www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Can not overload template method function with const. Bug?

reply "Tobias Pankrath" <tobias pankrath.net> writes:
You currently can't use Tuples with std.container.Array (Bug 


To fix this, I tried to overload the opEquals(R)(R rhs) of Tuple 
with
a const version opEquals(R)(R rhs) const. But the compiler does 
not use the overload if opEquals is called on a const tuple 
instance.

See this thread in d.learn 
http://forum.dlang.org/thread/vghlctzlpqxhoeojqqxt forum.dlang.org

The orignal bug is solved, if you mark the current opEquals 
const, but I don't know if this would brake code outside of 
phobos, so I tried the overload.

Coming to the question: Is it a bug that the const overload is 
not considered and what is the correct way to fix 
std.typecons.Tuple?
Aug 08 2012
next sibling parent reply kenji hara <k.hara.pg gmail.com> writes:
In the declaration of std.typecons.Tuple:

    bool opEquals(R)(R rhs) if (isTuple!R)
    {
        // [snip]
    }

    bool opEquals(R)(R rhs) const if (isTuple!R)
    {
        // same as non-const version
    }

Adding const version works correctly.
What's the problem?

Kenji Hara

2012/8/8 Tobias Pankrath <tobias pankrath.net>:



 To fix this, I tried to overload the opEquals(R)(R rhs) of Tuple with
 a const version opEquals(R)(R rhs) const. But the compiler does not use the
 overload if opEquals is called on a const tuple instance.

 See this thread in d.learn
 http://forum.dlang.org/thread/vghlctzlpqxhoeojqqxt forum.dlang.org

 The orignal bug is solved, if you mark the current opEquals const, but I
 don't know if this would brake code outside of phobos, so I tried the
 overload.

 Coming to the question: Is it a bug that the const overload is not
 considered and what is the correct way to fix std.typecons.Tuple?
Aug 08 2012
parent reply "Tobias Pankrath" <tobias pankrath.net> writes:
On Wednesday, 8 August 2012 at 15:01:23 UTC, kenji hara wrote:

 In the declaration of std.typecons.Tuple:

     bool opEquals(R)(R rhs) if (isTuple!R)
     {
         // [snip]
     }

     bool opEquals(R)(R rhs) const if (isTuple!R)
     {
         // same as non-const version
     }

 Adding const version works correctly.
 What's the problem?

 Kenji Hara


That's exactly what I've tried first. Using a git clone from 
yesterday and within phobos directory, using


 make -f posix.mak


everything works. Now I'm adding the const version of opEquals 
and I get: http://pastebin.com/akRdFfAJ

But it's similar with the reduced example from the thread in 
d.learn. Does that work for you?

I'm using dmd cef1bbfdcd9282934ff6f1b07617254753334799 and phobos 
90eb9313030340dc274d2fd82144a7f30d0e4ae5
Aug 08 2012
parent reply kenji hara <k.hara.pg gmail.com> writes:
2012/8/9 Tobias Pankrath <tobias pankrath.net>:

 On Wednesday, 8 August 2012 at 15:01:23 UTC, kenji hara wrote:

 In the declaration of std.typecons.Tuple:

     bool opEquals(R)(R rhs) if (isTuple!R)
     {
         // [snip]
     }

     bool opEquals(R)(R rhs) const if (isTuple!R)
     {
         // same as non-const version
     }

 Adding const version works correctly.
 What's the problem?

 Kenji Hara


 That's exactly what I've tried first. Using a git clone from yesterday and
 within phobos directory, using


 make -f posix.mak



 everything works. Now I'm adding the const version of opEquals and I get:
 http://pastebin.com/akRdFfAJ

 But it's similar with the reduced example from the thread in d.learn. Does
 that work for you?


I've tried following code now, and it is exactly the bug 8522.

struct Point
{
    bool opEquals(R)(R rhs) { return true; }
    bool opEquals(R)(R rhs) const { return true; }
}
void main()
{
    Point mp;
    const Point cp;
    assert(mp == mp);
    assert(mp == cp);
    assert(cp == mp);
    assert(cp == cp);
}

This code doesn't work with current git head of dmd, but works with my patch.
https://github.com/D-Programming-Language/dmd/pull/1075

Kenji Hara
Aug 08 2012
parent "Tobias Pankrath" <tobias pankrath.net> writes:
On Wednesday, 8 August 2012 at 15:54:43 UTC, kenji hara wrote:

 I've tried following code now, and it is exactly the bug 8522.

 struct Point
 {
     bool opEquals(R)(R rhs) { return true; }
     bool opEquals(R)(R rhs) const { return true; }
 }
 void main()
 {
     Point mp;
     const Point cp;
     assert(mp == mp);
     assert(mp == cp);
     assert(cp == mp);
     assert(cp == cp);
 }

 This code doesn't work with current git head of dmd, but works 
 with my patch.
 https://github.com/D-Programming-Language/dmd/pull/1075


Thank you. Your patch works for me as well.
Aug 08 2012
prev sibling parent reply kenji hara <k.hara.pg gmail.com> writes:
I found a dmd bug. Is this the bug which you have seen?
http://d.puremagic.com/issues/show_bug.cgi?id=8522

Kenji Hara

2012/8/9 kenji hara <k.hara.pg gmail.com>:

 In the declaration of std.typecons.Tuple:

     bool opEquals(R)(R rhs) if (isTuple!R)
     {
         // [snip]
     }

     bool opEquals(R)(R rhs) const if (isTuple!R)
     {
         // same as non-const version
     }

 Adding const version works correctly.
 What's the problem?

 Kenji Hara

 2012/8/8 Tobias Pankrath <tobias pankrath.net>:



 To fix this, I tried to overload the opEquals(R)(R rhs) of Tuple with
 a const version opEquals(R)(R rhs) const. But the compiler does not use the
 overload if opEquals is called on a const tuple instance.

 See this thread in d.learn
 http://forum.dlang.org/thread/vghlctzlpqxhoeojqqxt forum.dlang.org

 The orignal bug is solved, if you mark the current opEquals const, but I
 don't know if this would brake code outside of phobos, so I tried the
 overload.

 Coming to the question: Is it a bug that the const overload is not
 considered and what is the correct way to fix std.typecons.Tuple?
Aug 08 2012
parent "Tobias Pankrath" <tobias pankrath.net> writes:
On Wednesday, 8 August 2012 at 15:25:04 UTC, kenji hara wrote:

 I found a dmd bug. Is this the bug which you have seen?
 http://d.puremagic.com/issues/show_bug.cgi?id=8522


Using postfix const.
Aug 08 2012