| 5 Jul 2024 |
 ErichDonGubler | cargo clippy --workspace --all-features
warning: field `overrides` is never read
--> naga/src/compact/expressions.rs:6:9
|
4 | pub struct ExpressionTracer<'tracer> {
| ---------------- field in this struct
5 | pub constants: &'tracer Arena<crate::Constant>,
6 | pub overrides: &'tracer Arena<crate::Override>,
| ^^^^^^^^^
|
= note: `#[warn(dead_code)]` on by default
warning: field `locals` is never read
--> naga/src/front/wgsl/parse/ast.rs:145:9
|
115 | pub struct Function<'a> {
| -------- field in this struct
...
145 | pub locals: Arena<Local>,
| ^^^^^^
|
= note: `Function` has a derived impl for the trait `Debug`, but this is intentionally ignored during dead code analysis
warning: `naga` (lib) generated 2 warnings
| 01:51:11 |
| ErichDonGubler | * $ cargo clippy --workspace --all-features
warning: field `overrides` is never read
--> naga/src/compact/expressions.rs:6:9
|
4 | pub struct ExpressionTracer<'tracer> {
| ---------------- field in this struct
5 | pub constants: &'tracer Arena<crate::Constant>,
6 | pub overrides: &'tracer Arena<crate::Override>,
| ^^^^^^^^^
|
= note: `#[warn(dead_code)]` on by default
warning: field `locals` is never read
--> naga/src/front/wgsl/parse/ast.rs:145:9
|
115 | pub struct Function<'a> {
| -------- field in this struct
...
145 | pub locals: Arena<Local>,
| ^^^^^^
|
= note: `Function` has a derived impl for the trait `Debug`, but this is intentionally ignored during dead code analysis
warning: `naga` (lib) generated 2 warnings
| 01:51:16 |
| ErichDonGubler | Is there any reason we shouldn't just delete these fields? I don't think they're holding anything alive, but I feel like I might be missing something important. | 01:52:42 |
| ErichDonGubler | I hesitate in the case of Function::locals especially because it has quite a bit of documentation, and I'm not sure where it would migrate to. | 01:53:16 |
| ErichDonGubler | I see they're marked as pub, but they aren't actually exposed outside the crate. Are they supposed to be exposed to external users via Debug? | 01:57:16 |
 jimb | Almost nothing in compact should be public, outside the compact function itself. | 02:53:18 |
| jimb | It's supposed to be a self-contained thing. Everything should be at most pub(super) | 02:53:39 |
| ErichDonGubler | Yeah, I didn't think so. Sooo…this means we can delete unused stuff with prejudice, right? 😀 | 04:48:40 |
| 6 Jul 2024 |
|  thequstiner joined the room. | 05:32:09 |
| thequstiner | Hello are there any goo dexamples on how to covert a spv shader to glsl vertex/fragment shaders programmically? | 05:34:05 |
 wumpf | In reply to @thequstiner:4d2.org
Hello are there any goo dexamples on how to covert a spv shader to glsl vertex/fragment shaders programmically? check the naga-cli tool https://github.com/gfx-rs/wgpu/blob/25bc704e3515fdfd191bd414b6fa35e28f58d366/naga/README.md#conversion-tool | 08:16:22 |
| wumpf | had an idea how to avoid building msl-out on Windows / hlsl-out on Linux/Mac when a user is only after building wgpu-hal while not affecting the usecase of explicitely wanting to emit msl/hlsl on any platform
https://github.com/gfx-rs/wgpu/pull/5919
Was quite surprised this saved of 1.4s of the overall wgpu build times on my Windows machine!
The solution is a lil bit clunky tho, curious to hear what Naga folks thing :) | 10:14:20 |
| wumpf | * had an idea how to avoid building msl-out on Windows / hlsl-out on Linux/Mac when a user is only after building wgpu-hal while not affecting the usecase of explicitely wanting to emit msl/hlsl on any platform
https://github.com/gfx-rs/wgpu/pull/5919
Was quite surprised this saved of 1.4s of the overall wgpu build times on my Windows machine!
The solution is a lil bit clunky tho, curious to hear what Naga folks think :) | 10:14:29 |
| 7 Jul 2024 |
|  @ashley_:matrix.org changed their profile picture. | 11:22:52 |
| @ashley_:matrix.org left the room. | 11:24:31 |
| 9 Jul 2024 |
|  Yu-Wei Wu joined the room. | 09:32:44 |
| 10 Jul 2024 |
|  Ivo Leitão joined the room. | 22:15:50 |
| 14 Jul 2024 |
|  joshprk joined the room. | 07:58:53 |
 Natalie Cuthbert | Does naga build for wasm32-wasi? | 12:23:12 |
 cwfitzgerald | In reply to @ncthbrt:matrix.org
Does naga build for wasm32-wasi? It's io-less rust, so likely | 20:43:37 |
| 16 Jul 2024 |
|  unarelith joined the room. | 10:18:47 |
| 18 Jul 2024 |
| unarelith | Hello, I've been trying to use Naga but it fails on most compute shaders
Here is the error:
Function [1] 'main' is invalid:
The value [209] can not be stored
| 12:44:57 |
| unarelith | Download tmp.spvasm | 12:45:03 |
| unarelith | Do anyone understand this error? | 12:45:51 |
| unarelith | * Does anyone understand this error? | 12:46:05 |
| unarelith | * Hello, I've been trying to use Naga but it fails on most compute shaders
Here is the error:
Function [1] 'main' is invalid:
The value [209] can not be stored
Generating wgsl output requires validation to succeed, and it failed in a previous step
| 12:47:46 |
| 19 Jul 2024 |
 jasmine1717 | PartialEq is no longer implemented for naga types, which breaks naga_oil :/. Not sure how to work around this, it was used for fetch_if_or_append and fetch_or_append a lot. | 05:46:03 |
 teoxoy | Would using `.append()` instead cause issues? I think it makes sense that the types that don't go in the `UniqueArena` don't implement `PartialEq`. | 08:21:01 |
| teoxoy | Please open an issue though, it's hard to keep track of issues in multiple places. | 08:22:19 |
| 22 Jul 2024 |
|  @gschian0:matrix.org left the room. | 14:23:04 |
