| 3 Apr 2023 |
 cwfitzgerald | Both of my jobs were/are on the web, my one wgpu contacting thing was fit the web | 14:47:30 |
| cwfitzgerald | * Both of my jobs were/are on the web, my one wgpu contacting thing was for the web | 14:47:39 |
 jasperrlz | Gotcha. It's possible I'm not good at Rust to use it without feeling really constrained | 14:48:50 |
| jasperrlz | * Gotcha. It's possible I'm not good enough at Rust to use it without feeling really constrained | 14:48:58 |
| jasperrlz | but I feel like I've had to work around wasm-bindgen a whole lot | 14:49:06 |
| jasperrlz | maybe it's a lot nicer if you go full-on Rust rather than trying to call Rust WASM from JS | 14:49:36 |
| cwfitzgerald | In reply to @jasperrlz:matrix.org
but I feel like I've had to work around wasm-bindgen a whole lot Is there anything in particular that's you've been hitting a lot? | 14:51:47 |
| cwfitzgerald | Both of my jobs had a quite narrow interface (outside of WebGL), the ui was all traditional we dev (react) and had a "compute engine" in rust that did the heavy lifting | 14:52:58 |
| jasperrlz | The inability to send back structured data containing Vec's, demanding I write these weird little accessor structs every time | 14:54:13 |
| jasperrlz | https://github.com/magcius/noclip.website/blob/master/rust/src/halo/wasm.rs | 14:54:35 |
| jasperrlz | There's probably a better way to write that file, and I'd love to learn it, but the documentation has not been kind to me | 14:55:21 |
| 4 Apr 2023 |
 fl33tw00d | 
Download after.png | 17:27:39 |
| fl33tw00d | Very pleased with the profiler I wrote on top of my draft PR: https://github.com/gfx-rs/wgpu/pull/3636 | 17:28:25 |
| 10 Apr 2023 |
 kvark | new github search is so good, really making a difference for just reading other peoples code | 02:31:08 |
| fl33tw00d | In reply to @kvark:matrix.org
new github search is so good, really making a difference for just reading other peoples code This blog post explaining how they did it is cool: https://github.blog/2023-02-06-the-technology-behind-githubs-new-code-search/ | 08:55:25 |
 @heavyrain266:mozilla.org | Hmm, found some fancy adventages related to USD in Rust instead of C++. Now there is much better schema system that uses proc-macros and not jinja2 to generate C++ 😆 | 09:42:17 |
| @heavyrain266:mozilla.org | Lack of compositing engine doesn't hurt that much, but this way it will be entirely useless for tooling | 09:43:46 |
| @heavyrain266:mozilla.org | And now also have to rewrite Pixar's Hydra renderer to get real-time playground for it | 09:46:33 |
| 11 Apr 2023 |
|  evie_kurt001 joined the room. | 21:50:29 |
| 16 Apr 2023 |
|  @ultrakatiz:anontier.nl left the room. | 20:31:08 |
| 18 Apr 2023 |
| kvark | One of the most annoying issues with C++ codebases - you need linting to reduce shooting yourself in the foot, but linting in C++ (looking at you, clang-tidy) is extremely slow, to the point it becomes a technical challenge to do it consistently on CI.
It's night and day in Rust when I can just call clippy whenever I please. | 03:58:24 |
| 19 Apr 2023 |
 scoopr | and thats assuming you get somewhat consistent ccdb json from your buildsystem to even run the clang based tooling in the first place | 07:46:57 |
| kvark | Our engineer had to patch clang upstream to make it work in the first place… | 15:15:27 |
 minraws | In reply to @scoopr:matrix.org
and thats assuming you get somewhat consistent ccdb json from your buildsystem to even run the clang based tooling in the first place Having worked on a C++ linter I can assure you it's probably not the tool's fault and ecosystem and the language leave little room for hope. | 17:17:43 |
| scoopr | Oh indeed, I’m not blaming clang tools, it’s probably on the more successful side of having a deep understanding of code and not being totally impossible to roll out to a random codebase.. | 17:32:06 |
| 22 Apr 2023 |
|  Benjamin Swerdlow joined the room. | 20:17:57 |
| 24 Apr 2023 |
| jasperrlz | https://github.com/KhronosGroup/Vulkan-Docs/issues/2112 this is one of the weirdest rabbitholes I've ever dug down. Wave ops are fun! | 02:23:48 |
| Benjamin Swerdlow set a profile picture. | 16:45:51 |
| 26 Apr 2023 |
| fl33tw00d | I wrote a blog post about why WebGPU + ML is cool: https://fleetwood.dev/posts/running-llms-in-the-browser | 12:46:59 |
