03:24 <joseph_> teepee InPhase: Before I log off for the day, I wanted to let you know that I've added the first part of my report from tracing the render processes to my dev log repo (https://github.com/jbinvnt/gsoc-2022-dev-log). At the time of writing, I outlined a couple things I still need to investigate further (at this point I have begun those, but not organized my findings into a report yet). I'd value your feedback about whether

03:24 <joseph_> exploring too much further in all these areas would be a good use of time, although as previously discussed, certainly the glUseProgram call depending on showedges will be relevant

03:27 <joseph_> Then we can discuss where my incremental pull requests should begin, in terms of which functions I start editing and also whether my changes should be on their own base branch off of master (or whether changing the behavior of showedges should be addressed in its own, "regular" issue branch)

03:28 <joseph_> Finally, I discovered an issue with my dev environment that causes an attempt to copy the built executable out of the container before it has started. I will fix that tomorrow

03:38 <InPhase> https://github.com/openscad/openscad/blob/6aae7963498675dfb5bedaaef2cb8e56bdcddcab/src/glview/preview/OpenCSGRenderer.cc#L391

03:38 <InPhase> joseph_: Not your code of course, but you'll be touching this area. Is that a bug on 392 with that error check?

03:38 <InPhase> It looks like a classic case of writing an if without braces and then slapping another call after it.

03:42 <InPhase> teepee: You win the git blame game on that one. :) ^

03:44 <InPhase> joseph_: While hunting through git blame, I happened to notice that line 479 has another shader call with different conditions. Did you obtain an understanding of how that one is triggered?

03:48 <InPhase> joseph_: "Note made in post" links to a 404. Not sure if there was to be anything important there.

04:00 <InPhase> joseph_: As for Experimental on the VxO, that's nothing to be concerned about. Project protocol is to mark almost every change we can that could break core function in some manner as experimental and stick it under an option, and let people test it out for a bit. Then we fade away the experimental labels when things seem to work okay.

04:01 <InPhase> joseph_: So your whole project will also be marked experimental at first insertion.

04:02 <InPhase> joseph_: It will need its own configuration option set up for it, so that it can be turned on and off, and should be handled like the other experimental options.

04:44 <InPhase> joseph_: On branch strategies, I think centralizing it in your own single PR branch makes some good sense here. It's going to be a common need to touch existing areas with some tweaks to wedge this in, and showedges is perhaps just the first spot of many.

04:47 <InPhase> joseph_: On editing strategy I'd imagine something like demonstrate to yourself that you can actually change the shader appearance (sounds like you did that), then probably next setup the experimental configuration option for your work so that you can keep all the functional part of what you do within that configuration option. It's sort of like setting up a sandbox that behavioral changes go into, so

04:47 <InPhase> you build the sandbox first.

04:49 <InPhase> joseph_: Remember to periodically test your PR with the option on and off, to make sure it still works off. :) It's easy to make a breakage mistake and trap yourself in a tunnel only testing the new part. So raise awareness of it and keep periodically testing without it on.

07:36 <peepsalot> that craiyon site is finally working for me. it's cool but too bad that all generated images of people end up looking like Leatherface

08:30 <J1A8468> yeah  i think they just have restricted the recursion depth of the GAN,  -  just saw someone putting some bad looking AI generated Face into the faceApp beauty filter .. which really does the trick. (https://www.faceapp.com/ )

08:38 <Scopeuk> so thats an app that takes a photo and turns it into a "photo realistic" render from 10 years ago?

08:43 <J1A8468> it is just trained for faces ..   the  dall-e2 ( openai.com ) is not open, so i assume the crAIyon (dall e mini) is using much less ressources for generation

10:12 <muesli> Which version is this available in? https://en.wikibooks.org/wiki/OpenSCAD_User_Manual/WIP/Text_Metrics

10:23 <teepee> muesli: dev snapshot

10:24 <teepee> and needs to be enabled in preferences -> features

11:19 <muesli> teepee: thank you!