<kintel>
joseph_: OpenGL modernizing could be a decent project. There is probably some homework to do; determine how to deal with the existing open PRs, understanding the scope of the problem (e.g. how many legacy calls are left), decide how to deal with OpenCSG, whether to support WASM, GLES compatibility.
<kintel>
I think the main challenge with this project would be that it's mostly about refactoring and less about writing new features. Just make sure this is in-line with your wishes. This does, however, model real-world development pretty well, and could probably benefit from a more agile development workflow, with frequent Pull Requests and testable code.
GNUmoon has quit [Remote host closed the connection]
<mack>
Hi all. I downloaded and started my first use of openscad today by attempting to modify someone else's script. I was going to ask for guidance about the trouble I'm having with the threads library, but hit the Logs link first. I think I've found the problem -- there are at least 3, possibly a myriad, versions of threads.scad. I think I might have found the one with the module used in the script.
<mack>
I did indeed find the correct threads.scad. File modified and send to printer.
<InPhase>
Glad we did such a great job of helping you! ;)
<InPhase>
Also, welcome.
<mack>
I did find your threads lib, InPhase :) It wasn't the one, though.
<mack>
well, I assume. We'll see when the job finishes.
<InPhase>
Yeah. Sorry for contributing to the overlapping names. ;)
<InPhase>
Searching the called module name can help.
<mack>
right. After I realized there wasn't a "standard" one as the main web site led me to believe, that's what I did.
<mack>
it also doesn't help that several people have copied and/or modified files and put them on their github without attribution
<kintel>
Problem solved: alias openscad="mv ./openscad.exe ./offscreen.exe && ./offscreen.exe". Now, back to coding
<InPhase>
kintel: lol
<mack>
and you did help me -- the logs in the topic is what gave me the lightbulb moment.
<InPhase>
kintel: After that long on it, that sounds like a perfectly valid approach.
<kintel>
At least it allows me to run tests over ssh. For all other purposes: "don't ssh into Windows" is the advice until someone can reproduce this magical configuration :)
<lf94>
I have not forgotten about openscad + libfive
<lf94>
just been busy
<guso78>
did not want to conflict with anybody. this is python approach
<guso78>
(not using oracle approach)
<guso78>
building a tree instead. i believe the tree evaluation time needs to be super fast and it has the option, that the tree program can eventually be uploaded the the graphics card
Guest11 has joined #openscad
<lf94>
build da tree
<lf94>
oh, haha!
<lf94>
i didnt realize what i said was on-topic lol
<lf94>
very cool man !
<lf94>
that's just like my node-libfive bindings
<lf94>
except im seeing openscad here :o
<lf94>
guso78: yes, i think you're doing it right, but you lose some power of openscad, right?
<lf94>
i.e. you cant use the openscad language
<lf94>
this is incredibly cool regardless, awesome work
<lf94>
are you retired or have a lot of free time or just super focused? :D
Guest11 has quit [Client Quit]
<guso78>
no openscad power is still there. I just using python interpreter instead of scad interpreter
<guso78>
using openscad language is probably possible, but get it running in python was WAY easier for me.
<guso78>
still need to tidy up quite a bit but thats streight forward, i believe
<lf94>
ah
<lf94>
would you like to port one of my examples as a test?
<guso78>
you are talking about an SDF program right ?
<guso78>
so in your side you can display libfive in the web browser ?
<lf94>
no, it's node.js based, so outside of browser
<lf94>
i didnt want a browser based one on purpose
<lf94>
i wanted a REPL basically :D
<lf94>
so now i can use node + my editor + libfive to generate models nicely. only thing im missing is a view lol. I have to use an stl viewer for now.
<guso78>
ahh, right now it did not base on the libfive provided libs but i believe now its rather easy to adopt
<lf94>
you mean for yours?
<guso78>
yep , for mine
<lf94>
ah ok
<lf94>
i actually dont recommend it
<lf94>
reason: mkeeter is working on fidget
<lf94>
so libfive-stdlib work will not be portable
<lf94>
re-implement the sdfs in python with libfive primitives.
<lf94>
you'll thank me
<lf94>
i made a mistake not doing this
<guso78>
i think next step is to use libfive to use nice rounded joints for objects
<guso78>
thank you for the hint!
<lf94>
look at cube.smooth
<lf94>
ez pz :)
<guso78>
haha
<lf94>
this i took from curv3d, which took it from iquilez
<guso78>
actually i have two data types in python-openscad. one is PyOpenSCAD and one is PyLibFive
<lf94>
You may even want to stop work on libfive and immediately switch to fidget
<guso78>
last one just establishes the evaluation tree
<lf94>
the speed improvements are insane
<lf94>
he gets GPU comparable performance
<lf94>
(on GPU they only do interpreted, on CPU they now do JIT)
<guso78>
Ahh Fidget is the successor of LIbfive.
<lf94>
it's amazing. you'll see ive been commenting in places there
<lf94>
I convinced them to add a mesher very recently ;)
<lf94>
so it's on par now with libfive functionality essentially
<guso78>
i think the infrastructure is already there, its just a replacement of the engine
<lf94>
yea exactly
<guso78>
i will push my branch soon. just let me know in case you are interested
<lf94>
im good, you should push as soon as you need to :)
<guso78>
one advantage of push is to be safe against hard disc crash :b
<guso78>
(now i know why i do, just had another .git repo crash, but was able to recover)
guso78 has quit [Quit: Client closed]
fedorafan has quit [Ping timeout: 248 seconds]
fedorafan has joined #openscad
WayneS has joined #openscad
L29Ah has joined #openscad
fedorafan has quit [Ping timeout: 252 seconds]
use-value has quit [Quit: use-value]
fedorafan has joined #openscad
<lf94>
what kind of computer are you working with? ive never had a hd die on me ever, in my entire existence lol
<WayneS>
Some very cheap SSDs (usually those with low capacity) have extremely low write endurance and will break after just a few months.
castaway has quit [Ping timeout: 248 seconds]
<WayneS>
Anyway, OpenSCAD is very cool! So far I did not really understand traditional modeling software, but OpenSCAD clicked with me right from the start. Thanks to all the developers! <3
castaway has joined #openscad
<guso78[m]>
The hard disc IS Not the Problem. Rather the Problem IS that i use to hardly shut down my virtual Box. Still hard to believe that IT die Not sync all git Files after Minutes...
<kintel>
"...openscad/submodules/manifold/src/third_party/thrust/thrust/detail/type_traits.h:147:10: error: builtin __has_trivial_constructor is deprecated; use __is_trivially_constructible instead [clang-diagnostic-deprecated-builtins]"
<kintel>
guso78 I believe shaderinfo is a giant hack
<guso78>
ok. thank you
<guso78>
can you briefly explain what is shaderinfo is meant to be (it contains face and edge colors)
<guso78>
doesn't make shader info also make sense in F6 render mode?
<kintel>
I seem to remember that it was initially used to render wireframe in preview mode
<kintel>
..as it needs a different technique
<guso78>
ok. its a good method to destinguish between F5 and F6 render mode :)
<kintel>
Yeah, ideally, the mechanism for passing any info like that should be the same for both modes, as there could be good reasons for passing shader parameters also in F6 mode
<guso78>
actually i'd love to see shaderinfo ALWAYS used . then i could put texture info there as well and textures would also display with F6
<guso78>
will analyse the PR for myself.
<kintel>
In the end, "shaderinfo" is really just some information which should follow the mesh, like a material definition
<kintel>
Managing such things has been done so many times, which is why I've been wanting to use some sort of rendering library instead of inventing our own
<kintel>
..but to get there, we may need to refactor a lot of the rendering stuff to make these concepts cleaner in our own codebase
<guso78>
i think first goal could be to make sure, that shaderinfo is always not NULL. (because also vertex shader handles are constant). so code would not have to check for NULL ...
<guso78>
BTW today i managed to combine my local branch with libfive. i have merged klein bottle with a dish :)
<guso78>
probably will use successor "fidget" soon
aiyion has quit [Remote host closed the connection]
vsellier has quit [Ping timeout: 240 seconds]
vsellier has joined #openscad
aiyion has joined #openscad
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
guso78 has quit [Ping timeout: 260 seconds]
omegatron has joined #openscad
<lf94>
sweet :)
lkcl has quit [Read error: Connection reset by peer]