www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 6768] New: Problem with init of struct members in presence of templated opAssign

reply d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=6768

           Summary: Problem with init of struct members in presence of
                    templated opAssign
           Product: D
           Version: D2
          Platform: Other
        OS/Version: Windows
            Status: NEW
          Severity: critical
          Priority: P2
         Component: DMD
        AssignedTo: nobody puremagic.com
        ReportedBy: andrej.mitrovich gmail.com


--- Comment #0 from Andrej Mitrovic <andrej.mitrovich gmail.com> 2011-10-04
19:14:34 PDT ---
module test;

import std.traits;
import std.stdio;

struct Point 
{
    int x, y;
    void delegate() dg;

    // void opAssign(void delegate() rhs)  // ok, x and y initialized
    // {
    //     dg = rhs;
    // }

    void opAssign(T)(T rhs) if (isDelegate!T)  // x and y left uninitialized
    {
        dg = rhs;
    }

    void test() { dg(); }
}

class Foo
{
    this() 
    { 
        point = { writefln("Point: %s", point); };  // assign delegate
    }

    Point point;
}

void main()
{
    auto foo = new Foo;
    foo.point.dg();    // x and y are initialized
    foo.point.test();  // but here x and y are not initialized (??)
    foo.point.dg();    // again, not initialized (??)
}



I don't understand how calling dg() directly or indirectly via test() prints
different results for x and y. If I use the non-templated version of opAssign
then both calls are fine, with x and y being zero-inited.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 04 2011
next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=6768



--- Comment #1 from Andrej Mitrovic <andrej.mitrovich gmail.com> 2012-01-04
06:49:54 PST ---
In 2.057 all calls to the `dg` delegate print garbage values if the templated
opAssign was used:

Point: Point(4202703, 4931824, void delegate())
Point: Point(1244728, 4202612, void delegate())
Point: Point(4202726, 4931824, void delegate())

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 04 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=6768


Stewart Gordon <smjg iname.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code
                 CC|                            |smjg iname.com
           Platform|Other                       |All
             Blocks|                            |2573


--- Comment #2 from Stewart Gordon <smjg iname.com> 2012-02-12 13:05:02 PST ---
(In reply to comment #1)

 In 2.057 all calls to the `dg` delegate print garbage values if the templated
 opAssign was used:


Not for me (2.057 Win32):
-----
C:\Users\Stewart\Documents\Programming\D\Tests\bugs>bz6768
Point: Point(0, 0, void delegate())
Point: Point(1244752, 4202612, void delegate())
Point: Point(1244752, 4202612, void delegate())
-----
It's especially puzzling because after construction of foo, our code hasn't
changed either foo or foo.point, but something behind the scenes has caused it
to change.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 12 2012
prev sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=6768


Denis Shelomovskij <verylonglogin.reg gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |verylonglogin.reg gmail.com
         Resolution|                            |DUPLICATE


--- Comment #3 from Denis Shelomovskij <verylonglogin.reg gmail.com> 2012-11-12
12:25:48 MSK ---
The issue isn't with `opAssign`. It is because closure isn't detected:
---
import std.stdio;

void setDel()(Foo foo, void delegate() del)
{ foo.del = del; }

class Foo
{
    void delegate() del;

    void f()
    {
        writefln("%X, %X (instance and ref addresses)", cast(void*) this,
&this);
        void g() { writefln("%X, %X (instance and ref addresses from g)",
cast(void*) this, &this); }
        setDel(this, &g);
        writefln("%X (del.ptr)", del.ptr);
    }

    void callDel()
    {
        writefln("+callDel");
        del();
        writefln("-callDel");
    }
}

void main()
{
    auto foo = new Foo();
    foo.f();
    foo.del();
    foo.callDel();
    foo.del();
}
---

Output:
---
A01E70, 12FE58 (instance and ref addresses)
12FE58 (del.ptr)
12FE58, 12FE58 (instance and ref addresses from g)
+callDel
A01E70, 12FE58 (instance and ref addresses from g)
-callDel
12FE58, 12FE58 (instance and ref addresses from g)
---

Created a new issue with corrected title and description.

*** This issue has been marked as a duplicate of issue 8999 ***

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Nov 12 2012