I made no statements about licensing. It was all about updates and developer velocity. Lots of folks wanted to get patches into GCC and the GCC devs at the time were camping on the submission queue. The EGCS fork was a way around that, but a balkanized compiler is kinda like a poison pill, as soon as someone uses a feature in one and not the other, it bifurcates a codebase.
Think of what would have happened if some distros took alternate tooling paths? What about popular tools? We have to have both toolchains on our systems.
It is clear from how things turned out that EGCS wasn't about ownership or splitting the ecosystem. If one were to do this now, you would constantly rebase against upstream while you sort out the politics.
Clang/LLVM was both about implementation architecture and licensing, but had GCC been more modular, it might not have existed as well.
Think of what would have happened if some distros took alternate tooling paths? What about popular tools? We have to have both toolchains on our systems.
It is clear from how things turned out that EGCS wasn't about ownership or splitting the ecosystem. If one were to do this now, you would constantly rebase against upstream while you sort out the politics.
Clang/LLVM was both about implementation architecture and licensing, but had GCC been more modular, it might not have existed as well.