02:00 <gbruno> [github] jbinvnt synchronize pull request #4675 (GSoC 2023: Create VertexStateManager) https://github.com/openscad/openscad/pull/4675

02:01 <joseph_> teepee: I just made another commit that contains only an additional comment. Perhaps if the CI runs again, then there won't be mystery errors this time. Because all I changed previously was making something a const reference

11:26 <Guest53> Hello. Assume, I have an array of triangle vertices (each 3 vertices for one trangle). How can I write a function, that returns a new array, where each triangle is split in 3 triangles, by adding a 4th point in the middle?

11:46 <teepee> Guest53: this should work with nested list comprehension. it needs to be a single expression as functions are limited to that

12:19 <J23k32> subdivision a loop through the faces that creates for each face a new center point and then  3 new faces

12:20 <Guest53> assume module m(r, d); I'd write "r= (r==undef ? d/2 : r)". But I can't overwrite the parameter. Why did you decide, not to allow overwriting parameters?

12:21 <J23k32> https://en.wikipedia.org/wiki/Subdivision_surface    the problem is that without calculating new geometry you will not get a better resolution

12:22 <J23k> Guest53 you have constants that can't be overwritten - these are not variables

12:23 <J23k> Also your new triangles will probably removed when rendering if they are in the same plane as the other

12:24 <J23k> you can't create a cube with more than 16 triangles

12:24 <J23k> 12 Ü

12:31 <J23k> Guest53  iirc peeps wrote something https://github.com/thehans/FunctionalOpenSCAD/blob/master/subdivision.scad

12:33 <J23k> (for spheres)

14:58 <Guest53> Very complex. Thank you pointing me to this github

17:00 <gbruno> [github] jbinvnt synchronize pull request #4675 (GSoC 2023: Create VertexStateManager) https://github.com/openscad/openscad/pull/4675

17:03 <joseph_> teepee: I just removed the commit in my PR that changed to a const reference in the off chance that it will cause CircleCI to start behaving again. Also, I have submitted my portion of the midterm review because I felt ready to answer all the questions. If there's anything you'd like to discuss before submitting your own form, please let me know

17:16 <teepee> joseph_: I don't think that has anything to do with the failure. it's quite puzzling

17:16 <teepee> it looks like the base docker image was pushed 2 days ago

17:16 <teepee> no idea how that happened and the build shows cmake is installed but it's not in that image

17:16 <teepee> plus there's build setup for the repo, with auto-build *disabled*!

17:17 <teepee> and there was no push to the repo since april

17:17 <teepee> so i don't know what's going on, but we did end up with a broken docker image for building the appimage

17:18 <joseph_> Hmm. Is it only my branch that is having this problem?

17:18 <teepee> I would expect all new builds will fail with the same issue as it's independent from the main openscad repo

17:18 <teepee> the docker stuff is in a separate repo

17:21 <teepee> but as mentioned earlier, the circle-ci builds are basically the source for the builds we ship on the website

17:22 <teepee> all the test runs are on github, so if those are green, code is like 99.9% fine

17:25 <teepee> as for gsoc, I think we have a good midterm state with the current PR, I'm planning on posting this today

17:26 <teepee> I want to look through the PR changes again, but with all the latest updates, it's a good first step. looking at the diffs yesterday it looked good. I want to have a look at the whole render classes again

17:27 <teepee> but in general I think it's good to merge

18:11 <guso78> Haha i am curious how IT will merge with my local texture stuff

20:00 <teepee> joseph_: aaaand posted :)

20:01 <teepee> guso78: I suspect it will conflict, but this PR is mostly moving things to a common place, so we don't have 3 times code duplication for the rendering

20:01 <teepee> so it probably will not automatically merge, but it should be relatively ok to update

20:02 <teepee> the big question is still out on how to actually handle things overall. I've seen quite some chatter in regard to transfering additional data through the manifold CSG process

21:35 <gbruno> [github] gsohler synchronize pull request #4703 (Store Points in PolySet as Indexed Points) https://github.com/openscad/openscad/pull/4703