www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 6382] New: edge case with static foreach

http://d.puremagic.com/issues/show_bug.cgi?id=6382

           Summary: edge case with static foreach
           Product: D
           Version: D2
          Platform: Other
        OS/Version: Windows
            Status: NEW
          Severity: normal
          Priority: P2
         Component: DMD
        AssignedTo: nobody puremagic.com
        ReportedBy: andrej.mitrovich gmail.com



05:33:56 PDT ---
import std.typetuple;

struct Foo
{   
    this(void delegate() dg) { }
}

void test()
{
    Foo[] result;

    foreach (Type; TypeTuple!(int, int))
    {
        result ~= Foo( (){} );
    }
}

void main()
{
    test();
}

OPTLINK (R) for Win32  Release 8.00.12
Copyright (C) Digital Mars 1989-2010  All rights reserved.
http://www.digitalmars.com/ctg/optlink.html
C:\DOCUME~1\Andrej\LOCALS~1\Temp\.rdmd\rdmd-ctfe_bug22.d-97E30C11895AB443DA692FB4CE11A18B\ctfe_bug22-d-97E30C11895AB443DA692FB4CE11A18B.obj(ctfe_bug22-d-97E30C11895AB443DA692FB4CE11A18B)
 Offset 00960H Record Type 00C3
 Error 1: Previous Definition Different :
_D12createFields4testFZv12__dgliteral1MFZv

It seems as if the compiler generates two delegates with the same name inside
of test(), and they end up clashing. Of course, this function should be called
at compile time, not runtime. The fix is to use it as a template:

import std.typetuple;

struct Foo
{   
    this(void delegate() dg) { }
}

void test()()  // template func
{
    Foo[] result;

    foreach (Type; TypeTuple!(int, int))
    {
        result ~= Foo( (){} );
    }
}

void main()
{
    test!();  // call it as a template
}

I don't know if the first case is accepts-invalid or completely legal, but
linker errors are never nice to see. Anyway I thought this was worth putting
here.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jul 26 2011