Comments by "Mikko Rantalainen" (@MikkoRantalainen) on "Mastery Learning" channel.

@yifeiren8004 "Go is better than Rust" similarly how C# is better than C. Managed languages are easier to use but not suitable for all projects because garbage collection has non-neglible runtime costs, especially RAM requirements. As a prime example, C# runtime is actually implemented in C++ because C# doesn't handle such requirements. Similarly, Go project uses gccgo which is part of gcc and uses C++, too. I haven't seen comparisions for C# or Go but Java JVM typically needs 6x the RAM of a C or C++ program using similar algorithms to perform well. If you don't mind paying 6x for the RAM, Java is great.
75
@NoSpeechForTheDumb C++ would have been mostly about syntax only when it comes to use of the language in Linux kernel because you cannot use most of the stuff you mentioned within the special limitations of kernel mode programming. I totally agree that C++ is more than different syntax compared to C in generic programming tasks.
26
@KoflerDavid In addition, any new language must offer something but just different syntax. In case of Rust, the safeguards allowed by the language to avoid data races or memory corruption while still working in kernel space (as opposed to languages with managed memory / garbage collection). C++ would have been just about having different syntax.
23
@StringTerminator C# is better than C in same sense as Java is better than C: programming is easier because you don't need to think about releasing memory that much. Obviously the performance is much worse.
4
If I were to decide, all nesting should be made with tabs and a single nesting level is always a single tab. And all tabs on the line after first non-tab character would be always interpreted as error. That way nobody would be trying to align anything in the statements with tabs but the amount of visual indentation could be user decision instead of being hardcoded in the source code. For example, I prefer Allman-8 style indentation but I'm totally aware that most people consider 8 space width indentation excessive. My above definition would allow one developer to use 2 space indentations for the lines and I would be able to use 8 space indentations even though the binary file on the storage were exactly the same!
4
I interpreted that as doing the same mistakes for the kernel code (e.g. incorrect locking between threads for CPU native code), side-channel data leak vulnerabilities (Spectre-class vulnerabilities in hardware) and poor performance for TLB cache misses or the need to force clearing TLB cache. It would be great if RISC-V could avoid all those mistakes but I'm expecting them to hit at least some of them.
3