14:55 <Myster-D> Hi. Just a quick question. I use recent dev snapshots on Windows, and I had Fontconfig error messages stating that the default font config file was missing. I checked and yes, no "fonts.conf" in the fonts dir. Is it because I'm not using a release distribution? Did I miss something?

14:57 <Myster-D> To be precise, I only get these messages when I'm using openscad from the command line.

15:07 <InPhase> Can you show which exact error message you're getting? https://bpa.st if it's longer than one line.

15:09 <Myster-D> Fontconfig error: Cannot load default config file: No such file: (null)

15:09 <Myster-D> I could remove this error by copying the fons.conf file from an old 2021.01 official release to the fonts directory

15:13 <Myster-D> OS is Windows 10 21H2, OpenScad release is 2023.02.23 and 2023.02.15

15:15 <InPhase> Well there's a difference in the Makefile between release and the master branch where there used to be an install_fonts, and now that does not exist anymore.

15:15 <InPhase> Not sure if that used to be the thing triggered or not when doing a release build.

15:15 <InPhase> That would require checking up the process stream. I started by hunting for what touches the fonts.conf

15:17 <Myster-D> hum I see. Actually the error doesn't seem to prevent the rendering and the exporting to stl files anyway, but I was curious as to why this was happening.

15:18 <Myster-D> thanks for YOUR HELP

16:31 <gbruno> [github] ochafik opened pull request #4533 (Use Manifold for much faster CSG operations (WIP)) https://github.com/openscad/openscad/pull/4533

16:31 <teepee> O_o

16:31 <teepee> guso78[m]: look!!!

16:37 <J23> i already have projects that can't run on the official release as they render hours  (as i was told)

16:38 <teepee> if the claim of the Manifold name also proves true it would be not only a massive improvement in time but also in stable result

16:41 <teepee> and considering the author there's good reason to believe that claim

16:53 <gbruno> [github] ochafik edited pull request #4533 (Use Manifold for much faster CSG operations (WIP)) https://github.com/openscad/openscad/pull/4533

16:55 <JordanBrown[m]> InPhase: Myster-D : the (null) is never a good sign.

17:08 <gbruno> [github] ochafik synchronize pull request #4533 (Use Manifold for much faster CSG operations (WIP)) https://github.com/openscad/openscad/pull/4533

18:13 <gbruno> [github] ochafik edited pull request #4533 (Use Manifold for much faster CSG operations (WIP)) https://github.com/openscad/openscad/pull/4533

19:21 <teepee> what a world... very sad day yesterday, happy day of many returns today :)

20:17 <JordanBrown[m]> teepee: what news did I miss?

20:20 <teepee> the one with general interest would be the latest PR both by content and by person sending it

20:27 <InPhase> teepee: He's like a render improvement machine.

20:31 <teepee> taking on the complicated stuff without fear

20:32 <teepee> I suppose we need to decide on GSoC soon(-ish)

20:33 <InPhase> Are we back around the loop to that time again?

20:33 <teepee> BRL-CAD is in and I did activate last years tasks for now, but question is if we want to take part this year again

20:33 <InPhase> Time flies quickly.

20:34 <teepee> yes, chocolate rabbits all over ALDI already ;-)

20:35 <teepee> FreeCAD got convinced to join on their own at FOSDEM it seems, after talking to the GSoC people, maybe because there was no communication upfront

20:36 <teepee> I did see Sean talking about applying though via their chat thingy - which I forgot the name of

20:37 <teepee> zulip

21:04 <guso7890> Test123

21:04 <guso7890> I have Seen the PR About the manifold. I am wondering where IT helps Out already

21:05 <guso7890> In Chip Design there exists a Software now owned by Siemens which ist Falles calibre.

21:06 <guso7890> IT can drc check Chips really fast by doing Lots of 2d Unions, differences and intersections.

21:08 <guso7890> They Use a Secret Algorithm which ist owned by Linus Torvalds and IT makes the Stuff really fast compared to competitors. At least His Name appears when starting calibre.

21:09 <teepee> secret algorithm owned by linus? that does not sound right to me

21:09 <guso7890> I am confident that this Algorithm.could also applied to 3d Space instead of 2d space

21:10 <teepee> the go-to 2d stuff is Clipper

21:10 <teepee> at least in the open source world

21:10 <guso7890> The Mentor people at least kept IT Secret at least

21:11 <teepee> right, that they can do I suppose but I would assume it would have been published somewhere else

21:12 <teepee> well, he did work for one of the chip companies, so I guess it's possible

21:12 <teepee> transmeta

21:12 <InPhase> guso7890: What?

21:12 <InPhase> guso7890: Did someone slip something in your coffee this morning? :)

21:13 <guso7890> In Phase, His Name ist really mentioned when starting calibre software

21:15 <guso7890> I am really considering asking hin If the Algorithm , which makes calibre fast is available.

21:16 <teepee> nothing wrong with asking I suppose, but if he would be able to release it he probably would have done so

21:17 <guso7890> I am wondering which Core Algorithm manifold ist based on

21:18 <guso7890> Teepee, even He was Not allowed to Release IT, even telling Side facts would Help to find Out how He is doing

22:23 <Guest35> Hello everyone! I am getting Bing AI to generate an OpenSCAD model, it seems to have it almost renderable but the last few lines are giving me errors, is anyone able to help me get it working so I can see what Bings 3D modeling skills are like?

22:24 <Guest35> The error is on line 45, if anyone could get this rendering id be facinated to see how accurate it is https://pastebin.com/z4NqdHFi

22:31 <InPhase> I tried this with ChatGPT already, and it was impressively bad. ;)

22:31 <InPhase> The training process has no sense for visual or spatial information.

22:31 <Guest35> :')  yea but im curious to see in which ways it gets it wrong

22:31 <Guest35> Bing will likely be a lot more accurate

22:32 <InPhase> Line 45 just has a return that needs to be deleted.

22:32 <InPhase> glass-floor_ radius to glass-floor_radius

22:33 <InPhase> Although it has its variables all confused. There is a glass_floor_radius defined above, but not a glass or a floor_radius, so the _ cannot be a -

22:34 <Guest35> :'''') its awful, its just two cylinders. Thanks for the help!

22:34 <Guest35> glad i got it rendered, ill play more with it see if i can get it to do variables correctly

22:36 <InPhase> It seems to have made this mistake over and over again of the variables with dashes.

22:38 <InPhase> https://bpa.st/ZKFFM Here it is with all the typos fixed.

22:39 <InPhase> At least the obvious ones.

22:39 <InPhase> It is, of course, spatial nonsense. :)

22:39 <Scopeuk> guso78 if my experience with mentor is anything to go by the speed up comes from removing the sleep statements, that appears to be the questa Vs models I'm performance bump

22:40 <Scopeuk> Models I'm, silly auto correct

22:40 <InPhase> Guest35: It looks like 5 cylinders and a barely visible sliver of something.

22:42 <InPhase> Guest35: I was optimistic when I first tried it, but it would really take a different form of input processing to actually learn OpenSCAD usage. If you tried to learn it as a human without ever seeing the output of code, it would be extremely challenging. While it is technically possible, as far as I can tell none of the information out there is sufficient for this.

22:43 <Guest35> Lol thanks for everyones input!

22:44 <Guest35> It managed to make a decent human on the first try, thats pretty impressive, but yea clearly its not very good. https://imgur.com/a/pWcExNR

22:44 <Guest35> Special thanks to InPhase, thats awesome lol, just a bunch of cylinders everywhere

22:44 <InPhase> Well that's not bad. But maybe an example of that already existed.

22:44 <Guest35> yea you're probably right

22:51 <teepee> from what I've seen so far, I'd rate AI capabilities regarding openscad as useless

22:51 <teepee> text matching does not work for CAD

22:52 <Guest35> Not yet!:')  This is what it thinks a shrimp looks like, sorta close eh?https://imgur.com/a/61q7sXT

22:55 <Guest35> I think it could be useful to grab dimensions of existing objects and their corresponding RGB values, but after that yea useless

22:55 <Guest35> because it seemed to get dimensions and colours of CN tower right, jsut failed to render them

22:58 <teepee> those are easy to google I guess :)