Why the Visual.VisualOffset is double type but run the compositor in floats?

We will use Visual.VisualOffset to set the visual offset. The Visual.VisualOffset is in managed code. And the type of VisualOffset is Vector which is contained the double X and the double Y.

https://github.com/dotnet/wpf/blob/7e0195803bf0144daba4e5fd2d8be93774d043ef/src/Microsoft.DotNet.Wpf/src/PresentationCore/System/Windows/Media/Visual.cs#L3523-L3550

But we will use CMilVisual::ProcessSetOffset to set it in the MIL code. It will cast the double X and double Y to the float type. Just as the code says, The packet contains doubles. Should they be floats? Why are we using doubles in managed but run the compositor in floats?

https://github.com/dotnet/wpf/blob/7e0195803bf0144daba4e5fd2d8be93774d043ef/src/Microsoft.DotNet.Wpf/src/WpfGfx/core/resources/node.cpp#L425-L439

Can I know the reason?

[miloush]

My guess is because there is no Vector struct that holds floats, and all computations and math functions use doubles anyway. GPUs on the other hand can do much faster calculations with floats. Changing either side seems unreasonable, and changing the managed type to use floats - apart from compatibility issues - would move the burden of conversion on the developer (and from native to managed performance).

[vishalmsft]

@batzen The question exists in the code as comments. At some point of time, we may circle back to this one as code-review. This issue is practically a duplicate of issue already reported in code.