"I found it easier to program when I was trying to figure out the logic for myself rather than trying to figure out where in the infinite stack of documentation was the function I needed. So for me, programming is more like creating something rather than looking it up, and too much of today’s programming is more like looking it up." - K
Is the committee planning on working on the preprocessor? I don't see any reason for not boosting it. It's time for C to have real meta-programming. Would be nice to have local macros that are scoped.
On another note:
- Official support for __attribute__
- void pointers should offset the same size as char pointers.
- typeof (can't stress this one enough)
- __VA_OPT__
- inline assembly
- range designated initializer for arrays
- some GCC/Clang builtins
- for-loop once (Same as for loop, but doesn't loop)
I know some people are against metaprogramming because they believe the abstractions hide the intrinsic of how the underlying code will execute, but I would love to write substantial tests in C without relying on FFI to Python or C++ to perform property-based testing, complex fuzzing, and whatever. I feel metaprogramming would be a huge boon for C tooling and developer productivity.
In my point of view, there's a difference between abstraction created by the language, e.g. lambdas or virtual table in C++, and abstraction created by the programmers via the CPP.
The former is compiler dependent and you cannot know how it's implemented. The former is simple text substitution and you're the one implementing it. I often find myself creating small embedded languages in CPP for making abstraction, and I know exactly what C code it's going to generate and thus the penalty if there's any.
People that are afraid of the preprocessor simply don't understand how powerful it's in good hands.
