On the other hand, consider the following: Emscripten, Numba, Julia, Rust, Pyston, Pure, Polly, Mono-LLVM, ghc -fllvm, Terra, ... and those are just the ones I can think of in a few seconds. Several of these have the potential for huge impact over the next decade, and nearly all of them would have been significantly less viable either using GCC tooling (due to license-driven architectural decisions) or building a backend from scratch.
LLVM is known for being rather ruthless with their API refactorings. This has the twin advantages of keeping the code quality very high, and giving people an extra incentive to contribute upstream (e.g. the PS4 compiler backend). I won't go so far as to say this is intentional, but it does represent a rather amusing way to discourage users from keeping everything in-house.
LLVM is known for being rather ruthless with their API refactorings. This has the twin advantages of keeping the code quality very high, and giving people an extra incentive to contribute upstream (e.g. the PS4 compiler backend). I won't go so far as to say this is intentional, but it does represent a rather amusing way to discourage users from keeping everything in-house.