Pent Ploompuu

But why limit it to only nullability covariance? Making `Task` covariant on the type would be very helpful. Option 2 (`` in class types) doesn't work as discussed, but some...

I did some benchmarking with various data lengths (in bytes). JenkinsV2 is the version in this PR. Overall it looks like XxHash32 would be the best option (and it has...

xxHash64 is a bit faster for inputs as small as 24, but I'm not sure if the quality stays the same if we truncate the result from 64-bits to 32-bits?

> > Unconditionally using the intrinsic from Math.DivRem API is going to cause a lot of regressions > > I agree. I checked the current implementation vs. `DivRem` and with...

Adding a the hosting dependency just to validate a configuration value that is likely to be valid now (since both default to 30s on .NET 7) seems a bit much....