00:01 <gbruno> [github] MichaelAtOz edited issue #4429 (assigned variable from another variable is not shown on customizer) https://github.com/openscad/openscad/issues/4429

00:01 <joseph_> teepee kintel: Today I fixed the build dependency issues I was having. This allowed me to test some of the first bit of refactoring, which I pushed to a "gsoc-2023" branch in my fork. But I plan to rename that, considering that the goal this year is to have multiple, smaller PRs.

00:04 <joseph_> It will be useful to have a way of tracking all the smaller branches, for the purpose of making clear what is done for GSoC. How do you recommend I do that? For example, should each branch start with "gsoc-2023-"?

01:42 <kintel> joseph_ gsoc-2023 as a prefix makes sense as an easy-to-find way of locating PRs. Keeping a google doc or something as a log also makes sense. Also, if you collaborate with other on branches which happens not to be prefixed that's also good to mention as an outcome.

02:19 <gbruno> [github] kintel pushed 2 additions 1 modifications (Bump glib to 2.76.3 for macOS (#4663)) https://github.com/openscad/openscad/commit/f290a1e10c6cba143d0082b791336d676799fbc2

02:19 <gbruno> [github] kintel closed pull request #4663 (Bump glib to 2.76.3 for macOS) https://github.com/openscad/openscad/pull/4663

02:34 <gbruno> [github] kintel opened issue #4673 (Support building on macOS 13 and Xcode >= 14.3) https://github.com/openscad/openscad/issues/4673

02:36 <gbruno> [github] kintel edited issue #4673 (Support building on macOS 13 and Xcode >= 14.3) https://github.com/openscad/openscad/issues/4673

02:36 <gbruno> [github] kintel edited issue #4673 (Support building on macOS 13 and Xcode >= 14.3) https://github.com/openscad/openscad/issues/4673

04:31 <gbruno> [github] shayded-exe opened issue #4674 (UI font is tiny on nightly builds) https://github.com/openscad/openscad/issues/4674

17:31 <gbruno> [github] jbinvnt opened pull request #4675 (GSoC 2023: Create VertexStateManager) https://github.com/openscad/openscad/pull/4675

17:36 <joseph_> kintel teepee: The above pull request is intended to illustrate the approach I'm following for this refactoring. For now my VertexStateManager class is initialized with pointers to vertex_states and vertex_array, because logic elsewhere in OpenCSG/CGALRenderer still modifies those objects. Perhaps eventually though, that will be unncessary if the newly-added class can somehow take full ownership of them.

17:37 <teepee> joseph_: so who owns them now?

17:37 <lf94> did you use a borrow checker to check ownership

17:39 <joseph_> teepee: There are one or two exceptions, but usually vertex_array is a stack object inside the prepare function. The vertex_states (sometimes with a slightly different name) I think I have seen as a member variable but at least OpenCSGRenderer holds a smart pointer to it

17:42 <joseph_> What's happening is that my functions replace logic that would normally modify those objects just serially within a single Renderer function. So at least for now, I don't think it will get too complicated

17:43 <teepee> so for now the renderers own the data and the vertexstatemanager is a view layer on top?

17:43 <joseph_> VertexStateManager actually edits them. But yes

18:14 <teepee> sounds like a good first step, unify the interface which may help understanding the differences

19:03 <kintel> joseph_ Added a few comments. If you have plans for further refactoring which would invalidate my comments, let me know. In that case, writing up you plan in a doc/diagram or smth. would help.

19:33 <lf94> Makefile is so incredible

19:34 <lf94> As long as you stick to using it as a dependency graph creator

20:03 <InPhase> lf94: I use makefiles to make and deploy my LaTeX based resume converting it into many different versions and forms from the same source file, and it really fits this particular problem incredibly well. It's one of the few areas I use a manually edited Makefile regularly.

20:04 <lf94> I mean I'm using it for a production system to build all these microservices and to get some deps in the right places and it's soooo much nicer than any js-based shit

20:04 <lf94> webpack is f'ing pain

20:05 <InPhase> Like, make upload for posting it online requires the web pdf version and the html version, and the html version needs to be converted from the tex version. The web pdf version is built from the tex file using slightly different settings that removes my address and phone number. It's a wonderful little dependency graph.

20:34 <lf94> nice :)

21:47 <Scopeuk> That object class a depends on clas b with same name style is where basic make shines