www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - Compiler bug ?

reply "Temtaime" <temtaime gmail.com> writes:
Hello guys!

It seems that it is bug. And critical for me.
http://dpaste.1azy.net/b93f5776

Regards.
Jun 09 2013
next sibling parent reply "Kenji Hara" <k.hara.pg gmail.com> writes:
On Sunday, 9 June 2013 at 13:37:45 UTC, Temtaime wrote:

 Hello guys!

 It seems that it is bug. And critical for me.
 http://dpaste.1azy.net/b93f5776

 Regards.


The code should be rejected. It's a compiler bug.
But, you can do what you want as follows.

import std.stdio, std.typetuple;

struct A {
      ("1") int a;
      ("2") int b;
      ("3") int c;
}

auto bar(alias expr)()  property {
     writeln(expr);
}

void main() {
     A a;

     foreach(it; __traits(allMembers, A))
     {
         enum tp = [__traits(getAttributes, mixin(`a.` ~ it))];
         // or
         alias tp = TypeTuple!(__traits(getAttributes, mixin(`a.` 
~ it)));

         writeln(`passing: `, tp[0]);
         bar!(tp[0])();
     }
}

Kenji Hara
Jun 09 2013
parent "Temtaime" <temtaime gmail.com> writes:
Oh, i see.
Great thanks!
Jun 09 2013
prev sibling parent reply "Juan Manuel Cabo" <juanmanuel.cabo gmail.com> writes:
On Sunday, 9 June 2013 at 13:37:45 UTC, Temtaime wrote:

 Hello guys!

 It seems that it is bug. And critical for me.
 http://dpaste.1azy.net/b93f5776

 Regards.



I don't know whether it's a bug, but I managed to reduce the error
to the following code (removing __traits code):

     import std.stdio;
     import std.typetuple;

     void bar(alias expr)() {
         writeln(expr);
     }

     void main() {
         foreach(it; TypeTuple!("a","b","c")) {
             auto tp = TypeTuple!(it);

             bar!(tp[0])();
             writeln(`should be: `, tp[0]);
         }
     }


This prints:

     a
     should be: a
     a
     should be: b
     a
     should be: c

--jm
Jun 09 2013
parent "Temtaime" <temtaime gmail.com> writes:
Yes, it's strange. Using alias doesn't lead to this.
Jun 09 2013