http://d.puremagic.com/issues/show_bug.cgi?id=4237

           Summary: Link error with typedefs of the same name in a
                    function scope
           Product: D
           Version: 2.041
          Platform: x86
        OS/Version: All
            Status: NEW
          Keywords: link-failure
          Severity: normal
          Priority: P2
         Component: DMD
        AssignedTo: nobody@puremagic.com
        ReportedBy: rsinfu@gmail.com


--- Comment #0 from Shin Fujishiro <rsinfu@gmail.com> 2010-05-26 12:35:19 PDT ---
Typedefs in function scope don't get fully mangled with scope information.  As a result, the following code won't link:
-------------------- test.d
struct Struct(T)
{
    T value;
}
void main()
{
    {
        typedef int Number = 1;
        Struct!Number s;
        pragma(msg, "1: ", typeof(s).mangleof);
    }
    {
        typedef real Number = 2;
        Struct!Number s;
        pragma(msg, "2: ", typeof(s).mangleof);
    }
}
--------------------

On Linux and FreeBSD:
--------------------
1: S4test20__T6StructTT6NumberZ6Struct
2: S4test20__T6StructTT6NumberZ6Struct
test.o(.rodata+0x18): multiple definition of `_Dmain6Number6__initZ'
test.o(.rodata+0x14): first defined here
--------------------

On Windows:
--------------------
1: S4test20__T6StructTT6NumberZ6Struct
2: S4test20__T6StructTT6NumberZ6Struct
OPTLINK (R) for Win32  Release 8.00.2
Copyright (C) Digital Mars 1989-2009  All rights reserved.
http://www.digitalmars.com/ctg/optlink.html
test.obj(test)  Offset 0027DH Record Type 0091
 Error 1: Previous Definition Different : _Dmain6Number6__initZ
--- errorlevel 1
--------------------

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

http://d.puremagic.com/issues/show_bug.cgi?id=4237


Shin Fujishiro <rsinfu@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch
            Summary|Link error with typedefs of |Typedefs of the same name
                   |the same name in a function |cause initializer conflict
                   |scope                       |


--- Comment #1 from Shin Fujishiro <rsinfu@gmail.com> 2010-05-28 04:51:17 PDT ---
Maybe the example in #0 should be error.  I filed a separate bug 4245.

Still the problem exists.  As shown in the output example below, the two distinct typedefs get mangled to the same name.  Then the assertion fails due to the corrupted initial value.
-------------------- test.d
struct Struct(T) { T value; }
void foo()
{
    typedef int Number = 1;
    Struct!Number s;
    pragma(msg, typeof(s).mangleof);
    assert(s.value == 1);
}
void bar()
{
    typedef real Number = 2;
    Struct!Number s;
    pragma(msg, typeof(s).mangleof);
    assert(s.value == 2); // Assertion failure
}
void main() { foo(); bar(); }
--------------------
% dmd -run test
S4test20__T6StructTT6NumberZ6Struct
S4test20__T6StructTT6NumberZ6Struct
core.exception.AssertError@test(14): Assertion failure
--------------------

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

http://d.puremagic.com/issues/show_bug.cgi?id=4237



--- Comment #2 from Shin Fujishiro <rsinfu@gmail.com> 2010-05-28 04:54:49 PDT ---
Here's a trivial patch to dmd svn r502.

--- src/declaration.c
+++ src/declaration.c
@@ -296,6 +296,7 @@ Dsymbol *TypedefDeclaration::syntaxCopy(Dsymbol *s)
 void TypedefDeclaration::semantic(Scope *sc)
 {
     //printf("TypedefDeclaration::semantic(%s) sem = %d\n", toChars(), sem);
+    parent = sc->parent;
     if (sem == 0)
     {   sem = 1;
         basetype = basetype->semantic(loc, sc);

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

http://d.puremagic.com/issues/show_bug.cgi?id=4237



--- Comment #3 from Shin Fujishiro <rsinfu@gmail.com> 2010-05-28 06:02:42 PDT ---
The comment #2 patch set ->parent everytime = wrong.  This one is better:

--- src/declaration.c
+++ src/declaration.c
@@ -298,6 +298,7 @@ void TypedefDeclaration::semantic(Scope *sc)
     //printf("TypedefDeclaration::semantic(%s) sem = %d\n", toChars(), sem);
     if (sem == 0)
     {   sem = 1;
+        parent = sc->parent;
         basetype = basetype->semantic(loc, sc);
         sem = 2;
 #if DMDV2

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

http://d.puremagic.com/issues/show_bug.cgi?id=4237



--- Comment #4 from Kenji Hara <k.hara.pg@gmail.com> 2011-10-10 19:35:56 PDT ---
(In reply to comment #3)
> The comment #2 patch set ->parent everytime = wrong.  This one is better:
> 
> --- src/declaration.c
> +++ src/declaration.c
> @@ -298,6 +298,7 @@ void TypedefDeclaration::semantic(Scope *sc)
>      //printf("TypedefDeclaration::semantic(%s) sem = %d\n", toChars(), sem);
>      if (sem == 0)
>      {   sem = 1;
> +        parent = sc->parent;
>          basetype = basetype->semantic(loc, sc);
>          sem = 2;
>  #if DMDV2

Pushed as pull request: https://github.com/D-Programming-Language/dmd/pull/445

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

http://d.puremagic.com/issues/show_bug.cgi?id=4237


Walter Bright <bugzilla@digitalmars.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |bugzilla@digitalmars.com
         Resolution|                            |FIXED


--- Comment #5 from Walter Bright <bugzilla@digitalmars.com> 2011-10-11 15:01:46 PDT ---
https://github.com/D-Programming-Language/dmd/commit/1cc61cbccf64c65c3257326a1df02519c06808c1

https://github.com/D-Programming-Language/dmd/commit/a6586474a8cc7ae92f13afabfeffcaa5ac085d1d

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