Open mvaligursky opened 6 months ago
@mvaligursky I think we concluded this fallback behaviour would be managed by the user?
in a published app, we will try and create the device type based on user specification, and fallback to null renderer if none of those are available. User can handle this case.
But in the launcher, we should warn them if the device they selected for testing is not the one that is used. Launcher is our responsibility I think. Lets say their app handles null device. But in the launcher they want to reliably test against WebGL and WebGPU, and if they don't get the one they selected, we should tell them a fallback is used.
Is this still valid @mvaligursky
based on comments I would say yes, still valid.
In the launcher, the user can select
Prefer WebGPU
. But when the browser does not support WebGPU, we silently fallback to WebGL, giving the user a false sense ofit works on WebGPU
.In this case, we should some how inform the user of the fact this is WebGL and not WebGPU as requested.
Perhaps we can use the same system we use when the launcher is using a release candidate of the engine: