18 May 2024 |
jasperrlz | It looks like it needs the device to determine the caps | 00:26:03 |
Elabajaba | In reply to @cwfitzgerald:matrix.org It should be changed to be a pure function technically fixing this would be a breaking change, but would it be backportable? | 00:26:03 |
Elabajaba | In reply to @jasperrlz:matrix.org It looks like it needs the device to determine the caps couldn't you just pass in the features as a parameter on the function | 00:26:50 |
jasperrlz | Yeah, you could | 00:26:55 |
Elabajaba | In reply to @jasperrlz:matrix.org It looks like it needs the device to determine the caps * couldn't you just pass in the device's features as a parameter on the function | 00:26:57 |
jasperrlz | Or you could pass in the capability set manually, and have a capability_set_from_device function or something | 00:27:22 |
jasperrlz | But at that point it's not that much different from Validator::new | 00:27:39 |
Elabajaba | In reply to @jasperrlz:matrix.org Or you could pass in the capability set manually, and have a capability_set_from_device function or something i think the whole point is to not have to duplicate all the capability checks | 00:28:18 |
Elabajaba | In reply to @jasperrlz:matrix.org Or you could pass in the capability set manually, and have a capability_set_from_device function or something * i think the whole point of it being changed to put is to not have to duplicate all the capability checks | 00:28:48 |
Elabajaba | * i think the whole point of it being changed to pub is to not have to duplicate all the capability checks | 00:28:53 |
cwfitzgerald | In reply to @elabajaba:matrix.org technically fixing this would be a breaking change, but would it be backportable? We're doing a breaking backport soon, so yes | 01:06:21 |
Elabajaba | i'll make a PR for it tomorrow then | 03:50:06 |
schell | Are there any 'gotchas' in wgpu regarding concurrency on Windows? I ask because I ran my test suite on windows (after really slogging through getting a MSVC dev setup going) and found that the tests seem to stall, and running them with -j 1 and --num-threads=1 helps. | 19:53:34 |
schell | It's probably my code, but just wanted to make sure | 19:53:46 |
| vanta1 joined the room. | 20:27:06 |
cwfitzgerald | In reply to @schell:matrix.org Are there any 'gotchas' in wgpu regarding concurrency on Windows? I ask because I ran my test suite on windows (after really slogging through getting a MSVC dev setup going) and found that the tests seem to stall, and running them with -j 1 and --num-threads=1 helps. Our test suite requires each test run in its own process, or with -j 1 - it should tell you this though | 20:58:03 |
cwfitzgerald | The expected route is to run cargo xtask test, which requires cargo-nextest installed | 20:58:27 |
schell | Ok, so itโs not just me then ๐
| 21:46:35 |
cwfitzgerald | In reply to @schell:matrix.org Ok, so itโs not just me then ๐
I'm surprised it hung and didn't yell at you very loudly | 22:01:17 |
schell | Yeah, I'm not sure why, but many things about Windows dev allude me. It's interesting this doesn't happen on my mac, though. | 22:19:53 |
schell | * Yeah, I'm not sure why, but many things about Windows dev illude me. It's interesting this doesn't happen on my mac, though. | 22:20:32 |
19 May 2024 |
schell | What does it mean when I create an instance with wgpu::Backends::all() and then instance.enumerate_adapters() is empty? | 04:23:08 |
schell | I can see that wgpu::Instance::enabled_backend_features() returns Backends(VULKAN | GL) | 04:25:27 |
schell | Should I have used wgpu::Backends::PRIMARY ? | 04:28:42 |
cwfitzgerald | In reply to @schell:matrix.org What does it mean when I create an instance with wgpu::Backends::all() and then instance.enumerate_adapters() is empty? It means you don't have any affairs | 05:01:08 |
cwfitzgerald | * It means you don't have any adapters | 05:01:16 |
cwfitzgerald | What system are you on? | 05:01:26 |
schell | One is an AMD linux runner and the other is an Intel runner | 05:06:44 |
schell | I think it's a problem with whatever containerization is happening on the runner - sshing in and running the tests manually works fine | 05:08:01 |
cwfitzgerald | Weird | 05:34:49 |