Quote:Original post by _goat
C and C++ will teach you everything you need to know about the "fundamentals" of programming. I'm not talking about paradigms here, or areas of programming - I'm not talking about dynamic typing or advanced (some would argue "complete" is a better word) RTTI - I'm talking about how a computer, physically, does things. Memory addresses, offsets, alignment, registers, etc. C# hides some of these things.

How a computer physically does things is no more fundamental than type systems -- there's a strong case to be made that the type system is rather more fundamental than largely irrelevant implementation details like memory addresses and registers. That C# hides those things should be a strong tip off that the physical details are not fundamental at all.

What is fundamental (as far as imperative programming goes anyway) are variables, control flow and control structures, functions, and basic logic. C# and C++ both have plenty of that.

Quote:All this is true in the theory, but tell me, do you know a c++ programmer who is not farmiliar with addresses or registers ? IMO learning c/c++ implies the learning of basic machine-level programming, and every c/c++ book should at least cover bare minimum of this.

I suppose that depends on perspective. From where I'm sitting, I see a lot of C++ programmers who do not understand registers or memory at all. They might think they do -- they got the textbook explanations of pointers and registers, after all -- but they only have those simplified views and very little idea of the reality of how things work.

C++ -- well, really C programmers -- will deal with pointers a lot. But how many of them understand addressing modes? Segments? Virtual memory? Pages? Page faulting? Translation lookaside buffers? All they know is the comfortable illusion of a flat memory space which has been created for them. A lot of beginners don't even realize that they can't access other programs' memory simply by setting pointers to certain values.

Learning C++ doesn't magically gain you machine level knowledge. If anything, it can create serious misconceptions about how things work.

Quote:Original post by RedDrake
All this is true in the theory, but tell me, do you know a c++ programmer who is not farmiliar with addresses or registers ? IMO learning c/c++ implies the learning of basic machine-level programming, and every c/c++ book should at least cover bare minimum of this.

I certainly am not familiar with addresses or registers. Actually, I would say that my knowledge of mainstream CPU registers and of mainstream computer memory layout is insignificant and sometimes nonexistent.

EDIT: to quote Promit, I have no idea what addressing modes, segments, virtual memory, pages, page faulting and translation lookaside buffers are, or how they work.

I do not consider myself to be a bad C++ programmer, though, quite the contrary, and I am extremely familiar with the guarantees that C++ makes over the underlying hardware—the irony of this is that this allows me to write safer and more portable code than the average C++ programmer, because I don't mix up platform-specific details and language-enforced concepts. I have never found the need to know about the specifics of my hardware unless using that hardware was my primary objective—which is why I am also highly familiar with the ARM processors found in 2000-2004 PocketPCs and their corresponding assembly, and also with the hardware and PTX assembly of the GeForce 8800.

But at that point, I'd say a programmer needs to learn about machine-level programming when he actually needs to do machine-level programming, not when learning a self-advertised general-purpose language.