Validation in CTS tends to do a good job of covering edge cases for limits. We in Firefox have noticed that CTS' validation generally doesn't try to cover edge cases for the bounds of integral values, and this seems interesting to add. This is valuable because:
- Obviously, validation of bounds always needs to be robust for these edge cases.
- Less obviously, rendered diagnostics should also not encounter overflow.
Adding coverage for (1) is essential for ensuring that an implementation safely handles inputs, independent of things like maximum values for limits. (2) is nice, and generally comes for free with (1).
JS doesn't tend to give direct access to overflow cases for GPUSize64s, but this seems achievable with APIs that accept GPUSize32s offsets and sizes.
Validation in CTS tends to do a good job of covering edge cases for limits. We in Firefox have noticed that CTS' validation generally doesn't try to cover edge cases for the bounds of integral values, and this seems interesting to add. This is valuable because:
Adding coverage for (1) is essential for ensuring that an implementation safely handles inputs, independent of things like maximum values for limits. (2) is nice, and generally comes for free with (1).
JS doesn't tend to give direct access to overflow cases for
GPUSize64s, but this seems achievable with APIs that acceptGPUSize32s offsets and sizes.