Open
Description
- This issue is blocking
- This issue is causing unreasonable pain
Today dotnet repos that build cross-platform native code all define their own settings for CMake. This can lead to differences like different optimizations / symbols / shared libraries etc. At the very least it's a ton of duplication and redundant work.
We should try to find how much of the dotnet/runtime native CMake infrastructure is shareable and put that in arcade. Then we can have repositories use that infrastructure rather than their locally defined scripts. One candidate we have for this is dotnet/machinelearning, but I bet there are others.