01:43 <Guest2933> Anybody notice how they have perspective and orthogonal mixed up now?=$

03:06 <InPhase> They who? Are those OpenSCAD devs messing things up again?

10:58 <juri_> InPhase: guess it's time for everyone to switch to implicitcad, then. ;)

11:21 <crazy_imp> uhm, openscad window not in focus - translation input from spacemouse still works (unexpected because no focus), rotation input doesn't (as expected)

12:07 <gbruno> [github] printerderby closed issue #4179 (Performance issues with 2022.03.14 on higher $fn values) https://github.com/openscad/openscad/issues/4179

12:34 <gbruno> [github] t-paul pushed 2 modifications (Merge pull request #4226 from MichaelPFrey/mfr-2022-04-18-pref Fix preference setting for RotateVPRelGain.) https://github.com/openscad/openscad/commit/294ab163bc70128c870ea51ebcf58d4fcebbfba9

14:13 <gbruno> [github] MichaelPFrey assigned issue #4030 (Option to clear console on each Preview/Render) https://github.com/openscad/openscad/issues/4030

15:22 <buZz> crazy_imp: yeah thats a ancient bug

15:22 <buZz> at least 4 years :P

15:25 <gbruno> [github] t-paul pushed 12 modifications (Merge pull request #4222 from MichaelPFrey/mfr-2022-04-15 Trace: show parameters of user modul calls, make trace depth configurable.) https://github.com/openscad/openscad/commit/b25862ffdbe0fab1cfd060d5f2c91a654effe78e

15:29 <gbruno> [github] t-paul pushed 3 modifications (Merge pull request #4221 from MichaelPFrey/2022-04-15-include-use-newline Warning when a \n is found in use<> or include<>.) https://github.com/openscad/openscad/commit/b9fff2b9132333ea3b337ba47d4ef230b4da9491

15:55 <crazy_imp> buZz: is there a ticket for it? https://github.com/openscad/openscad/issues/3724 sounds like quite the opposite :D

15:55 <buZz> i have never seen it work -ANY- different

15:56 <buZz> crazy_imp: spacemice are -not- joysticks, obviously

15:56 <crazy_imp> yeah, but still focus and input related ;)

15:56 <buZz> crazy_imp: that issue indeed confirms the lack of sane defaults or configurability

16:00 <crazy_imp> buZz: not sure how much of this is linked to the fact that the used input libs continue to provide input without checking for the focus

16:00 <teepee> from OpenSCAD perspective spacemouse are really just joysticks ;-)

16:00 <buZz> right, so why is rotation and translation different from the 1 joystick

16:00 <buZz> because they arent

16:01 <teepee> from software side, they are identical on Linux going through the linux joystick driver

16:01 <teepee> some special stuff going on using HID or SpaceNav but that's not a huge difference in the end

16:08 <teepee> hmm, where's my spacemouse then

16:14 <buZz> mine is so ancient , it uses that weird daemon

16:14 <veverak1> Hi folks

16:16 <veverak1> by any chance, anybody know a good algorithm for this 2D problem? : I have a 'rectangle' with fixed dimensions, a sequence of points with x,y,theta (coordinates and rotation). I want to take the rectangle and sweep it over the sequence of points to generate a polygon of the "path" (it's a car)

16:16 <veverak1> How to get the polygon of the path?

16:16 <veverak1> (fast)

16:20 <teepee> no idea, but it's probably somewhere in there: https://github.com/openscad/openscad/pull/918

16:23 <gbruno> [github] t-paul closed issue #4030 (Option to clear console on each Preview/Render) https://github.com/openscad/openscad/issues/4030

16:23 <gbruno> [github] t-paul pushed 4 modifications (Merge pull request #4228 from MichaelPFrey/mfr-2022-04-18-consol Add auto clear console option to preferences) https://github.com/openscad/openscad/commit/ca2ba5f64c8d2be2ccf6394331d74d2b0647cb7c

16:23 <veverak1> wow, thanks!

16:24 <teepee> I wish he'd left some pointers on how it works, I'm still a bit sad seeing that rotting away

16:24 <teepee> the focus thing looks like a simple bug

16:25 <teepee> when having the preferences window open and focussing that, it stops all input

16:25 <teepee> I guess it's just some copy&paste code that should have been unified into a single place

18:03 <gbruno> [github] t-paul pushed 2 additions 4 modifications 4 removals (Merge pull request #4227 from MichaelPFrey/mfr-2022-04-18-mousewheel Ignore mouse wheel in preferences.) https://github.com/openscad/openscad/commit/59386b965a53254dbf824afd0e549eaf8162074b

18:12 <gbruno> [github] t-paul pushed 24 modifications (Merge pull request #4220 from MichaelPFrey/mfr-2022-04-14 Fix and regenerate ord-tests.) https://github.com/openscad/openscad/commit/260d3ba45f2212b231e9b7e0395e1b4ffbc7bf6b

19:54 <AndrewStone[m]> I don't know if this is the spot to talk about implementation details. But I'll just ask my question now and if its not the right place, feel free to let me know!

19:55 <AndrewStone[m]> I'm looking at StackCheck::inst() and its use (for example) in UserModule.cc.

19:57 <AndrewStone[m]> (nvm figured it out)

20:00 <teepee> AndrewStone[m]: it's the right place too :)

20:02 <AndrewStone[m]> thx! I'm working on deferring computation to other threads so that the UI remains responsive... obviously StackCheck therefore needs to be a little more sophisticated!

20:03 <teepee> ah, yes. right now there's only the CGAL thread which is the mesh calculation part

20:04 <teepee> the 2 other big chunks "compile" and "preview" are on the GUI thread. so you are looking at the compile part?

20:05 <teepee> there's another trap, well, one I know of, that there's some static information used

20:05 <teepee> I tried to at least make it more obvious but that comes into play when spawning separate threads too

20:11 <AndrewStone[m]> I've made a general job executor (in threads), and then a slot that executes an arbitrary function to defer UX calls back into the GUI thread. I found the real hiccup in compile is actually execution of compileDone(bool didchange), so so-far I've just deferred that (instantiateRoot). But I think updateCompilerResult() is also causing a problem.

20:12 <AndrewStone[m]> So that's working with a StackCheck workaround. This first step is a little easier than a generalized multi-thread perf speedup, because I'm just moving computation out of the GUI thread so presumably there remains only 1 thread actually doing "real work". But can you point me to the static vars?

20:16 <AndrewStone[m]> (no its not updateCompilerResult... its something executing after that)

20:16 <teepee> StaticModuleNameStack in UserModule

20:17 <teepee> I think that's one of the reasons for that GUI lock in MainWindow that prevents parallel execution

20:19 <teepee> I believe all that stuff now has actually a place to go, as there's the EvaluationSession now

20:19 <AndrewStone[m]> perfect, I won't have to make something similar

20:28 <teepee> I'm not sure it covers everything already, but as it's meant to handle the memory management, it should be at least in scope of the whole compilation part

20:35 <AndrewStone[m]> great, I'll fool around and if I get something that seems stable open an MR if you want one but I'm pretty new to this codebase so will probably try it in 2 parts. Part 1 is just defer expensive operations out of the gui thread to keep the gui responsive (like you'll be able to edit or scroll around the code while its computing). Part 2 is actually multithreading the compile/compute. This is more where we'd need every global that

20:35 <AndrewStone[m]> depends on compile context to be put into that Session object. There's also another problem with that which is to figure out WHEN to launch in a new thread. I already tried doing so in NodeVisitor, but that was foolish as I'm sure you know.

20:35 <AndrewStone[m]> sry, gtg

20:35 <teepee> sounds good

21:08 <gbruno> [github] t-paul pushed 3 modifications (Merge pull request #4229 from MichaelPFrey/mfr-2022-04-18-error-log Error log improvements.) https://github.com/openscad/openscad/commit/110ca328899e031c5fbd21f8caf34734a24b3d7e