The last half is basically his own opinions stated as truth from God, which, to me, makes him look like a complete tool. Personally I'm not a fan of parens around the conditional in ifs, but I don't go around spouting that it is the only true way.

It basically makes me write off the first half, even if it is correct (I'm not a linux dev, so I don't know).

Added: From the reply on lkml I'm not the only one.

It's a style that has been recognized as humorous for decades before the corporate CoC-wielding bureaucrats moved in. Much more refreshing than passive aggressive replies like this one:

https://lore.kernel.org/lkml/[email protected]/

I think Rust is not a culture fit for the Linux kernel and its adoption will further undermine Linux.

Humorous? I don't think how anyone could think stuff like "I cannot view with less or vim. Therefore it looks not at all." as even remotely funny.

Sounds just like "neckbeard yells at cloud" style to justify that the world does not do exactly what he wants.

The first point is quite valid though. From "https://doc.rust-lang.org/reference/memory-model.html" where it says "Rust does not yet have a defined memory model"

the unstated part here is that in the absence of a compiler-visible atomic modelling system, an atomic model is implemented in `core::atomic` and LLVM. similarly, a `kcore` library could provide atomic types just like the way the LKMM is defined as a library in-tree.

this is not a language limitation, as the language is able to emit LLVM atomic operations just fine. like allocations, it needs people with a defined goal (like the kernel) to provide a target for implementations to reach; proäctive creation of a model without knowing if it has an audience is not worthwhile, which is why it hasn't been done