It feels a bit unfair to lay the blame for all of this on C. All other mainstream compilable languages compile to similar abstract machines in a similar fashion to C.
Low-level, high-level… such categorization is arbitrary semantics. An old and arbitrary IT meme, like vim vs emacs.
No useful outcomes are achieved through the debate. We learn nothing about a previously unknown capability of our technology or self structuring pointless taxonomy.
“Oh look a grammatically correct sentence.” is all this rises to.
Our computers though more powerful don't seem to work any better than they did before in my user experience of 30 years: stalls, hang ups, crashes and pop up windows telling me to do something...
Why do we trust this consumer level technology to do mission critical work: running our lives or ruining it?
Low level means simple translation to assembly, not the gigabytes it takes to make a compiler or the crazy implementations, I wonder why Randal Hyde's HLA is not popular or maintained by the community.
This is partly fighting a strawman. No performance aware programmer worth their salt assumes a fast PDP-11. C is still loved because it provides control over memory layout and memory management, maybe the two most important things if performance is needed.
This old canard again? Hasn't this been debunked multiple times already?
"C is not a low-level language"
Uh, okay, then what is?
What are you comparing it to? Low-level/high-level is not a binary switch, it's a spectrum; exactly how many languages are lower-level than C and how many are higher?
Arguably then only High-level Assembly [1] meets the criteria. Or perhaps micro-coding.
That said with caches, branch predictors, register renaming, op-cracking, reordering, it's hard to say that assembly language is even low-level anymore because it's significantly abstracted over what is going to occur in the chip.
> Hasn't this been debunked multiple times already?
No. One of the main problems with the C user base is that there's a sizable number of developers who attach some sort of special status to C as the closest thing to assembly language and get incredibly belligerent with compilers when they fail to reproduce the exact assembly they were expecting. (Cryptographers angry that the compiler is getting rid of their constant-time tricks is one of the more notable cases, no matter how many times compiler vendors keep saying "stop using C if you want constant time, we don't guarantee it, we won't guarantee it").
> What are you comparing it to?
The reality is that C isn't meaningfully closer to assembly language than just about any other non-runtime based language--things like Fortran, Pascal, Rust, Zig, etc. are all around as good as C at representing machine hardware. Honestly, I'd say Rust is marginally better than C, by virtue of the fact that it allows multiple return types and lacks some of the implicit conversion inanities of C.
> Uh, okay, then what is?
The closest notable language you'll probably find is... LLVM IR (although I'd caveat that with LLVM IR isn't meant to actually be written by hand). LLVM does hew too close to the C abstract machine to be truly effective in this role, but it does make the crucial first step of having a distinction between registers and memory that is absent in all of the regular programming languages I'm aware of.
By raw compute power though it's pretty much x86, ARM, and whatever proprietary internal machine code Nvidia and AMD use on their GPUs that make up the vast majority.
