Will LLILC render your compiler superfluous?

In case you haven't heard. the .NET foundation has been working almost a year to create an LLVM-based MSIL compiler (http://www.dotnetfoundation.org/llilc) which will allow C# to machine code transformations through LLVM (http://llvm.org). The focus now is on JIT but, according to the intro at https://github.com/dotnet/llilc, an AOT compiler is planned.
Has anyone investigated the extent to which the flingOS compiler will be necessary after the release of LLILC and how much "purpose overlap" already exists?

Comments

  • Nevermind. A few days research and I realized why is was a naive question. Sure LLILC will open up C# to LLVM optimizations and additional platforms, but apps still need, at a minimum, some run-time garbage collector, even if it's statically linked; and that's already available.
    For example I just used https://github.com/dotnet/corert on Ubuntu to compile "Hello World" and for that one-line main method it produced 126,023 lines of C++ code (hello.cpp) and a 4MB executable -- and that's a 'Release' version!

    I do like the FlingOS project. As an educational resource I think it deserves it's own section at osdev.org!
Sign In or Register to comment.