02:55 <joseph_> teepee InPhase: It looks like we'll want to jointly consider how best to handle the following situations: choosing the value of shaderinfo (and when it should be nullptr), and the conditions for activating glUseProgram with the progid of the shader initialized in Renderer's constructor. The approach would need to be compatible with the relevant differences between Renderer and all its derived classes. Hopefully by midday tomorrow

02:55 <joseph_> I'll have cleaned up my notes to include links to specific line numbers in VBORenderer, etc. and published them so we can have a more in-depth discussion.

02:56 <joseph_> I will also search through the blame for some earlier commits in case there is a specific community member who might have insight about certain aspects of CGALRenderer (currently the blame for the most recent commit always shows teepee because of the beautify script). I was hoping that activating the ExperimentalVxORenderers feature and adding a call to glUseProgram in CGALRenderer::draw would cause the shader to be used for the

02:56 <joseph_> F6 render, but instead nothing is rendered. More information about what I've uncovered so far will be available once my latest notes are organized enough to be legible

03:26 <InPhase> joseph_: You still around?

03:28 <InPhase> joseph_: I don't quite understand the nature of the question. I would think this would flow directly from whether or not the feature is enabled and a custom shader is provided.

03:29 <InPhase> joseph_: Are there cases where that starting logic misbehaves?

07:19 <cheater> hi

07:21 <cheater> i want to order a sheet metal cutout (cnc routed) from pcbway and i need to deliver a drawing. what's better for that purpose, openscad or freecad? i don't know either program. the part is relatively simple, it's basically a square with some smaller squares cut out from the inside. i need to indicate things like tolerances. i just need a 2d drawing, since it's sheet metal.

07:28 <teepee> cheater: if there's single line cuts somewhere, OpenSCAD may cause some issues as it operates on closed objects, even in 2D

07:29 <teepee> e.g. you can't have 2 rectangles touching on an edge and keep that edge. it will merge into a single bigger rectangle

07:29 <teepee> another option might be librecad as that's dedicated 2d cad

07:30 <teepee> I'm sure freecad can do that too, but I don't have much experience with that, but people in that channel probably will give advise on that :)

07:51 <crazy_imp> cheater: which file formats do they accept? that may pose some restrictions too

07:52 <crazy_imp> (and if they offer a preview for your uploaded files you can try without loosing money and time)

07:53 <crazy_imp> s/loosing/losing/

07:53 <cheater> teepee: what's a single line cut?

07:53 <cheater> ah

07:55 <cheater> crazy_imp: 3D CAD: *.step, *.stp, *.x_t , *.iges, *.sldprt 2D Drawing: *.dwg, *.dxf, *.pdf

07:56 <crazy_imp> https://en.wikibooks.org/wiki/OpenSCAD_User_Manual/3D_to_2D_Projection

08:23 <J1A8462> go with squares export as dxf

11:35 <Fyr> hello, I need to analyze stability of a thing I created with OpenSCAD. how can I, at least, calculate projection of density?

11:55 <teepee> there's nothing included directly, but I think there's some tools for that using STL as input

11:58 <teepee> or as alternative, I believe FreeCAD has some FEM stuff included now and also has an OpenSCAD workbench for loading OpenSCAD designs

12:13 <J1A8462> not sure what density is as this would require to define materials .. but you can cut through a model (or use projection) to see thickness

12:17 <Fyr> J1A8462, I want to make sure that the model won't fall.

12:17 <Fyr> I don't know what my eye can say about thickness.

12:41 <InPhase> Fyr: Like a center of mass calculation?

12:44 <Fyr> something like that

12:44 <Fyr> more like distrubution of mass

12:45 <InPhase> Fyr: It seems meshlab has a geometric measures feature that might help with center of mass.

12:46 <InPhase> "distribution of mass" is not a well-defined concept. The distribution of mass is roughly the shape itself.

12:46 <InPhase> (The only reason it's not exactly the shape is because of options for non-100% infill.)

12:46 <Fyr> InPhase, the shape is hollow.

12:48 <Fyr> the entire shape, the shape of the void are difficult.

12:49 <InPhase> Center of mass of the surface is a much simpler calculation, such that one could write code for that oneself with only a modest geometry effort. I'm not sure if meshlab specifically has a surface calculation for that.

12:50 <InPhase> The calculation is to just iterate over the triangles and add up their area times their center position vector in a loop, then divide by the summed surface area.

12:51 <InPhase> If you select the ascii output you could write that in a pretty small Python script.

12:54 <Fyr> InPhase, https://www.thingiverse.com/thing:5410657

12:54 <othx> Fyr linked to "Outline vase by mishkin2" on thingiverse => 1 IRC mentions

12:54 <Fyr> surface calculations won't help.

12:58 <J1A8462> you don't need calculation for that .. as it is vertical .. the problem is what you put into the vase

12:59 <peeps[zen]> as I understand from the keywords of projection, density and distribution: sounds like you would want to project a grid of rays through a geometry. for each ray, record the total length which intersect the object, then basically view those data points as an input to surface() ?

13:01 <peeps[zen]> there's not any particular straightforward way to do a detailed analysis like that via openscad

13:08 <peeps[zen]> oh, just looked at the vase example. I wouldn't really describe that as a projection of density, its more a very abstract representation of shape derived from projection outlines, from couple orthogonal planes

13:20 <InPhase> Fyr: Yeah, basically if the thing inserted is centered or balanced toward the back, that's stable. If the thing inserted leans outward and it's heavy it will tip. Make sure the thing inserted is lighter than the printed vase, and not super long, and it will be stable. If it's a heavy object inserted, it needs to tilt back or it will fall.

13:21 <Fyr> it was just an example of a complex shape.

13:21 <InPhase> Fyr: That is not a good vase design for a boquet of flowers. :) It's suitable only for small things like inserted.

13:21 <Fyr> my shape is even more complex and hollow.

13:22 <InPhase> If truly hollow (like zero infill), then it is, in fact, a surface mass calculation.

13:22 <Friithian> put it into blender, make it ridgid body, see what happens

13:23 <InPhase> Fyr: The engineer in me will suggest finding an artistic way to add stability so it's not an issue in the first place. But the mathematician in me says you can just add up the surface center of mass from the STL file like I explained, and know the limits from that.

13:32 <peeps[zen]> Fyr: is my grid of rays example more or less what you are asking for then?

13:34 <peeps[zen]> Fyr: maybe a better question is, what do you need this "mass distribution" for? what would you do with that information?

13:36 <peeps[zen]> you mention analyzing stability, but stability in what context?

13:39 <peeps[zen]> of course none of my messages reach them *rolleyes*

13:40 <Friithian> is it really irc if someone doesn't disconnect when you're trying to help

13:43 <J1A8462> i like that rigid body approach ..  or  print it  small and see if it tips  over

13:45 <Friithian> throwing into blender and putting a rigid body on it and a plane would be pretty quick and show any obvious issues, but printing it would be the best way to see if it'll tip. If only we knew the reason they needed this info

13:47 <J1A8462> .oO in the end it always is a dildo

13:48 <Friithian> well I hope they're using body safe materials

14:42 <J1A8462> i mean if it is likely to tip over .. just add a big suction cup at the bottom

23:06 <joseph_> InPhase: A reproduction of my current approach is now pushed. I thought it would be easiest to show the changes in the form of a draft PR within my own fork, so not to conflict with development in the actual openscad repo. https://github.com/jbinvnt/openscad/pull/1. This shows how I removed the showedges requirement so that the shader is used in F5 preview once the VBO feature is on. But for the F6 render, my call to glUseProgram

23:06 <joseph_> leads to nothing rendered. I didn't see any OpenGL errros logged

23:08 <brianx> nevermind, i killed the process and it's gone.

23:17 <teepee> joseph_: did you try running with debug to enable those PRINTD statements?

23:18 <teepee> maybe that helps tracking down which of the draw code is active

23:23 <teepee> otherwise there's also renderdoc for tracing OpenGL calls, but it's a bit difficult to use, at least I did struggle a bit making sense of it's traces

23:35 <joseph_> teepee: I didn't know a debug mode needed to be enabled in order for PRINTD to be displayed. This might be my issue. I have used the debug build args during compilation of course, but I didn't see anything in the docs about an environment variable or similar requirement to show debug messages at runtime. What do I need to change?

23:40 <teepee> I think it's independent from code debug stuff. you can enable via --debug=<SourceFileName>,...

23:40 <teepee> or just --debug=all should be quite verbose

23:44 <teepee> yep, openscad --debug=CGALRenderer works even on the release version

23:45 <teepee> the output is not great though, at least for the render code