[llvm-dev] RFC: changing variable naming rules in LLVM codebase

[David Greene via llvm-dev]

Rui Ueyama via llvm-dev <llvm-dev at lists.llvm.org> writes:

>  - LLVM's `/*foo=*/`-style comment to annotate function arguments need
>  to be handled automatically to make the tool scalable. So is the
>  doxygen @parameter.

This is a bit of a side note, but in my own work I've more recently
tried to move from this style:

foo.h

int foo(int a, bool doSomething);

foo.cpp

x = foo(a, /*doSomething=*/true);
y = foo(a, /*doSomething=*/false);

to something like:

foo.h

inline constexpr DoDoSomething   = true;
inline constexpr DontDoSomething = false;

int foo(int a, bool doSomething);

foo.cpp

x = foo(a, DoDoSomething);
y = foo(a, DontDoSomething);

One doesn't need the inline variable (and thus not C++17) if one uses
macros or enums or something else less elegant.

This kind of thing is slightly more cumbersome to do if the parameter
takes a wide range of values, but the most common place I see the former
style is for Boolean arguments.  Nevertheless, the wide-ranging case can
be handled:

bar.h

inline int Threshold(int x) { return x; }

int bar(int a, int threshold);

bar.cpp

x = bar(a, Threshold(1000));
y = bar(a, Threshold(100));

With either technique the "named values" could be wrapped in their own
namespaces to avoid collisions.

I wonder if there is any appetite for doing something similar in LLVM.

                       -David