<joseph_>
teepee peepsalot: Feel free to mention my username in any relevant discussion to send a push notification, otherwise I might miss it (in this case though it happened to not get buried by other messages). Anyway I was traveling over the weekend but last night and this morning I completed work on several issues. First, the errors that occurred with the lazy union tests were apparently not due to an edge case in the resulting
<joseph_>
polyset, as I had initially predicted. Instead it seemed to be that lazy union were the only ones that caused multiple polysets to be rendered, and I didn't yet have a conditional to prevent add_shader_data from being called repeatedly
<joseph_>
Second, I allowed wireframe mode to work again, which was previously causing another CI test to fail. It should be noted that my fix was to go back to the old-style OpenGL and polyhedrons only for wireframe, because a rewrite for shader-controlled edges is out of scope at this time
<joseph_>
To address peepsalot's question about colors for cut faces, in fact my changes do allow this to be supported with shaders (I didn't judge it to be optional because otherwise the CI tests would fail). My solution is flexible and allows a shader to read the property of whether a face is cut. The default shader just makes them green to fulfill the test cases, but other shaders could change reflectance etc. for cut faces only
epony has quit [Remote host closed the connection]
epony has joined #openscad
<joseph_>
In summary, the pipeline is passing on my pull request. For some reason, only on my machine fastcsg-cgalpng_polyhedron-tests fails, which I suspect might be due to a dependency mismatch. I can troubleshoot it later but that's not part of the GSoC work. teepee, what are your thoughts about priorities for the final week of work before submitting?
<teepee>
joseph_: as discussed earlier, I have asked Sean 2 weeks ago for setting up the project as 20 week project, but I did not get a confirmation
<teepee>
if you are ok to go with next week being final, I can check with him again and if it's not changed yet to keep that setup
ur5us has joined #openscad
<teepee>
if we can keep the cut information for a bit, that's nice. as discussed with peepsalot I would not be terribly fuzzed about it going away as we need to do that eventually
J1A84 has quit [Quit: Client closed]
<teepee>
not ideal to change this, but in the end while useful it still seems like a feature implemented "because we can" simply because CGAL does emit that information
J1A84 has joined #openscad
<gbruno>
[github] kwikius reopened issue #4351 (Feature request : Extend array syntax to allow indices to be generally accessible via '.' syntax by allowing optional names for elements) https://github.com/openscad/openscad/issues/4351
Guest74 has quit [Quit: Client closed]
kwikius has joined #openscad
<kwikius>
After thinking about the feature allowing array indices to be named, I think it doesnt intefere with the swizzling useage. Like many other things in openscad you can override the defaults. if you want to do ar = [z=1,x=20,barg = 3]; then you should be allowed to do so. lookup can search for the user defined version first.
<teepee>
but what's the argument of having it other then "it's possible with great difficulties" ?
<teepee>
the object/dict/whatever-it's-named data structure will still be needed
<kwikius>
Its great documentation. ... and the difficulties are overrated.
<teepee>
I can't agree to that looking at github issues and mailing list discussions for the past 8 years
<kwikius>
The array can use either names or indices. Can the object use indices ?
<teepee>
it's probably fine for people who are used to slightly strange computer languages
<kwikius>
Personally I find the swizzling bizarre. What would have been wrong with a function?
<teepee>
right now they can't, it would be possible to implement I suppose, not sure it's a good idea
<teepee>
it's like 96 functions or so?
<joseph_>
teepee: I can see in my GSoC dashboard that the final week has been updated to Oct 2 - Oct 9. As a side note I believe this means my coding has been extended to 16 weeks total (measured from June 13) instead of 20 weeks. By my assessment, at this point everything from my proposal is fulfilled except an error popup box if a selected shader does not compile/link. Then there is documentation/review/polishing of my work. Let me know
<joseph_>
if I've missed something
<kwikius>
not really swiz = swizzle(ar,"barg");
<teepee>
joseph_: so I guess it's not changed from the initial 16 weeks Sean did. but I'd be ok to keep it
<kwikius>
Is anyone using swizzle? Ive looke but I can't find any uses so afr
<teepee>
it's a pretty new feature, and specifically not in a release version
<kwikius>
Is it experimental. Cos it probably should be :)
<teepee>
no, I don't think it is
ur5us has quit [Ping timeout: 248 seconds]
<teepee>
joseph_: I'll build a new version with your latest changes. did you change the GUI part? I was wondering if maybe adding some sort of default shader selection would be nice. similar to the colorscheme feature
<teepee>
where it picks up files from a system and a user folder automatically. the "open file" is still nice for maximum flexibility
<kwikius>
So swizzle is experimental, but not marked as such?
<teepee>
it's not marked as experimental, so it's not experimental :)
<teepee>
hmpf, why is it not showing the CircleCI builds?
<joseph_>
teepee: The GUI has not changed since late Aug, it still uses an "open folder" arrangement. For the timeline I agree that the existing Oct 2 - Oct 9 final week will likely be enough time to finish those remaining tasks. I think there was a miscommunication before, because on Sep 12 my message was just affirming that the existing 4week extension was good, not requesting 4 additional weeks beyond that
<teepee>
joseph_: ah, ok. that's fine, I'll check with Sean then so we keep it at 16 weeks (hopefully he's not changing it before he sees that mail :)
<teepee>
what do you think about that direct selection feature? I guess at minimum it could look for the 2 files to be present, or it could use a simple json file as the code snippets do
<joseph_>
teepee: Thanks, and I understand that the way I phrased that previous message might have made it seem like I was asking for even more time. I agree it would be good to say that 16 weeks is enough, although if he does change it before seeing your newest message I don't believe there is any downside
<kwikius>
So look at at this way. By being able to name array elements you can override 96 functions. Anyway got to go... bye!
<teepee>
joseph_: only downside would be later end which also means later stipend
<joseph_>
teepee: Good point, so let's just try confirming it won't get extended further. For the direct selection feature, I'd have to take some time to review how the color schemes are handled and assess how easy that is to translate to the shader selection. Unless you have a strong opinion that it should be included in GSoC, I'd prefer this to be something I investigate after the end (when I am just a regular volunteer contributor). If
<joseph_>
I'm correctly understanding what you suggest, it is a visual improvement that wasn't specifically in scope of my original proposal
<teepee>
ok, mail sent
<joseph_>
teepee: However, for specifically the part about verifying the presence of the frag/vert shader files, that would fall under error checking which was in my proposal
<joseph_>
So that would be included in the part where I would check to make sure that the shader compiles properly and create an error message if it does not
<teepee>
right, so adding more checks and clear error messages for the "open file" way?
<teepee>
that would be good, yes
ur5us has joined #openscad
<teepee>
testing with a couple of models looks good so far, not all the shaders seem to work, so that detailed error checking and robust behavior is a good conclusion
<teepee>
maybe also some cleanup of the example shaders giving them good names. I really like the "fakeGlints" one, some others are probably not that useful for end user
teepee_ has joined #openscad
teepee has quit [Ping timeout: 258 seconds]
teepee_ is now known as teepee
kwikius has joined #openscad
kwikius has quit [Client Quit]
jjs has quit [Ping timeout: 244 seconds]
ToAruShiroiNeko has quit [Remote host closed the connection]
ToAruShiroiNeko has joined #openscad
Non-ICE has quit [Quit: Screw you guys! I'm going home!]
la1yv_j has quit [Remote host closed the connection]
la1yv_j has joined #openscad
la1yv_j has quit [Remote host closed the connection]
la1yv_j has joined #openscad
<gbruno>
[github] kwikius edited issue #4351 (Feature request : Extend array syntax to allow indices to be generally accessible via '.' syntax by allowing optional names for elements) https://github.com/openscad/openscad/issues/4351