> The sad thing is that there is so much NIH around MLIR. The infrastructure investments in LLVM are massive, but now lots of people are rushing off to reinvent

> The sad thing is that there is so much NIH around MLIR. The infrastructure investments in LLVM are massive, but now lots of people are rushing off to reinvent the world rather than bringing the ideas to LLVM itself. What if your idea is completely opposite to LLVM's goals? For example, you want a very small (e.g. in the sense of LOC) compiler?