I am allowed to create a type tuple of function templates, but then it seems I am not allowed to use a static foreach on that type tuple:


import std.typetuple;

int foo(T)(T x) {
    return 0;
}

void main() {
    alias TypeTuple!(foo, foo) t2;
    static assert(t2[0](0) == 0); // OK
    static assert(t2[1](0) == 0); // OK
    foreach (t; t2) {} // Error
}


Is this expected, a bug in my code, a known bug in DMD, a new DMD bug, a limitation meant to be removed, etc?

Bye,
bearophile

On 2011-03-04 04:37, bearophile wrote:
> I am allowed to create a type tuple of function templates, but then it seems I am not allowed to use a static foreach on that type tuple:
>
>
> import std.typetuple;
>
> int foo(T)(T x) {
>      return 0;
> }
>
> void main() {
>      alias TypeTuple!(foo, foo) t2;
>      static assert(t2[0](0) == 0); // OK
>      static assert(t2[1](0) == 0); // OK
>      foreach (t; t2) {} // Error
> }
>
>
> Is this expected, a bug in my code, a known bug in DMD, a new DMD bug, a limitation meant to be removed, etc?
>
> Bye,
> bearophile

Maybe you can try something like typeof(t2) in the foreach.

-- 
/Jacob Carlborg

Jacob Carlborg:

> Maybe you can try something like typeof(t2) in the foreach.

If you mean code like this:

import std.typetuple;

int foo(T)(T x) {
     return x;
}

void main() {
     alias TypeTuple!(foo, foo) t2;
     foreach (i, t; typeof(t2)) {
         pragma(msg, t);
         assert(t2[i](i) == i); // OK
         //assert(t(i) == i); // Not OK
    }
}

It seems "t" is not useful...

Is all this Bugzilla-worthy?

Thank you, bye,
bearophile

bearophile Wrote:
> It seems "t" is not useful...

Yep, foreach over tuples is kinda messed up.
Same with reference tuple foreach: http://d.puremagic.com/issues/show_bug.cgi?id=2411
Direct access doesn't work but accessing via [i] does.

On 2011-03-04 13:19, bearophile wrote:
> Jacob Carlborg:
>
>> Maybe you can try something like typeof(t2) in the foreach.
>
> If you mean code like this:
>
> import std.typetuple;
>
> int foo(T)(T x) {
>       return x;
> }
>
> void main() {
>       alias TypeTuple!(foo, foo) t2;
>       foreach (i, t; typeof(t2)) {
>           pragma(msg, t);
>           assert(t2[i](i) == i); // OK
>           //assert(t(i) == i); // Not OK
>      }
> }
>
> It seems "t" is not useful...

Yeah, that's really annoying.

> Is all this Bugzilla-worthy?
>
> Thank you, bye,
> bearophile



-- 
/Jacob Carlborg

bearophile <bearophileHUGS@lycos.com> wrote:

> Jacob Carlborg:
>
>> Maybe you can try something like typeof(t2) in the foreach.
>
> If you mean code like this:
>
> import std.typetuple;
>
> int foo(T)(T x) {
>      return x;
> }
>
> void main() {
>      alias TypeTuple!(foo, foo) t2;
>      foreach (i, t; typeof(t2)) {
>          pragma(msg, t);
>          assert(t2[i](i) == i); // OK
>          //assert(t(i) == i); // Not OK
>     }
> }
>
> It seems "t" is not useful...
>
> Is all this Bugzilla-worthy?

Yes.

-- 
Simen