<J2364424>
I need some way to show clear infos in the console .. just got this PM »I'm getting errors in line 5: "ERROR: Assertion '(Version >= useVersion)' failed: "lib version undef detected, install 22.078 ub.scad library‼ «
<J2364424>
As missing libraries are probably a quite common issue maybe the parser could display a popup if the "use" or "incude" or "import" files are missing
<teepee>
I would hate actual popup windows, but there's probably plenty of way improving the error window behavior and content
<J23>
JordanBrown[m] I am using assert so there is the warning and the assert failed in red .. but users are still confused - i mean this one had at least the console open but obviously wasn't able to understand the message. That made me think - someone with no knowledge about scad or how software works is opening a file from thingy/printables
<J23>
and it doesn't work. If a pop up would take his attention with " your script uses a file that is not accessible " please get that file and copy to this location : "current path" - then two buttons " i understand" (close) - " more information" which leads to manual/docu wiki
<J23>
I am too not a friend of popups but a missing file is a condition that is so severe and new users can be quite unfamiliar that multiple files needed - and as we don't have a container that includes all files ( i think i have seen a feature request/issue for that one"
ToAruShiroiNeko has quit [Ping timeout: 268 seconds]
ToAruShiroiNeko has joined #openscad
ccox has quit [Ping timeout: 264 seconds]
ccox has joined #openscad
<Scopeuk>
This falls back ultimately to the package manager concept, I know someone put forward a proposal using a vcpkg varient a little while back but as far as I'm aware it sort of died
foul_owl has quit [Ping timeout: 248 seconds]
foul_owl has joined #openscad
LINKO has joined #openscad
fling has quit [Remote host closed the connection]
<LINKO>
Hi. I'm completely new to OpenSCAD, because I found a file I wanna customize and print, and it came with a .scad-File.
<LINKO>
Is there any way to boost performance on the Windows version? On the example-files it seems fine, but with the file I wanna work on it's pretty much unusable.
<LINKO>
I have a fairly beefy PC (5900X, 32GB RAM, RTX 3080). The latest release 2021.01 is completely unusable and basically crashes after loading the given file. The latest SNAPSHOT-Version is a fair bit better, but still pretty bad.
<LINKO>
I have tried running the Linux-Version of 2021.01 through WSL2 which honestly performs even better than the Windows-Snapshot, but it's still far from great.
<J23>
but if the design is bad - there is not much, like every software can be slowed down
<J23>
you can start with setting $fn=0; $fs=3; at least for the preview
<J23>
line 53 .. just set to 0
<J23>
if you select shell style cylinder this works much better .. else you need to add "render()" in line 117 before the difference - that is slowing down a lot
<J23>
however there are still several design flaws causing unpleasant user experience
<J23>
(also fast CSG is failing and can't be used for that reason)
<HimeHaieto>
linko: one of the bigger things to keep in mind about render performance with openscad/cgal is that it's all purely single threaded on your cpu and afaik there's never been any gpu acceleration either
<HimeHaieto>
the gui may use opengl, but the actual render doesn't, so it won't get any faster with additional cores or a better gpu
<HimeHaieto>
it probably shouldn't crash like that though, not sure what might cause that
<HimeHaieto>
fyi the unmodified source did a full render (to stl) for me in 20 minutes, and a png preview in 20 seconds
<LINKO>
J23 Thank you very much. It's a lot more responsive now and preview is pretty much instant now (still using WSL). However when trying to render I get this error with your code: "ERROR: The given mesh is not closed! Unable to convert to CGAL_Nef_Polyhedron."
<J23>
Yes that is what i meant that there are issues with the geometry - maybe with the 2021 version this will work
<J23>
I tried to fix the non overlapping surfaces but it is bit a pain to find all of them .. you can use ! to only render a sub part to check if that is working
<J23>
so if you use "shell style= cylinder" then the connecton_groove causes the issue.
<LINKO>
Ok thank you so much. I can work around it.
<J23>
LINKO so here you go https://bpa.st/T74XW i am not sure why this happens but i changed the order of the union and difference and now it is rendering fine
<J23>
teepee i wonder why render() works fine in preview but not in render - i thought render() is doing the same F6 is doing?
<teepee>
it's doing the same in the sense of forcing mesh calculation at this point it's written as F6 does for everything
<teepee>
the display code and the top-level operation might still be different to full F6
<HimeHaieto>
making me feel weird for never once using the gui application
<J23>
teepee so i have had a union with render() in front .. displays proper ( took some time till the roof worked with the offsets ) .. then i hit F6 and get "mesh not closed" .. tried flush cache but still (fast CSG) And i set all values to fix values so preview and render uses the same .. bit annoying that i can't check this
<J23>
in preview
<HimeHaieto>
I actually never even realised there was one at all for maybe the first few weeks I used it
<J23>
HimeHaieto so you don't use preview and always a full render ?
<J23>
or just an external editor
<teepee>
hmm, not sure if fast-csg causes some extra strange effects, it's possible I suppose as it tries to prevent extra conversions because it introduces an additional type of internal mesh data structure needed for those CGAL functions
<J23>
it is terrible if starting to use all these experimental features .. you won't miss them but it can get tough if complexity increases
<teepee>
yep, that's the unfortunate side effect
<teepee>
that's the price we play for only being able to progress with things if people find the time
<J23>
oh btw did it make sense that animations saves a backup file for each frame? - doesn't that slow things down?
<teepee>
backup? I've not seen that
<teepee>
that sounds indeed strange
<J23>
the backup is written each F5 .. so animation triggers that each frame
<J23>
( at least the console is telling me that a backup is saved each frame)
<teepee>
oh, not an image backup, scad backup
<J23>
yes
<teepee>
that makes no sense either
<J23>
i mean if you change things while in animation - so if there is some detection if something was changed since the last backup and it is not just $t
<teepee>
I guess detecting actual change in the text should be fine. after all that's what the editor contains
<teepee>
regardless of parameters put into it that could change the actual resulting mesh
<J23>
if that costs less resources than saving the file Ü
<veverak>
eeeeeh
<veverak>
let's say that I want to design a thing with GT2 belts used inside
<veverak>
what is the best openscad lib currently to generate GT2-compatible gears?
<teepee>
I don't know any that does those, but I never had a close look
<teepee>
I think Brandano on #reprap did something related to GT2 stuff
<teepee>
but I don't know if that's online somewhere
<HimeHaieto>
j23: I've used preview a little in the form of png output, but for any more serious stuff, yeah, I always do full render and use separate software to view the stl...or if not, it's because I'm loading the scad source directly into the likes of freecad instead (in particular, so that I might be able to export step...)
<J23>
veverak mine has precise GT2 belt profiles
<veverak>
J23: do you have a link or something?
<LINKO>
J23 I'll try it later. Thanks again.
<LINKO>
If I find some free time it definitely seems like it would be worth to dive a bit deeper into OpenSCAD.
<LINKO>
I really like the idea of CAD by coding, being a software dev myself.
<HimeHaieto>
I do most of my work more "blind" in that I might write and write, but not interactively "see" the result...I generally don't feel like I need to, I can usually understand what I'm doing well enough without that
<HimeHaieto>
I rather agree about the "cad by coding" concept...I actually originally started with just freecad, but it felt horrible trying to get anything done, or worse, changing anything too much after I already did stuff
<InPhase>
guso787: The github actions are controlled under ./.github/ and ./scripts/ as teepee highlighted, and will alter behavior immediately in a PR. I made use of that when altering the testing dependencies, also to include additional Python support, but in that case in the environment. It altered the build process and testing in the PR successfully before merge.
<InPhase>
guso787: This is why first time contributers require approval before testing will run on a PR. Github was having problems with bots submitting PRs containing crypto miners.
<HimeHaieto>
maybe it's partly because I'm working under certain restrictions, but I would already know things like how I wanted particular shapes of exact sizes at exact coordinates, etc, and it just seemed like a borderline nightmare having to constantly fight it to do what I wanted, whereas in openscad it was really easy
<J23>
veverak just compared mine with teepees https://imgur.com/a/aLpFni9 .. strange they don't fit exactly .. even if i remove the clearance and the evolute generation you finde it on openscad.org/libraries ub..scad
<Scopeuk>
That looks like there is a mixed qt setup on their, double check the latest version of qt networking package is installed?
<J23>
i am annoyed that these marks are only at the center .. but if you are at [20,20] and zoom in .. they are gone
castaway has joined #openscad
epony has joined #openscad
<teepee>
rapha: that error message is a bit strange, I don't see a qt-network package for arch
<teepee>
do you have qt5-multimedia installed?
<teepee>
aaand fixed, switch on the outside is so much better than opening the battery compartement twice every day to switch it on and off... https://imgur.com/a/4kXR5qB
<teepee>
also there's the question of how often you can mess up that simple wiring. the answer appears to be 3 :-)
<teepee>
rapha: oh, as Scopeuk already pointed to the version numbers mismatch, probably between qt5-multimedia and qt5-base
<teepee>
why would that be allowed by the package manager anyway
fling has joined #openscad
<J23>
teepee nice fit .. and that black glitter filament looks great
<teepee>
yep, I do like that "blue pearl" from "Das Filament"
la1yv_a has joined #openscad
la1yv_a has quit [Quit: Leaving]
castaway has quit [Ping timeout: 246 seconds]
linext has quit [Read error: Connection reset by peer]