• Walter Bright <newshound2 digitalmars.com> Oct 21 2010
• Juanjo Alvarez <fake fakeemail.com> Oct 21 2010
• bearophile <bearophileHUGS lycos.com> Oct 21 2010
• Juanjo Alvarez <fake fakeemail.com> Oct 21 2010
• bearophile <bearophileHUGS lycos.com> Oct 21 2010

Walter Bright <newshound2 digitalmars.com> writes:

http://news.ycombinator.com/item?id=1814887

Some commentary on D there.

Juanjo Alvarez <fake fakeemail.com> writes:

On Thu, 21 Oct 2010 10:51:14 -0700, Walter Bright 
<newshound2 digitalmars.com> wrote:
 http://news.ycombinator.com/item?id=1814887



 Some commentary on D there.


I always enjoy a good rant about some language.  When I'm intrigued 
by some  language rants are the first things I google. 

I wish Go, sorry, D had named arguments too. Its a simple feature 
that nuke a trunkload of function overloading.

bearophile <bearophileHUGS lycos.com> writes:

Juanjo Alvarez:

 I wish Go, sorry, D had named arguments too. Its a simple feature 
 that nuke a trunkload of function overloading.


I use named arguments every day in Python, they make the code more readable and
safer at the calling point, because there's less risk of passing wrong data.

And recently I have seen that named arguments are widely used in SPARL Ada too,
to make code less bug-prone. If a feature is seen as good in both Python and
SPARK (that are two very different language), then it's good.

This is the syntax in Ada (and SPARK):
http://en.wikibooks.org/wiki/Ada_Programming/Subprograms#Named_parameters
And in recent C#:
http://msdn.microsoft.com/en-us/library/dd264739.aspx

On the other hand, currently there are many D2 features that are unfinished and
buggy, so adding even more stuff is not a good idea. And I think named
arguments are a purely additive change. So Walter may add them later when the
current features are implemented well enough. Currently it's much more
important to focus on eventually needed non-additive changes instead.

Bye,
bearophile

Juanjo Alvarez <fake fakeemail.com> writes:

On Thu, 21 Oct 2010 15:45:55 -0400, bearophile 
<bearophileHUGS lycos.com> wrote:
 On the other hand, currently there are many D2 features that are 


And I think named arguments are a purely additive change. So Walter 
may add them later when the current features are implemented well 
enough. Currently it's much more important to focus on eventually 
needed non-additive changes instead.

Very true. I also use named arguments constantly in Python. 

Since I'm not the only one that would like to have it in D 2.1 I'll 
go into gredy mode and add that expanding a dictionary as named 
arguments to a function is pretty useful too, but since in D's hashes 
the values must be of the same type for a declared hash it would be 
less useful, except for Variant[string] hashes maybe, which is a 
little convoluted.

bearophile <bearophileHUGS lycos.com> writes:

Juanjo Alvarez:

 expanding a dictionary as named 
 arguments to a function is pretty useful too, but since in D's hashes 
 the values must be of the same type for a declared hash it would be 
 less useful, except for Variant[string] hashes maybe, which is a 
 little convoluted.


In D2 there are even typesafe Variadic Functions for dynamic arrays and class
objects:
http://www.digitalmars.com/d/2.0/function.html

An example:

class Foo {
    int x;
    string s;

    this(int x_, string s_) {
        this.x = x_;
        this.s = s_;
    }
}

void test1(int[] array ...) {}
void test2(int x, Foo f ...) {}

void main() {
    test1(4, 5, 6);
    test2(1, 4, "text");
}


So far typesafe variadic functions for class objects is an useless feature for
me.

Once named arguments are present, typesafe variadic functions for associative
arrays may look like:

void test3a(int[string] aa ...) {}
void test3b(Variant[string] aa ...) {}

You may call them like this:
test3a(x: 1, y: 2, good: 3);
test3b(x: 1, y: 2, good: "right");

Bye,
bearophile