Cartesian product of immutable ranges

[matovitch]

On Sunday, 26 January 2014 at 22:52:56 UTC, Jakob Ovrum wrote:
> On Sunday, 26 January 2014 at 22:32:32 UTC, matovitch wrote:
>> You mean that two input ranges are created from the immutable 
>> arrays when I call the function ?
>>
>> Zip doesn't compiles while zip compile. :/
>
> `Zip` must be explicitly instantiated, which allows you to 
> attempt to instantiate it with head-immutable slices, which 
> isn't allowed because head-immutable slices aren't input ranges 
> due to a lack of working `popFront()`.
>
>> Here is the implementation of zip :
>>
>> auto zip(Ranges...)(Ranges ranges)
>>    if (Ranges.length && allSatisfy!(isInputRange, Ranges))
>> {
>>    return Zip!Ranges(ranges);
>> }
>
> `zip` is a range constructor function. These are used to allow 
> IFTI (Implicit Function Template Instantiation) to infer 
> template arguments, so the user doesn't have to provide them 
> explicitly.
>
> With IFTI, head-immutable and head-const are stripped from 
> slice types, so even though you pass immutable(int[]), the 
> parameter type is set to be immutable(int)[], i.e. an mutable 
> slice of immutable integers. Only the latter is a valid input 
> range.

Thank you very much for these very clear explanations ! I am 
quite tired tonight and shouldn't have started to code in the 
first place anyway. ;-)