TheAssassin has quit [Remote host closed the connection]
TheAssassin has joined #openscad
<sparsh-n>
A bit late, but I periodically look at the irc logs and noticed there was discussion around 2 days ago about prospective contributors/mentees not being too active in the IRC chat. After the small QKeySequence fix PR I haven't found too much time to work on OpenSCAD lately. Once my uni exams are over, I hope to be more involved :)
mmu_man has quit [Ping timeout: 276 seconds]
sparsh-n has quit [Quit: Client closed]
kintel has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
stealth_ has quit [Quit: Leaving]
mmu_man has joined #openscad
ubitux has left #openscad [WeeChat 4.6.1]
Guest31 has joined #openscad
Guest31 has quit [Client Quit]
teepee_ has joined #openscad
teepee has quit [Ping timeout: 264 seconds]
teepee_ is now known as teepee
mmu_man has quit [Ping timeout: 252 seconds]
mmu_man has joined #openscad
mmu_man has quit [Ping timeout: 248 seconds]
mmu_man has joined #openscad
kintel_ has joined #openscad
<kintel_>
sparsh-n Thanks for reaching out, and good to know you keep an eye on IRC logs :)
<kintel_>
In terms of Windows-related topics, here are two small issues which could be good to demonstrate ability to understand and address Windows-specific issues. Windows-specific filesystem-related issues tend to be very common:
<kintel_>
In terms of demonstrating knowledge and ability to build knowledge of the Windows ecosystem/platform itself, that's a bit harder
<teepee>
plus we need to discuss that there's "enough" code going on, it's google summer of *code* after all
<kintel_>
There was some discussion here earlier on the merits of MSYS2 vs. other build environments vs. WSL. Any insight you can offer that helps people understand what it takes to Build-run-debug on Windows vs. distribute binaries, and drawbacks/benefits etc., would greatly improve trust :) Not asking you to do crazy amounts of work, but pick something and run with it
<kintel_>
Yeah, ability to write code is also something that should be demonstrated. While one-liner fixes can be pretty hard to pin down, it's still a one-liner :) For the Windows project, the amount of code written _may_ end up being mostly scripting, cmake, configuration etc., but demonstrating the ability to engage with such would be great.
<kintel_>
Same goes for the other proposed gsoc projects - it's about demonstrating skills relevant to the project + ability to clearly communicate ideas and deal with offline/async communication like IRC/GitHub/email efficiently
mmu_man has quit [Ping timeout: 265 seconds]
mmu_man has joined #openscad
<kintel_>
On another topic: I think the next step in the Manifold saga is how to deal with caching. We currently have caching enabled, but it has problems: 1) We don't have a way of estimating cache size for Manifold trees and 2) Caching every sub-result of Manifold forces us to evaluate each operation of our own CSG tree which circumvents Manifolds optimization algorithms.
<kintel_>
Not quite sure how to address this. The naive idea is to estimate cost prior to operations based on heuristics and only cache expensive nodes (possibly with a validation post-CSG that our heuristics were correct), but that still early ideas stage.
<kintel_>
Ideally, this would be built into Manifold, but it feels like that's pretty far away still.
<kintel_>
Ideas are welcome.
<kintel_>
pca006132 If you have any further ideas, please share too :)
kintel_ has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<pca006132>
I think it will be better to cache only operations that are expensive. We probably need an estimator for this, maybe depending on the size estimate for the operands.
<pca006132>
from a PL perspective, it may be possible to estimate where the changes can occur and use that information to determine what to cache
<pca006132>
but that is definitely non-trivial, and only works with animation
kintel has joined #openscad
<kintel>
pca006132 My thinking so far is that the stuff we send to Manifold is pretty limited; it's essentially pure CSG ops
<kintel>
..plus hull, and some other minor ops
<kintel>
..but a simple OpenSCAD construct, like minkowski with two operands, can turn in to a quite significant set of manifold operations
<kintel>
..so if we can estimate the size/cost of such a set of operands, that could help determine which OpenSCAD subtrees should be dispatched to Manifold as a unit, then cache that result
<kintel>
Perhaps building some helpers to map between OpenSCAD node trees and Manifold node trees would be helpful. I'll try to tinker with this when I have time.
<kintel>
Not sure yet if I consider this a blocker for making Manifold default or not. But I think it's definitely a release blocker
kintel has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
sinned69158 has quit [Quit: Poof goes the quack]
sinned691586 has joined #openscad
stealth_ has joined #openscad
averdow has joined #openscad
<averdow>
about how much RAM is required to compile from source? It just chewed through 16GB and I'm not sure if I should throw some swap space at it or if there's a leak
kintel has joined #openscad
<kintel>
averdow For single-threaded compilation, I believe it should stay under 2GB, but if you massively parallelize, it will scale accordingly
feep has quit [Ping timeout: 252 seconds]
TheAssassin has quit [Remote host closed the connection]
TheAssassin has joined #openscad
califax has quit [Remote host closed the connection]
teepee has quit [Remote host closed the connection]
teepee has joined #openscad
califax has joined #openscad
<teepee>
averdow: using clang instead of gcc might also help, on my 32gb notebook I usually limit to -j4
averdow has quit [Ping timeout: 240 seconds]
feep has joined #openscad
aiyion has quit [Remote host closed the connection]
aiyion has joined #openscad
aiyion has quit [Remote host closed the connection]
aiyion has joined #openscad
sparsh-n has joined #openscad
<sparsh-n>
I was reading a blog post about different ways to represent rotations, and was wondering if OpenSCAD already has support for Quaternion based rotations? I'm seeing only regular rotation matrices so I'm guessing it isn't currently supported. A relevant link I found for this: https://15462.courses.cs.cmu.edu/spring2023/lecture/3drot
GNUmoon has quit [Remote host closed the connection]
<sparsh-n>
I noticed there was some short discussion about the use case of quaternions for rotations earlier in march from the irc chat, so one counterexample mentioned for using quaternions from 2 years ago sent here is https://marctenbosch.com/quaternions/
sparsh-n has quit [Client Quit]
GNUmoon has joined #openscad
kintel has quit [Quit: My Mac has gone to sleep. ZZZzzz…]