His approach is a lot more nuanced and tactical than it is often portrayed.
Those don't actually contradict it in any way that I can see.
> His approach is a lot more nuanced and tactical than it is often portrayed.
Sure. That doesn't mean that the long-term goal isn't as I've described it (and, in fact, RMS has pretty explicitly said it is for him, though that may or may not be true of the FSF.)
He supported BSD licences for Xiph codecs because he recognised the impact of network effects,
You claim that both these thing are considered "evil", "unconscionable" and "categorically wrong" by him regardless of whatever benefits they create. Clearly RMS disagrees, and is happy to use them as tools to achieve his long term goals in specific circumstances.
> The existence of LLVM is a terrible setback for our community precisely because it is not copylefted and can be used as the basis for nonfree compilers -- so that all contribution to LLVM directly helps proprietary software as much as it helps us.
... which pretty much is the answer to your last question.
You can create proprietary software _with_ GCC just fine, and he has never proposed any measures with which to impose any restrictions on the final output binaries you create with GCC so I don't really follow your reasoning.
The 'problem' with LLVM from his view is that it is easy to incorporate into proprietary software, as well as easy to use in conjunction with proprietary software (through plugins for instance), which means that there is potentially much less incentive to contribute your code back when it's easy to maintain it in proprietary form, which then leads to less 'free' software.
His cynicism has a tendency to come true, so it will be interesting to see if LLVM open source development will continue to flourish or if we will see a slow but steady transition of 'real improvements' happening in proprietary plugins or forked versions rather than contributed back to the open source project.
In some ways we might be seeing it already, last time I heard (which was in september IIRC), Apple had not contributed their A7 support to the open source versions of Clang/LLVM, and are instead keeping it proprietary and only available on the Clang/LLVM versions they ship with XCode on their OSX platform.
He makes a lot of good points and is very aware, but his extreme ideological dedication is what drives people off. Which is a shame, because he's done a lot to convince and further adoption of free software in government and education, plus there's plenty of things one can learn from him.
This presents an interesting (to RMS) quandary. I take it as a given that he prioritizes the freedom of software over its technical merits (as discussed in other threads). But, even just from the freedom perspective: LLVM increases the amount of proprietary software produced, and also increases the amount of free software produced. Does RMS consider this a worthwhile trade-off? Or does he think that less software overall, and less free software in particular, is favorable to more free software existing and also some proprietary software.