teepee changed the topic of #openscad to: OpenSCAD - The Programmers Solid 3D CAD Modeller | This channel is logged! | Website: http://www.openscad.org/ | FAQ: https://goo.gl/pcT7y3 | Request features / report bugs: https://goo.gl/lj0JRI | Tutorial: https://bit.ly/37P6z0B | Books: https://bit.ly/3xlLcQq | FOSDEM 2020: https://bit.ly/35xZGy6 | Logs: https://bit.ly/32MfbH5
<J23K40> if anyone has an idea what causes this - some windows cache ? ..  can't reboot the PC every time
<J23K40> i now have 3seconds preview time for a simple sphere(50);
<J23K40> and even a slight view change uses 100% cpu (core) for several seconds
<peepsalot> what is that screenshot? doesn't look like openscad renderer?
<J23K40> its from meshmixer  as my openscad is totally useless now .. every action freezes the software for 10 or more seconds
<peepsalot> J23K40: do you have VBO enabled?
<J23K40> in openscad?
<peepsalot> yes, its another experimental feature
<J23K40> vertex buffer objects?
<peepsalot> yes
<J23K40> i think yes ..
<J23K40> yes all 4 vertex object render .. .but  in an older build 2023.1  it  still works fine
<peepsalot> what build is slow? did you build it yourself or its a dev snapshot?
<J23K40> have disabled them (on the old version) and the new version is still not starting without  30seconds delay after the welcome screen
<peepsalot> having VBO enabled can speed up preview responsiveness (camera movements etc.) by something like 25x
<J23K40> 2023.05.05
<J23K40> this issue  emerge after oscad crashes and close  unexpectetly
<J23K40> and it is not just the preview - every program interaction like opening a menu takes now  5-10 seconds
<peepsalot> dunno, check task manager for zombie process maybe
<peepsalot> maybe graphics driver related
<J23K40> is there any other process than openscad.exe ?
<peepsalot> no, unless you run openscad.com
<J23K40> last time the issue was gone after rebooting but directly emerged again after updating openscad ..  the strange part is that the older build is working fine
<J23K40> OK  i just renamed the exe and no it is working fast again ..  somehow windows stores  something  that is slowing down this process
<peepsalot> pastebin your library info from the newer build
<J23K40> i the renamed it to openscad.exe and it is slow again
<peepsalot> renamed what? what was it called before?
<J23K40> openscad.exe ↦  openscad2.exe  ..
<J23K40> https://bpa.st/5XXU2   here the library info
<peepsalot> are you using a space mouse?
<J23K40> yes
<J23K40> seems windows tracked the rename as the shortcut is still working
<peepsalot> are you calling with a relative path? .\openscad.exe instead of just openscad.exe so you know which one is being run?
J23 has joined #openscad
<J23> i starting from the program folder where it is installed .. and the shortcut is showing the correct  version  - if i change the extension too the shortcut stopped working
<peepsalot> what does this command output from command line: where openscad*
J23K40 has quit [Quit: Client closed]
<J23> you mean the folder where it is installed ? ..  E:\Prog\OpenSCAD2023.05
<gbruno> [github] chotchki opened issue #4649 (Manifold crashes on second render on macOS) https://github.com/openscad/openscad/issues/4649
<gbruno> [github] chotchki edited issue #4649 (Manifold crashes on second render on macOS) https://github.com/openscad/openscad/issues/4649
<J23> as i have other versions in other folders there are quite a lot openscad*  on my system
<gbruno> [github] chotchki edited issue #4649 (Manifold crashes on second render on macOS) https://github.com/openscad/openscad/issues/4649
<J23> E:\Prog\OpenSCAD2023.05>where openscad*
<J23> E:\Prog\OpenSCAD2023.05\openscad.com
<J23> E:\Prog\OpenSCAD2023.05\openscadnew.exe
<peepsalot> where tells you which executables are found on the PATH
snakedd has quit [Ping timeout: 248 seconds]
<peepsalot> i guess you have "." (current directory) on your path
<J23> at least i can work with renaming the file if it is getting slow again
<J23> you mean a dot in the foldername causing this?
snakedd has joined #openscad
<peepsalot> no i was just trying to understand how you are invoking openscad. are you clicking it from file manager or running from command line?
<peepsalot> only thing I can think of that might affect performance dependent on filename would be application specific nvidia settings. there is a tab in nvidia settings where you can configure process specific settings. make sure there's no openscad stuff there maybe.
<J23> i am clicking the file  and  i am using the shortcut from the windows taskbar
<J23> and i am clicking the openscad.exe file from the installation folder
<peepsalot> i see. nevermind all the path stuff then.
<peepsalot> was just trying to confirm that you were running the exe which you thought you were running. if you call from command line, then it depends on how your path looks and the first matching executable it finds. but if you click the exe directly then its irrelevant
<J23> hmm there are quite a lot programs but no openscad listed  (nvidia settings)
<J23> if windows identify the process only by name - maybe the crash caused some corrupted data in cash that is  causing the slow down  but strange it is not  renewed after closing the process
<J23> cache
use-value has quit [Remote host closed the connection]
use-value has joined #openscad
kintel has joined #openscad
<kintel> Fwiw, I also had strange issues in Windows, where I needed to rename my .exe to a specific filename to make it work. I never figured out how and why Windows seems to remember the filename. In my case, it didn't even matter which folder the exe was in.
<kintel> Next step is to declare my Windows dead and reinstall it : /
mmu_man has quit [Ping timeout: 246 seconds]
ali1234 has quit [Remote host closed the connection]
ali1234 has joined #openscad
<peepsalot> its kind of interesting if windows really auto tracks renames for link targets, i don't think i've tried it
ali1234 has quit [Remote host closed the connection]
ali1234 has joined #openscad
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Sauvin has quit [Read error: Connection reset by peer]
ur5us has quit [Ping timeout: 240 seconds]
peeps[zen] has joined #openscad
peepsalot has quit [Ping timeout: 240 seconds]
Sauvin has joined #openscad
Scopeuk has quit [Quit: Scopeuk]
Scopeuk has joined #openscad
<J23> remember the windows95 Easter egg  where you got the microsoft team image by creating and renaming a desktop folder https://en.wikipedia.org/wiki/List_of_Easter_eggs_in_Microsoft_products
J23 has quit [Quit: Client closed]
J23 has joined #openscad
KimK has quit [Remote host closed the connection]
KimK has joined #openscad
use-value has quit [Ping timeout: 264 seconds]
use-value has joined #openscad
guso78 has joined #openscad
<guso78> https://imgpile.com/i/9PpT1w this was created by intersecting a big item array with a big cylinder. Does OpenSCAD have something like an "overlap/touch" select operator to only pick the complete ones ?
<J23> no but you can make an intersection with a  grid that uses the norm as distance from center
<Scopeuk> no, you would have to do it through validation/conditional inside your array/looping behavior to take out anything where dia small cylinder + sqrt(x^2+y^2) > big circle dia
<guso78> yes, there are many workarounds for a circle shape, but imagine, the outer shape is more general/organic
<Scopeuk> yes, that is a more difficult problem and there is nothing in the language I know of which would solve that for you
<guso78> i believbe than openscad would benfit to have such an overlap select next to union/difference/intersection. maybe cgal could handle that ...
<J23> Grid (20,es=5)if(norm($pos)<20)cylinder(2);
<J23> if you can describe the form as  polygon you can check if parts are inside a boundary
<guso78> yeah of course, there always workarounds. question was if if can be implemented directly ;)
<J23> we both already said NO Ü
<J23> i have seen an approach that used a grid of tiny cubes that were scaled after a difference with the desired shape
<J23> as they are tiny the chances are small to cut one in half
<guso78> yeah, this is a clever approach
<guso78> once we isolate single objects from the mesh, we could calculate their volume and check if its not changed. calculating volume is a quite fast  operation.
mmu_man has joined #openscad
teepee_ has joined #openscad
teepee has quit [Ping timeout: 240 seconds]
teepee_ is now known as teepee
<J23> there was already the idea to have a volume in the summary of the render - so if we have something like v1=render.volume(){}  you could compare them
<J23> due to floating points and grid or rotations the volume may have some fluctuation though
<guso78> i already have a much better algrithm in mind!
<guso78> convert the mesh into indexed mode, and sort the resulting polygons (higher number is later)
<guso78> then do and intersection with the other shape and compare polygon listt again
<guso78> then only missing thing is to output "complete" solids
<guso78> there can be an option to keep or to discard cut objects
guso78 has quit [Quit: Client closed]
guso78 has joined #openscad
hyperair has quit [Ping timeout: 250 seconds]
hyperair has joined #openscad
<gbruno> [github] ochafik synchronize pull request #4642 (Avoid excessive reallocations in PolySet append_poly + append_vertex flow) https://github.com/openscad/openscad/pull/4642
J23 has quit [Quit: Client closed]
<gbruno> [github] ochafik edited pull request #4642 (Avoid excessive reallocations in PolySet) https://github.com/openscad/openscad/pull/4642
J23 has joined #openscad
<gbruno> [github] ochafik synchronize pull request #4642 (Avoid excessive reallocations in PolySet) https://github.com/openscad/openscad/pull/4642
<gbruno> [github] ochafik edited pull request #4642 (Avoid excessive reallocations in PolySet) https://github.com/openscad/openscad/pull/4642
little_blossom has joined #openscad
<gbruno> [github] ochafik synchronize pull request #4648 (Speed up script evaluation by precomputing identifier hashes and caching builtin function lookups) https://github.com/openscad/openscad/pull/4648
use-value has quit [Remote host closed the connection]
use-value has joined #openscad
guso78 has quit [Quit: Client closed]
snakedd has quit [Ping timeout: 240 seconds]
guso78 has joined #openscad
mmu_man has quit [Ping timeout: 240 seconds]
snakedd has joined #openscad
<gbruno> [github] ochafik opened pull request #4650 (Reserve ValueMaps in ContextFrame when possible to avoid hashtable reallocs) https://github.com/openscad/openscad/pull/4650
<gbruno> [github] ochafik opened pull request #4651 (Check/remove duplicate variable assignments earlier to speed up hot codepaths) https://github.com/openscad/openscad/pull/4651
<gbruno> [github] ochafik opened pull request #4652 (Reserve lists (VectorType) systematically) https://github.com/openscad/openscad/pull/4652
<gbruno> [github] ochafik opened pull request #4653 (Turn simplify_function_body into a virtual method) https://github.com/openscad/openscad/pull/4653
<lf94> J23: haha, thank you.
<lf94> Of course... minkowski for fillet
<lf94> The openscad code looks better than of opencascade variants too
<guso78> minkowski for fillet ? minkowski for offset ?
<lf94> so basically
<lf94> minkowski is like "flood fill" but for 3d
<lf94> you have your brush (a sphere) and say "fill this volume with the sphere"
<gbruno> [github] ochafik opened pull request #4654 ([Do not merge] Umbrella PR for non-CSG speedup initiatives) https://github.com/openscad/openscad/pull/4654
<lf94> This makes the resulting space has a fillet on the edges
<gbruno> [github] ochafik edited pull request #4654 (Umbrella PR for non-CSG speedup initiatives [Do not merge]) https://github.com/openscad/openscad/pull/4654
<gbruno> [github] ochafik edited pull request #4654 (Integrate non-CSG speedup initiatives [Do not merge]) https://github.com/openscad/openscad/pull/4654
<gbruno> [github] ochafik edited pull request #4654 (Integrate non-CSG speedup initiatives [Do not merge]) https://github.com/openscad/openscad/pull/4654
<J23> fillet is what fills an edge ..  minkowski rounds edges
<guso78> did minkowski gets fast lately with the floodfill in this channel? O:3 O:3 ?
<guso78> it always used to be  very cpu expensive ...
<gbruno> [github] ochafik edited pull request #4654 (Integrate non-CSG speedup initiatives [Do not merge]) https://github.com/openscad/openscad/pull/4654
<gbruno> [github] ochafik edited pull request #4654 (Integrate non-CSG speedup initiatives [Do not merge]) https://github.com/openscad/openscad/pull/4654
<J23> it is quite fast with manifold (or fast csg)
<gbruno> [github] ochafik edited pull request #4654 (Integrate non-CSG speedup initiatives [Do not merge]) https://github.com/openscad/openscad/pull/4654
<lf94> minkowski is generally expensive, but in fairness, SDF version is "offset" which gives a nice rounding too.
<lf94> And SDF in its entirety can be expensive
<lf94> so, usually I see minkowski as "cheating", but these days I take it back.
<guso78> in SDF you can get decide  to  get a very course result in design phase und set perfect  resolution only in the end.
mmu_man has joined #openscad
<guso78> in  scad world you very very limited abilities using $fn, $fs, $fa ..
<guso78> right now developping the overlap operator
<guso78> overlap() {a, b} outputs  only solids  within a, which overlap with shapes in b
L29Ah has quit [Ping timeout: 240 seconds]
<J23> minkowski  can be used if the  edge count is set like on a cube  however hull is faster  and minkowski also get very slow on concave objects
<J23> but sometimes it is faster then doing the math for a proper solution Ü
<gbruno> [github] ochafik edited pull request #4654 (Integrate non-CSG speedup initiatives [Do not merge]) https://github.com/openscad/openscad/pull/4654
<lf94> guso78: yea. That was my idea for my "real time developing" - have it render better and better res
<lf94> guso78: want to do me a favor and fix an fstl bug that'll help me with that? :D.
<lf94> Otherwise think I gotta write my own viewer
<guso78> haha, but i just realized its not openscad
<lf94> Yeah I know x)
<lf94> but you've got time!
<lf94> :D
<lf94> all good, I dont mean to pressure you lol.
<lf94> not nice of me...
<guso78> i think its  easy to fix. you need to temporarily diasable the event checker on changed files  during file write
cl50 has joined #openscad
<cl50> ciao  a tutti
cl50 has quit [Client Quit]
L29Ah has joined #openscad
J23 has quit [Quit: Client closed]
J23 has joined #openscad
L29Ah has left #openscad [#openscad]
L29Ah has joined #openscad
<gbruno> [github] ochafik synchronize pull request #4651 (Check/remove duplicate variable assignments earlier to speed up hot codepaths) https://github.com/openscad/openscad/pull/4651
snakedd has quit [Quit: Leaving]
L29Ah has left #openscad [#openscad]
L29Ah has joined #openscad
L29Ah has left #openscad [Error from remote client]
L29Ah has joined #openscad
L29Ah has left #openscad [#openscad]
L29Ah has joined #openscad
L29Ah has left #openscad [#openscad]
L29Ah has joined #openscad
guso78 has quit [Ping timeout: 245 seconds]
Paul32 has joined #openscad
Paul32 has quit [Quit: Client closed]
peeps[work] has joined #openscad
<peeps[work]> <guso78> overlap() {a, b} outputs only solids within a, which overlap with shapes in b
<peeps[work]> aka intersection?
J23 has quit [Quit: Client closed]
J23 has joined #openscad
<J23> yes but oSCAD intersection also cut through a solid if it is partly within the shape
L29Ah has left #openscad [#openscad]
Paul23 has joined #openscad
<guso78[m]> J23 yes, this IS Why i need an additional Filter Algorithm
<guso78[m]> ITS intersection plus Filter/sort
<Paul23> hi everyone! https://imgur.com/a/5YYZoKa -- "parentless" curly brackets don't seem to create a variable scope, $children and children([1,2,3]) also ignore them. Are there any cases when adding curly brackets inside a block *does* change the behavior?
<Paul23> in other words it seems like wrapping a few lines in curly brackets never changes the program. am I missing an example where they'd matter?
Paul23 has quit [Quit: Client closed]
L29Ah has joined #openscad
Paul30 has joined #openscad
<lf94> > leaves
L29Ah has quit [Ping timeout: 248 seconds]
<guso78[m]> Maybe IT makes Sense to extend the overlap and Sort the Input solida by Parameter constraints Like Volume, surface.area,.faces ,edges, Corners, manifoldness, holes, Container Angles and much more
Paul30 has quit [Quit: Client closed]
<gbruno> [github] ochafik synchronize pull request #4644 (Speed up hull() w/ better reserve & less alloc) https://github.com/openscad/openscad/pull/4644
guso78 has joined #openscad
peeps[work] has quit [Remote host closed the connection]
peepsalot has joined #openscad
teepee_ has joined #openscad
teepee has quit [Ping timeout: 240 seconds]
teepee_ is now known as teepee
<linext> i'm trying to use offset to make the shape of an SVG image thicker
<linext> the final rendering comes out looking weird
<lf94> increase faces? or whatever
<lf94> increase the resolution
<teepee> try with latest version
<teepee> that looks like the manifold bug that was fixed some time ago
<linext> it occurs with cgal too
<linext> if i exclude the cylinder() from the code, it comes out fine
<teepee> hmm
<teepee> I guess you can try with larger $fn
<teepee> but svg used to be on the high side anyway
<linext> if i put a cylinder under that shape, it gets weird
<teepee> that's preview though?
<linext> nope, that's the render
<linext> preview looks normal with offset
<teepee> oh, right, no CSG operation
<teepee> that is manifold
<linext> i believe what i did in the past was use another program to merge the two shapes
<teepee> again, try latest version, https://github.com/openscad/openscad/pull/4599 fixed an issue with exactly that effect
<linext> cgal also isn't right: https://i.ibb.co/6mX232w/image.png
<teepee> fix 13.4.
<teepee> do a cache clear
<linext> ah worked
<teepee> yay!
<teepee> time to check the speed of manifold for those designs :)
<teepee> good chance it's faster than the external tool you used before
<linext> not sure
<linext> i was using pymesh and inkscape i believe
<linext> the manifold version is probably faster
<teepee> it's at least worth a try
<linext> yea, i tested it
<linext> cgal takes about 10 seconds, manifold about half a second
<linext> maybe i'll go back and get rid of pymesh
<linext> it requires a docker container
guso78 has quit [Quit: Client closed]
guso78 has joined #openscad
<linext> i just ordered some more parts on JLCPCB
<linext> trying out a new thumb tab design for the leatherman free P series
<linext> a set screw is going to be used to attempt to spread the steel halves apart slightly
<linext> that way no protruding screw head
<linext> held in place by friction force
<linext> current design: https://www.ebay.com/itm/285260022017
<guso78> https://imgpile.com/i/9Yd2zg my algorithm has worked immeditaly after it compiled ...
guso78 has quit [Ping timeout: 245 seconds]
ur5us has joined #openscad
guerdy has joined #openscad
rogeliodh5 has joined #openscad
mlaga97_ has joined #openscad
ur5us has quit [*.net *.split]
peepsalot has quit [*.net *.split]
pie_ has quit [*.net *.split]
abff has quit [*.net *.split]
germ has quit [*.net *.split]
guerd871 has quit [*.net *.split]
rogeliodh has quit [*.net *.split]
gimzmoe has quit [*.net *.split]
mlaga97 has quit [*.net *.split]
ubitux has quit [*.net *.split]
rogeliodh5 is now known as rogeliodh
<lf94> guso78_: paul got back to us B)
J23 has quit [Quit: Client closed]
J23 has joined #openscad
gimzmoe has joined #openscad
use-value has quit [Remote host closed the connection]
L29Ah has joined #openscad
Tabaliah has joined #openscad
Tabaliah has left #openscad [#openscad]
ur5us has joined #openscad
pie_ has joined #openscad
abff has joined #openscad
peepsalot has joined #openscad
germ has joined #openscad
ubitux has joined #openscad
KimK has quit [Quit: Leaving]
peepsalot has quit [Ping timeout: 240 seconds]