I haven't looked at the GCC code myself, but what I heard is that you can see that it's a codebase that grew massively over a large timespan and things that sho

I haven't looked at the GCC code myself, but what I heard is that you can see that it's a codebase that grew massively over a large timespan and things that should be separated are not. At least a few years ago there were actually two different IR formats (the newer one being based on SSA) which are converted between each other several times. I think GCC's IR does not have a canonical text representation, this is really helpful for understanding and debugging in LLVM. Still also LLVM has its weird parts, like their variable naming scheme