void func(string[2] a) {}

void func2(T...)(T args) {
   static assert(is(typeof(args[0]) == string[2]));
}

void func3(T...)(T args) {
   static assert(args[0].length == 2);
}

func(["",""]); // Okay
func2(["",""]); //  Error: (is(string[] == string[2LU])) is false
func3(["",""]); // Error: _param_0 cannot be read at compile time

Is this the intended design? Is there a workaround which allows me to ensure that the parameter is exactly 2 length?

On Tuesday, 12 March 2013 at 21:47:02 UTC, Zach the Mystic wrote:
> void func(string[2] a) {}
>
> void func2(T...)(T args) {
>    static assert(is(typeof(args[0]) == string[2]));
> }
>
> void func3(T...)(T args) {
>    static assert(args[0].length == 2);
> }
>
> func(["",""]); // Okay
> func2(["",""]); //  Error: (is(string[] == string[2LU])) is false
> func3(["",""]); // Error: _param_0 cannot be read at compile time
>
> Is this the intended design? Is there a workaround which allows me to ensure that the parameter is exactly 2 length?

I'm just going to assert at runtime instead of at compile time.
assert(args[0].length == 2);

That's a good workaround, only a minor inconvenience it can't be caught at compile time.

On 03/12/2013 04:10 PM, Zach the Mystic wrote:
> On Tuesday, 12 March 2013 at 21:47:02 UTC, Zach the Mystic wrote:
>> void func(string[2] a) {}
>>
>> void func2(T...)(T args) {
>> static assert(is(typeof(args[0]) == string[2]));
>> }
>>
>> void func3(T...)(T args) {
>> static assert(args[0].length == 2);
>> }
>>
>> func(["",""]); // Okay
>> func2(["",""]); // Error: (is(string[] == string[2LU])) is false
>> func3(["",""]); // Error: _param_0 cannot be read at compile time
>>
>> Is this the intended design? Is there a workaround which allows me to
>> ensure that the parameter is exactly 2 length?
>
> I'm just going to assert at runtime instead of at compile time.
> assert(args[0].length == 2);
>
> That's a good workaround, only a minor inconvenience it can't be caught
> at compile time.

Actually, this is not about variadic function templates. The type of an array literal is a slice:

    static assert(is(typeof(["",""]) == string[]));

This doesn't help you but it is possible to cast:

    static assert(is(typeof(cast(string[2])["",""]) == string[2]));

or use a variable with an explicit type:

    string[2] var = ["", ""];
    static assert(is(typeof(var) == string[2]));

Ali

On Tuesday, 12 March 2013 at 23:35:28 UTC, Ali Çehreli wrote:
> On 03/12/2013 04:10 PM, Zach the Mystic wrote:
> > On Tuesday, 12 March 2013 at 21:47:02 UTC, Zach the Mystic
> wrote:
> >> void func(string[2] a) {}
> >>
> >> void func2(T...)(T args) {
> >> static assert(is(typeof(args[0]) == string[2]));
> >> }
> >>
> >> void func3(T...)(T args) {
> >> static assert(args[0].length == 2);
> >> }
> >>
> >> func(["",""]); // Okay
> >> func2(["",""]); // Error: (is(string[] == string[2LU])) is
> false
> >> func3(["",""]); // Error: _param_0 cannot be read at compile
> time
> >>
> >> Is this the intended design? Is there a workaround which
> allows me to
> >> ensure that the parameter is exactly 2 length?
> >
> > I'm just going to assert at runtime instead of at compile
> time.
> > assert(args[0].length == 2);
> >
> > That's a good workaround, only a minor inconvenience it can't
> be caught
> > at compile time.
>
> Actually, this is not about variadic function templates. The type of an array literal is a slice:
>
>     static assert(is(typeof(["",""]) == string[]));
>
> This doesn't help you but it is possible to cast:
>
>     static assert(is(typeof(cast(string[2])["",""]) == string[2]));
>
> or use a variable with an explicit type:
>
>     string[2] var = ["", ""];
>     static assert(is(typeof(var) == string[2]));
>
> Ali

Thanks. I see that the type is a slice, but it doesn't stop func(string[2]) from taking ["",""] at compile time. I think it might have been felt to be problematic not to allow it. I think it's inconsistent though. Might be harder to include in complex parameter deduction than in explicit parameter lists.

On 03/12/2013 10:47 PM, Zach the Mystic wrote:
> void func(string[2] a) {}
>
> void func2(T...)(T args) {
>     static assert(is(typeof(args[0]) == string[2]));
> }
>
> void func3(T...)(T args) {
>     static assert(args[0].length == 2);
> }
>
> func(["",""]); // Okay
> func2(["",""]); //  Error: (is(string[] == string[2LU])) is false
> func3(["",""]); // Error: _param_0 cannot be read at compile time
>
> Is this the intended design?

Yes.

> Is there a workaround which allows me to
> ensure that the parameter is exactly 2 length?

http://d.puremagic.com/issues/show_bug.cgi?id=9712
http://d.puremagic.com/issues/show_bug.cgi?id=9711

On Tue, 12 Mar 2013 17:47:00 -0400, Zach the Mystic <reachzach@gggggmail.com> wrote:

> void func(string[2] a) {}
>
> void func2(T...)(T args) {
>     static assert(is(typeof(args[0]) == string[2]));
> }
>
> void func3(T...)(T args) {
>     static assert(args[0].length == 2);
> }
>
> func(["",""]); // Okay
> func2(["",""]); //  Error: (is(string[] == string[2LU])) is false
> func3(["",""]); // Error: _param_0 cannot be read at compile time
>
> Is this the intended design? Is there a workaround which allows me to ensure that the parameter is exactly 2 length?

Actually an array literal's default type is a slice, not a fixed-sized array.  With IFTI, the type is decided by the call, not how you use it the parameters.

I filed an enhancement request a while back that would help fix this.

http://d.puremagic.com/issues/show_bug.cgi?id=4998

-Steve

On Wednesday, 13 March 2013 at 22:14:26 UTC, Timon Gehr wrote:
> On 03/12/2013 10:47 PM, Zach the Mystic wrote:
>> void func(string[2] a) {}
>>
>> void func2(T...)(T args) {
>>    static assert(is(typeof(args[0]) == string[2]));
>> }
>>
>> void func3(T...)(T args) {
>>    static assert(args[0].length == 2);
>> }
>>
>> func(["",""]); // Okay
>> func2(["",""]); //  Error: (is(string[] == string[2LU])) is false
>> func3(["",""]); // Error: _param_0 cannot be read at compile time
>>
>> Is this the intended design?
>
> Yes.
>
>> Is there a workaround which allows me to
>> ensure that the parameter is exactly 2 length?
>
> http://d.puremagic.com/issues/show_bug.cgi?id=9712
> http://d.puremagic.com/issues/show_bug.cgi?id=9711

Man, Hara Kenji is so fast!!!!!! I didn't even know my problem wasn't just part of the design, and already it's close to being a fixed bug/enhancement!

On Thursday, 14 March 2013 at 20:59:38 UTC, Zach the Mystic wrote:
> On Wednesday, 13 March 2013 at 22:14:26 UTC, Timon Gehr wrote:
>> On 03/12/2013 10:47 PM, Zach the Mystic wrote:
>>> void func(string[2] a) {}
>>>
>>> void func2(T...)(T args) {
>>>   static assert(is(typeof(args[0]) == string[2]));
>>> }
>>>
>>> void func3(T...)(T args) {
>>>   static assert(args[0].length == 2);
>>> }
>>>
>>> func(["",""]); // Okay
>>> func2(["",""]); //  Error: (is(string[] == string[2LU])) is false
>>> func3(["",""]); // Error: _param_0 cannot be read at compile time
>>>
>>> Is this the intended design?
>>
>> Yes.
>>
>>> Is there a workaround which allows me to
>>> ensure that the parameter is exactly 2 length?
>>
>> http://d.puremagic.com/issues/show_bug.cgi?id=9712
>> http://d.puremagic.com/issues/show_bug.cgi?id=9711
>
> Man, Hara Kenji is so fast!!!!!! I didn't even know my problem wasn't just part of the design, and already it's close to being a fixed bug/enhancement!

(I meant to thank you too, Timon Gehr)