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
mtm has quit [Ping timeout: 260 seconds]
mtm has joined #openscad
ali1234 has quit [Ping timeout: 265 seconds]
ali1234 has joined #openscad
guerd has joined #openscad
califax has quit [Remote host closed the connection]
califax has joined #openscad
lockywolf has quit [Quit: ZNC 1.9.1 - https://znc.in]
lockywolf has joined #openscad
J24k has quit [Quit: Client closed]
J24k has joined #openscad
fling has quit [Remote host closed the connection]
fling has joined #openscad
J24k48 has joined #openscad
J24k has quit [Ping timeout: 240 seconds]
mmu_man has quit [Ping timeout: 252 seconds]
foul_owl has quit [Ping timeout: 244 seconds]
foul_owl has joined #openscad
teepee has quit [Remote host closed the connection]
teepee has joined #openscad
J24k48 has quit [Quit: Client closed]
J24k48 has joined #openscad
foul_owl has quit [Ping timeout: 252 seconds]
foul_owl has joined #openscad
hyperair has joined #openscad
drfff has joined #openscad
teepee has quit [Quit: bye...]
teepee_ has joined #openscad
teepee_ is now known as teepee
teepee_ has joined #openscad
teepee has quit [Ping timeout: 264 seconds]
teepee_ is now known as teepee
mtm has quit [Ping timeout: 246 seconds]
mtm has joined #openscad
mmu_man has joined #openscad
f0lkert has quit [Quit: ZNC - https://znc.in]
Guest9 has joined #openscad
Guest9 has quit [Quit: Client closed]
snaked has quit [Remote host closed the connection]
extor has quit [Quit: ZNC 1.8.2+deb2ubuntu0.1 - https://znc.in]
f0lkert has joined #openscad
TheCoffeMaker has quit [Ping timeout: 260 seconds]
TheCoffeMaker has joined #openscad
little_blossom has quit [Quit: little_blossom]
little_blossom has joined #openscad
f0lkert has quit [Changing host]
f0lkert has joined #openscad
<L29Ah> omg new openscad release takes ages to run
<L29Ah> over 10 times more than the previous one
<L29Ah> oh, "release" i mean, 20241114 snapshot vs 2021.01
<InPhase> ? Did you enable manifold?
<InPhase> It should be like 50x faster for most models. :)
<L29Ah> yes, it's enabled by default in Gentoo it seems
<L29Ah> the problem is not model rendering time, it's just the time it takes to show the window
<L29Ah> it's slow even w/o any models
<L29Ah> seems like it spends all the time in QFontEngineFT::loadGlyph
<InPhase> That's unusual.
<InPhase> L29Ah: It must be some specific issue going on with your platform or some system library setup. It might be worth sharing this detail information on an issue.
<InPhase> L29Ah: 2024-12-11 (the latest one posted) takes the same time as normal to launch for me. Can you try the AppImage directly from the site?
<L29Ah> narrowed it town to my ~/.config/fontconfig/fonts.conf
<InPhase> L29Ah: I also just tried both nightly snapshots that were released as AppImages on 2024-11-11, and the one on 2024-11-14, and the next one on 2024-11-18, and all of them launch at a normal speed for me. So test those, so we can see if it's a build issue with how gentoo has done things.
<L29Ah> minimal repro: http://0x0.st/XFLN.conf
<L29Ah> seems like openscad is upset when it's being told what font to use
<InPhase> Does this happen if you use a different font?
<L29Ah> InPhase: it doesn't happen with Noto Sans but happens with GNU Unifont; perhaps it's upset by bitmap fonts in particular
<InPhase> Unusual, but interesting. Do any other recent build Qt programs have issues?
<InPhase> Also... I just today noticed that nightly builds of OpenSCAD have different inferior fonts with them to the ones my own builds from the master branch pick up... Apparently this has been going on for a while, but I never noticed.
Guest77 has joined #openscad
<InPhase> I'm not sure exactly why they're picking up different menu and other gui element fonts. I guess the AppImage must bundle its own or something, and they're just inferior fonts to the ones my system normally uses.
Guest77 has quit [Client Quit]
Guest77 has joined #openscad
<L29Ah> InPhase: no
<L29Ah> things like qutebrowser, marble and qbittorrent feel fine
<L29Ah> my random gdb backtrace might help: https://0x0.st/XFLm.txt
<InPhase> https://imgur.com/a/lOqddZ0 On the right is my most recent build. On the left is what the nightly builds have apparently been looking like for at least a year and a half: https://imgur.com/a/lOqddZ0
Guest77 has quit [Quit: Client closed]
<InPhase> L29Ah: So the harfbuzz library is decompressing a whole bunch of unicode crap?
<InPhase> L29Ah: So gimp and thunderbird are showing up in my dependency tree as also using harfbuzz, but very old versions. We seem to be by default building with 2.6.4
<InPhase> L29Ah: How about gedit? It looks like that's using a reasonably up-to-date harfbuzz through its libpango dependency.
<InPhase> And gedit normally loads quick.
<InPhase> L29Ah: Also, I tried your font config thing, but it didn't trigger a problem. But I probably don't have the underlying system setup to trigger whatever is going wrong.
<L29Ah> gimp loads as fast as usual
<L29Ah> installing gedit to check
<tcurdt-> I am little puzzled. I have a svg that I import. When I try to linear_extrude it, it just disappears. Not even an error message. Any pointers?
<L29Ah> my system only has one harfbuzz
<L29Ah> gedit runs instantly
<InPhase> tcurdt-: What happens if you do linear_extrude(height=1) offset(0.01) import(...)
<tcurdt-> InPhase I just got it working. It was related to using %
<InPhase> L29Ah: Can you determine with harfbuzz version that gedit is using?
<InPhase> tcurdt-: Oh.
<tcurdt-> I've see that a couple of times before that % makes things disappear when they should not. But I could not yet find a pattern.
<tcurdt-> It seems to be related to % with difference ... but I am not sure yet
<InPhase> tcurdt-: I could see how using it inside a difference could potentially do weird unintuitive things. There are probably a lot of edgecases untested.
<tcurdt-> Next time I encounter it I will open a ticket. Maybe it helps to track down.
splud has quit [Ping timeout: 246 seconds]
<L29Ah> InPhase: 10.1.0
<L29Ah> the same as everything else
<InPhase> Oh. That's one of those projects that rapidly increased its version numbering rate I see. :)
<InPhase> I had to check to see how they got to 10 so fast. lol
<InPhase> L29Ah: Then I don't really have an explanation for what's different. You could open an issue, but it's probably not fixable unless it's something others are experiencing (so that more info is available) or if you could find more info.
<InPhase> L29Ah: Right now it's looking like some sort of perfect-storm kind of bug, but we don't know where the storm is.
<L29Ah> i could do a bisect, but it will be quite clunky due to how much have changed since the last release
splud has joined #openscad
germ_ has quit [Read error: Connection reset by peer]
germ has joined #openscad
mmu_man has quit [Ping timeout: 246 seconds]
mmu_man has joined #openscad
TheCoffeMaker has quit [Remote host closed the connection]
TheCoffeMaker has joined #openscad
TheCoffeMaker has quit [Ping timeout: 272 seconds]
TheCoffeMaker has joined #openscad
<ali1234> is there a way to do relative colours?
<ali1234> like color_scale() or something?
<L29Ah> seems like it's normal for C++ code to cease being buildable after half a year :[
<pca006132> L29Ah: can you do `perf record -g openscad` or something and upload the trace?
<pca006132> you probably want to build openscad as RelWithDebInfo though
<L29Ah> ok, hang on
<ali1234> reading scrollback - could this be a font cache issue?
<ali1234> building the font cache for certain huge unicode fonts (like roboto) can take ~20 seconds. distributions ship an already populated cache to avoid this on first boot, but there was at least one bug where the cache was invalid and always got rebuilt anyway
<ali1234> could very much be connected to openscad using the wrong font for gui, if the font it decided to use isn't cached
<L29Ah> filed a bug while it rebuilds w/ symbols: https://github.com/openscad/openscad/issues/5495
<gbruno> [github] l29ah opened issue #5495 (openscad takes a long time to show the main window when a bitmap font is used) https://github.com/openscad/openscad/issues/5495
<teepee> that does not sound like an openscad issue
<L29Ah> ali1234: terminus isn't a huge unicode font, neither any other software experiences this
<pca006132> seems that qt6 doesn't work with bitmap font properly... idk
<ali1234> bitmap fonts can never be huge since pcf fonts only support 16 bit codepoints :)
<L29Ah> seems like qt-6.7.3 isn't 6.2.1
<teepee> that sounds way more likely with the stack trace
<L29Ah> ah, that's kde version, not qt
<L29Ah> and the user noted it affects every qt software, while all the other qt6 stuff i have works fine
<pca006132> interesting
<pca006132> anyway, will just wait for your perf trace
<teepee> deleting the font cache as suggested by ali1234 sounds like a simple test too
guso78k has joined #openscad
<ali1234> deleting the cache would just force a rebuild so i'm not sure that would do anything. it would still be slow whether that is the cause or not
guso78k has quit [Client Quit]
guso78k has joined #openscad
<guso78k> Still I understand why CSG tree of that show more than 300 hierarchy levels to generate this BOSL2 - cuboid
<teepee> it was on windows, but the effect was that a broken cache file caused the rebuild that never resulted in a valid cache
<teepee> or stracing where the cache tries to write, maybe it's a wrong location
<ali1234> yeah - perhas bulding the cache and then attempting to write it somewhere that is read only, and silently failing
<ali1234> but deleting the real cache wouldn't solve that
<L29Ah> pca006132: seems like it takes ages for the window to show up under perf; would an interrupted one suffice?
<L29Ah> ok it finished
<gbruno> [github] pca006132 opened issue #5496 (Shader error for preview window) https://github.com/openscad/openscad/issues/5496
<gbruno> [github] pca006132 edited issue #5496 (Shader for preview window using deprecated OpenGL features) https://github.com/openscad/openscad/issues/5496
Guest52 has joined #openscad
<teepee> pca006132: wrong Qt build in that issue, GLES backend is not supported
<Guest52> Hi, I'm seeing a bunch of boardgame inserts on printables and some of them in openscad, but solid. I thought, wouldn't it be nice to skip out on the infill? But I also don't think this is easily achieved... or maybe not even relevant due to the low infill. What's your opinion?
<teepee> Guest52 that's what slicers usually do better
<pca006132> L29Ah: sadly not much can be confirmed from the trace other than it is probably having a cache miss and is extracting the font for some reason...
<pca006132> teepee: I see, will look at that
<teepee> pca006132: I don't know if it's still the same for qt6 but qt5 had a switch to either do GLES or DesktopOpenGL
<gbruno> [github] pca006132 closed issue #5496 (Shader for preview window using deprecated OpenGL features) https://github.com/openscad/openscad/issues/5496
<pca006132> L29Ah: it seems that fonts-conf has environment variables that can print more verbose debug information to show what is happening
<pca006132> maybe you can have a look at that and see if it is telling us something is wrong...
Guest52 has quit [Quit: Client closed]
guso78k has quit [Quit: Client closed]
<pca006132> teepee: it seems that it requires setting up some opengl profile stuff (https://forum.qt.io/topic/140469/using-legacy-opengl-calls-with-qt6/6)
<pca006132> not familiar with opengl
<teepee> that's for the client application
<teepee> I think that should be all covered, Qt just needs to be built with --opengl instead of --gles3
<teepee> well, that was how it worked for qt5
<pca006132> hmmm, but I can't see any build options related to this, or qt packages
<teepee> hmm, let me check the debian builds
guso78k has joined #openscad
Guest27 has joined #openscad
Guest27 has quit [Client Quit]
<teepee> found something for raspi using: -DINPUT_opengl=es2
<teepee> qt config docu is still impossible to find
<pca006132> it says INPUT_opengl is not used by the project
<pca006132> anyway, probably not very important for now, I just wanted to try QT6 on nixos to reproduce the font issue mentioned above
fling has quit [Ping timeout: 264 seconds]
fling_ has joined #openscad
fling_ is now known as fling
califax has quit [Remote host closed the connection]
califax has joined #openscad
teepee_ has joined #openscad