with exact same orientation, but i am resistant to call my function once after render and before doing measurement as it takes some cpu time ...
<guso78k>
still have other languages like lua in mind, but i did not yet find the 2nd best match(python is my #1)
<kintel>
I wonder if we could convince Manifold to support allowing multiple triangles to have the same face ID - that could solve your problem
<guso78k>
it would ease my problem. i would only allow edges which have faces with different FaceID's on either side, but still i have to build an Edge-Database ...
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<guso78k>
as i dont have the perfect solution yet, i accept the fact, that measurement can also select inter-polygon edges.
<guso78k>
I think we will anyway need to do some pre-measurement processing in future, this will enable us to do fancy face selection, as seen in prusa slicer.
<guso78k>
latest measure PR also adds a "ruler" as red line to show the distance, where the measurement was taken, however ideally we copy the code to draw x/y/z axes and use that one create the "ruler"
guso78k24 has quit [Ping timeout: 250 seconds]
guso78k has quit [Ping timeout: 250 seconds]
kintel has joined #openscad
<kintel>
..or ideally refactor the axes code to modern OpenGL, to avoid building more technical debt :)
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<pca006132>
run the benchmark and see the results... that can help us tune the algorithm performance a bit
<pca006132>
windows is not supported as the benchmark code I copied uses cxxabi for demangling and print out the function names, which is not supported by msvc (mingw/msys2 may work?)
<t4nk_fn>
because it was complaining about quickhull
<pca006132>
with MANIFOLD_PAR=TBB
<t4nk_fn>
no sources given
<pca006132>
did you do recursive clone? we have submodules :P
<t4nk_fn>
lol
<t4nk_fn>
of course not, how dare you ;)
<pca006132>
we used to have many more submodules
<pca006132>
and quickhull will become optional for manifold build later, so users will not notice they forgot to do recursive clone if they did not enable building that
<t4nk_fn>
no hidden telemetry, right?
<t4nk_fn>
or open, for that matter ;)
<pca006132>
no, it is too much work to hide that into our code
<t4nk_fn>
lol
<t4nk_fn>
euh, buddy, I just built manifold from a build dir I created, with you suggested TBB thingy
<pca006132>
interesting, AMD behaves very differently from Intel
<t4nk_fn>
I'm guessing you are all on way better hardware
<pca006132>
but not users have the same hardware as ours, so it is important to get more data from the users
<pca006132>
when doing this kind of performance tuning
<t4nk_fn>
sure
<pca006132>
btw what gcc version are you using?
<t4nk_fn>
13.2.1 20240210 and I think my mem is running at 2100MHz
<pca006132>
the stable_sort in your system seems suboptimal... I am not expecting my simple parallel version to beat the stl sort for small vectors (on other machines it is indeed slower than the stl version)
<pca006132>
if you look at the results we had on the github issue, you will notice that sort_* for 1K array size is usually slower, because the merge sort I am using is not well optimized for single core
<pca006132>
but in your system it seems to be faster than the STL sort
<pca006132>
no idea why
mmu_man has quit [Ping timeout: 252 seconds]
kow__ has joined #openscad
mmu_man has joined #openscad
drkow has quit [Ping timeout: 272 seconds]
Guest14 has joined #openscad
cart_ has quit [Ping timeout: 264 seconds]
<t4nk_fn>
just let me know if there's anything else you'd like and I'll see what I can do if I see it.
<pca006132>
Maybe compare the current master branch performance (elalish's branch) with the performance after my PR. Just time both tests/manifold_test and extras/perfTest. CPU time and wall clock time are both important to know
<pca006132>
ideally after these tuning we should reduce unnecessary parallelization, so the CPU time may be slightly a bit better
<t4nk_fn>
so I just applied your pr as a patch on top of the current master branch
guso78_ has joined #openscad
guso78 has quit [Ping timeout: 240 seconds]
lf94 has quit [Ping timeout: 252 seconds]
lf94 has joined #openscad
guso78k has joined #openscad
drkow has joined #openscad
kow__ has joined #openscad
drfff has quit [Ping timeout: 240 seconds]
drkow has quit [Ping timeout: 272 seconds]
jts_ has joined #openscad
jts has quit [Ping timeout: 260 seconds]
guso78k95 has joined #openscad
guso78k has quit [Quit: Client closed]
mmu_man has quit [Ping timeout: 246 seconds]
Guest38 has joined #openscad
Guest38 has quit [Client Quit]
mmu_man has joined #openscad
guso78k95 has quit [Ping timeout: 250 seconds]
drkow has joined #openscad
kow__ has quit [Ping timeout: 246 seconds]
mmu_man has quit [Ping timeout: 252 seconds]
mmu_man has joined #openscad
J24k79 has quit [Quit: Client closed]
J24k79 has joined #openscad
hyvoid has quit [Server closed connection]
hyvoid has joined #openscad
Guest9 has joined #openscad
<Guest9>
Hi all. I'm new to 3D modeling and OpenSCAD but I am finding it really useful the more I play with it. One thing - I use a Mac, and the Github binary isn't signed; is anyone working on getting the signing process working? I am a long time Mac developer and could help with that.
<Guest9>
Happy to file a bug in Github if that's the best way to keep track of the issue.
<teepee>
there's a long discussion on that already. if you have an idea how to get this woking that would be nice
<Guest9>
It looks like there are some nice tools now for signing and notarizing builds with Github workflows. The trick is to have an Apple developer account - if OpenSCAD.org doesn't have one already then that's probably the best way to do it, rather than an individual developer account.
<Guest9>
(I know someone in Developer Relations who could help with that.)
<teepee>
i can reactivate our company account, I assume that would work
<Guest9>
Sure, if you are OK with the company being the "distributor" of the binaries
<teepee>
the builds are on CircleCI though, Github runs the test framework, but is not used for building binaries for distribution
<Guest9>
I can take a look at this and comment in the bug. I'll take a look at the bundle ID, entitlements, etc. and then see how automatic signing could work
<Guest9>
I'm sure there's a way; I'll look at how it builds on CircleCI - haven't used that before
<Guest9>
Ah thanks. The trick is you need some way to store secrets for the developer certificates; CircleCI must have a way to do that for other things that need authentication.