John Carmack's Comments On C/C++

[plonoma]

It's not about how hard it is to substitute auto.
It's about how easy it can be made to figure out which auto to substitute.

[log0]

The choice of the word auto as a keyword for automatically determining type is an abomination in terms of expressiveness.
The word autotype should have been used.
(Auto is just too short and ambiguous. People are going to make mistakes easier.)

Also in re factoring projects where the word auto had another meaning it's now more of a problem.
You have to define an alias to know what has been re factored and what not. (Must see the difference easily remember, can't use things that rely on keywords being different.)

You mean too short and ambigous compared to short, long, double?

And if you find it too hard to substitute auto, you might be using the wrong source code editor.

[plonoma]

How about this:

writeln("Today is January ", 15, ", ", 2013);

Or even:

printf("Today is January %d, %d\n", 15, 2013);

Really like the first one.
Would like to do the following:

uint intvariable = 5;
setWrite(std::cout);

writeln("sghsfmhj",12," tyjumdnjmjdsgjhpg ", 2013," pesriprtu", intvariable);
write(" pegihjesj");
writeln(123);
writeln({5}133);

Would produce the following three lines of output:

sghsfmhj12 tyjumdnjmjdsgjhpg 2013 pesriprtu5
 pegihjesj123
  133

(The {} is used for format specifying but is in front or behind the object to be converted to a string.)
Also being able to do toString();
With comma separated stuff like in the example.
Also being able to do a toStringNl();
Same as the previous function but with an endline behind it.
Being able to use the << operator here:
<<("sfsgfsf ", 45," sdrghjdsgol");

(also use of {} with multiple parameters, Example: {6,"^"} Would reserve six places and have as second parameter a filler for the empty space, also multiple {} on one, cascading to form more interesting outputs)

Short, easy readable, see what you get out of it, concise, very functional and high-level.

[plonoma]

The choice of the word auto as a keyword for automatically determining type is an abomination in terms of expressiveness.
The word autotype should have been used.
(Auto is just too short and ambiguous. People are going to make mistakes easier.)

Also in re factoring projects where the word auto had another meaning it's now more of a problem.
You have to define an alias to know what has been re factored and what not. (Must see the difference easily remember, can't use things that rely on keywords being different.)

[F i L]

And for the reference, here's the D version (with an additional case and that is syntactically valid):

template foo(T, O : int) 
{
    auto foo(T x, O y)
    {
    	return x * y;
    }
}

int main()
{
    string  s = "text"; // 's' is text
    auto i = 0;      // 'i' is an int
    auto f = 0.0 ;   // 'f' is a double

    auto r = foo(s, i) ;// error: can't pass text as a first parameter
    auto r2 = foo(f, i) ;// works: because compiler can multiply a 'int' and a 'float'
    auto r3 = foo(i, i) ;// works: compiler can compile 'int' and 'int'
    auto r4 = foo(i, f) ;// error: can't pass non-int as second parameter
    return 0;
}

That D code can be even prettier:

auto foo(T)(T x, int y)
{
    return x * y;
}

void main()
{
    auto s = "text"; // 's' is text
    auto i = 0;      // 'i' is an int
    auto f = 0.0 ;   // 'f' is a double

    auto r1 = foo(s, i); // error: can't pass text as a first parameter
    auto r2 = foo(f, i); // works: because compiler can multiply a 'int' and a 'float'
    auto r3 = foo(i, i); // works: compiler can compile 'int' and 'int'
    auto r4 = foo(i, f); // error: can't pass non-int as second parameter
}

[Grawp]

Well, even the very basic printing to console. C is consistent in that it uses a function for that, printf(). And C++ uses the "cout" notation that looks way out of place.

Well, the printing to console does not have anything to do with the language itself! You have to understand that C++ != STL and C != libc. Let me explain:
cout is actually an object and you just call operator "<<" on it. cout is not a keyword, it's just name for a struct that someone made and the "<<" operator was just simply chosen because it looks nicer than (+,-,*...). You don't have to use the STL library. It's absolutely not mandatory or recommended. (Embedded systems don't use STL, Qt doesn't use STL...)
Anyway, these operator overloading is a beautiful thing. You can do something like complexnum3 = complexnum1 + complexnum2 and you can define what + and = do. It's like automatic calling some function complexsum(c,a,b) without any overhead over ordinary function. The cout << is just like that. You can you use printf. You can write somehting new. You can even use inline assembly like in C.

[nslay]

The first one was D, the second one was C.



Yea, it was an example that people picked up. In general I just feel that C++ is too reliant on all their operators, so code is no longer as easily readable if you are not familiar with the meaning of each.

I think you're exaggerating C++'s reliance on operators since only two are used to define an extensible and cascaded input/output mechanism.

Function overloading and the cascading behavior of operators make a strong case for this scheme ... no matter how unusual it looks.

In the end, most C++ programmers don't write their own operators (except for extending stream operators). Those that do are often doing it for mathematical objects (where it makes sense) or string-like objects (where + and "concatenate" are universally synonymous).

[GreatEmerald]

The first one would never work in C. See stdarg.h for more information. In C++11, you can use variadic templates for a writeln() function as you described. But then, at the very worst, the compiler will produce a unique function for every output format you use. More likely, the compiler will inline such templated writeln() functions.

The first one was D, the second one was C.

Also, as to "printf() vs cout", that just seems like such a stupid and unnecessary discussion.
You're not discussing languages, only conventions of standard libraries at best.

Yea, it was an example that people picked up. In general I just feel that C++ is too reliant on all their operators, so code is no longer as easily readable if you are not familiar with the meaning of each.

C++(11) does *exactly* this:

And for the reference, here's the D version (with an additional case and that is syntactically valid):

template foo(T, O : int) 
{
    auto foo(T x, O y)
    {
    	return x * y;
    }
}

int main()
{
    string  s = "text"; // 's' is text
    auto i = 0;      // 'i' is an int
    auto f = 0.0 ;   // 'f' is a double

    auto r = foo(s, i) ;// error: can't pass text as a first parameter
    auto r2 = foo(f, i) ;// works: because compiler can multiply a 'int' and a 'float'
    auto r3 = foo(i, i) ;// works: compiler can compile 'int' and 'int'
    auto r4 = foo(i, f) ;// error: can't pass non-int as second parameter
    return 0;
}

[plonoma]

Syntax being easy to read, expressive and flexible seems like it's the most important thing.
(I notice that with the << it's easier to read, produces smaller code and is (less error prone?) shorter than printf.)
(The use of the %s, %d specifiers has an issue since the type specification seems to conflict with the DRY principle of software engineering. )

[log0]

Also, he's now fighting to get the auto keyword to be used as a form of template for the next revision:

auto foo(auto i, auto j) {
    return i * j;
}

OMG, me want this!!!