We're seeing a resurgence in multiple architectures as the world hurtles away from desktops.
While it's good to have GNU alternatives that do some things better, portability is more important than ever. This means we need to keep adding toolchains for new arches as they come up. For example, you missed Snapdragon and A4, which together are probably around a billion devices.
I don't want to detract from LLVM, but the community needs to keep the GNU one prospering if we want to avoid proprietary tool hell.
Aye, and the rest of the LLVM toolchain is supporting a huge number of architectures, some of which, like GPUs, that the GNU toolset doesn't support. Like any project, LLD is supporting the most popular architecture first, with roadmapped items for further architectures once that is finished. Keep in mind that binutils didn't support all the architectures it did with v1, it took porting work as well.
Furthermore, I'd disagree that the GNU binutils are the only toolset that are keeping us from proprietary hell. The LLVM development model, with fairly frequent releases and updates gives a huge push for toolchain users to contribute their patches. Sony, for example, had been grumpily looking at all the new features in LLVM being added as they were developing the toolchain for the PS4 because they couldn't add their contributions into the LLVM toolchain, and couldn't import the LLVM updates to avoid DNF syndrome.
I'd even go as far as to say that LLVM is our best hope of keeping the GNU project honest. A competing, Free, toolchain means that the GNU project cannot feign ignorance of the new features that llvm is bringing to the development table. The abilities that llvm offer are slowly trickling into gcc, even if certain parties are fearful of what effects those changes might bring. The right mix of cooperation and competition can and does bring wonderful things.
While it's good to have GNU alternatives that do some things better, portability is more important than ever. This means we need to keep adding toolchains for new arches as they come up. For example, you missed Snapdragon and A4, which together are probably around a billion devices.
I don't want to detract from LLVM, but the community needs to keep the GNU one prospering if we want to avoid proprietary tool hell.