liittle question about reduce

[monarch_dodra]

On Saturday, 29 September 2012 at 23:07:30 UTC, Timon Gehr wrote:
> On 09/30/2012 01:04 AM, bioinfornatics wrote:
>> hi,
>>
>>
>> int[] list = [ 0, 1, 1, 2, 3, 4, 5, 6, 7, 8, 9 ];
>>
>> why this do not works ?
>> list.reduce!( (a,b) => a + b )( 0 ); // sum all elements
>>
>> but by this way that is ok:
>> reduce!( (a,b) => a + b )( 0, list ); // sum all elements
>>
>>
>
>
> Because of the parameter order.
>
> 0.reduce!((a,b)=>a+b)(list); // works

Yeah... UFCS sometimes doesn't lend itself all that well to 
certain functions.

This because UFCS was invented later in D's life cycle. It would 
have been better if reduce's range was defined as the "first" 
argument, rather than the "last".

This is especially true, because you don't have to specify the 
seed:
reduce!( (a,b) => a + b )( list ); //OK
reduce!( (a,b) => a + b )( 0, list ); //OK

This reads very odly to me. I know this is not a case of "default 
argument", but I don't like the change of usual behavior. I'd 
have expected this as valid syntax:
reduce!( (a,b) => a + b )( list, 0 ); //Should be the valid 
syntax.

Too late to change it now I guess! (unless we create a duplicate 
function called accumulate or something, but won't happen).

Anywhoo, if you don't specify the seed (which you don't have to 
here), then you can just use:

     int[] list = [ 0, 1, 1, 2, 3, 4, 5, 6, 7, 8, 9 ];
     list.reduce!( (a,b) => a + b )(); // sum all elements

If you wanted the sum of the 10 first integrals, this also works:

     iota(0,10).reduce!"a+b"().writeln();

I really like the trailing writeln() :D UFCS is BY FAR one of the 
things I enjoy the most in D (not the most important feature, but 
the most enjoyable)