| 6 Oct 2022 |
 jinleili | Latest wasm-pack was updated 4 months ago, cannot used to compile wgpu 0.14 | 05:47:58 |
 cwfitzgerald (wgpu, rend3) | ono, what'd we do | 05:48:44 |
| jinleili | Need wasm-pack updates, we can't do much. now can only use wasm-bindgen directly. | 05:54:54 |
| cwfitzgerald (wgpu, rend3) | is there a bug? | 06:18:03 |
| jinleili | It could be my own problem, and I need to take a closer look. | 06:44:49 |
 maxammann | I noticed that wasm-pack uses some quite outdated dependencies like an old wash-opt | 06:59:23 |
| cwfitzgerald (wgpu, rend3) | Yeah, it's a pretty constant problem | 07:00:23 |
 multisn8 (I am a sign) | With the new CompositeAlphaMode, two things
- the comma at
automatically，depending is actually NOT ,, it's ，, character U+FF0C. Guess someone was in the flow
- if I happen to care about the alpha_mode, can I somewhere look up what's supported where? I think I'd like to use PreMultiplied but that won't make sense if it spuriously fails
| 07:06:51 |
| cwfitzgerald (wgpu, rend3) | Opaque is the only guarentee iirc , depends a lot on both vendor and backend | 07:10:21 |
| multisn8 (I am a sign) | aaaaAAAAaaa I already see myself sinking through issues about "y no transparency, it worked on my friend's PC" but thank you, that's good to know | 07:11:43 |
| cwfitzgerald (wgpu, rend3) | You need to provide us one of the supported modes, so it'll be a validation error if you don't | 07:28:37 |
|  @kim_martial:matrix.org joined the room. | 10:51:54 |
 ryan | I need some seedable 3D noise, but totally random values between 0.0 and 1.0, not perlin clouds. Can the noise crate do that, or is there something better? | 11:33:24 |
 scoopr | if something returns random u32, then dividing by u32::MAX gets you 0..1 range, right? | 12:10:36 |
| scoopr | I've occasionally used a "squirrel noise", inspired by this talk https://www.gdcvault.com/play/1024365/Math-for-Game-Programmers-Noise | 12:11:32 |
| ryan | I need the "something that returns random u32" | 12:20:53 |
| ryan | but it needs to take an x,y,z input | 12:21:14 |
| ryan | I'll just go with let seed = (6543 + x as u64) * 63456 + (7423 + y as u64) * 23982 + (4643 + z as u64) * 5634; | 12:26:56 |
| multisn8 (I am a sign) | In reply to @r4tty:matrix.org
I need some seedable 3D noise, but totally random values between 0.0 and 1.0, not perlin clouds. Can the noise crate do that, or is there something better? Hashes I guess | 13:00:31 |
 nical | cwfitzgerald (wgpu, rend3) and/or kvark do you know why the need for a full blit to resolve the attachment to the window in the gles backend? I'd like to document it for later (it's quite costly). | 13:20:35 |
 groves | nical: there's some background here https://github.com/gfx-rs/gfx/issues/3469 | 13:37:49 |
 wumpf | oh man that is the kind of writeup I've been looking for yesterday :D | 13:38:27 |
| groves | (this issue is in gfx, but that's where the original egl presentation logic came from https://github.com/gfx-rs/gfx/commit/94b71d5fc555f75999ca0fd7e0b72d7cc0e2d5d5#diff-3c3708b8ab2a59f5c903de589330d635515c80df4ccc5071e0d2db70a1e77227R276-R309) | 13:38:33 |
| nical | groves: thanks! | 13:39:47 |
| wumpf | I know I keep coming back to the same thing over and over again here, but something the issue doesn't mention is the need to flip y because of flipped texture space. Yes if a user is aware of it that's not a problem, but it's hard to deal with on a wgpu level | 13:41:19 |
| nical | Good to know | 13:49:49 |
| nical | I know ANGLE has a similar code path with a big sad blit to deal with the y flip issue as well | 13:50:16 |
| nical | IIRC there's an extension to disable the blit and deal with the inverted y manually which is good enough | 13:51:37 |
| wumpf | I think that would make a ton of sense for wgpu | 13:57:57 |
| @kim_martial:matrix.org left the room. | 16:00:17 |
