Re: [RFC PATCH v3 0/8] Remote helpers smart transport extensions

Nanako Shiraishi <nanako3@lavabit.com> writes:

Nico already gave you correct and more concise version; this more verbose explanation is primarily for you who said a few times that you are not fluent in C. Others can skip this message without missing anything.

I haven't asked people why they choose to write like this:

	char* string;

beyond "that is how we were taught and what we are used to".

But if I have to guess, it is because it makes it appear as if you are being consistent between basic types like "int" and pointer types like "char *". E.g. these both look like "<type> <variable>;"

	int     variable;
        char*   variable;

The example in CodingGuidelines gives a clear illustration that this however is not the way how C language works. IOW, the syntax for declaration is not "<type> <var1>, <var2>,...;". E.g.

	char*	var1, var2;

declares var1, which is a pointer to a char, and var2, which is a char. That is what Nico means by '* is an attribute of the variable and not of the type'.

The only sane way to read declaration in C is to read from inside to outside, starting from a variable name. When you read "char **argv", for example, you treat it as "char (*(*argv))", and read it from the variable name:

    - It declares argv; what's the type of it I wonder...

    - It is a pointer, because the asterisk in front of it tells us that
      we can dereference it. Now, what's the type of the result of
      dereferencing argv, i.e. "*argv", I wonder...

    - That "*argv" is a pointer, because the asterisk in front of it again
      tells us that we can dereference it. Now, what's the type of
      dereferencing that "*argv", i.e. "**argv", I wonder...

    - Ah, it is a "char", that is what this declaration says.

And the style we use (which is the same as Linux kernel style) naturally matches the way how you read this declaration.