http://d.puremagic.com/issues/show_bug.cgi?id=2460

           Summary: Ref functions can't be template functions.
           Product: D
           Version: 2.020
          Platform: All
        OS/Version: All
            Status: NEW
          Keywords: rejects-valid
          Severity: minor
          Priority: P4
         Component: DMD
        AssignedTo: bugzilla@digitalmars.com
        ReportedBy: hskwk@inter7.jp


Ref functions can't be template functions

I guess DMD parser failed to bind `ref' to ref functions if the template is
written in function template syntax.
It is not serious problem for me.


real t;
ref real f(S)(S rhs)
{
        size_t tmp;
        return t;
}

template g(S)
{
        ref real g(S rhs)
        {
                size_t tmp;
                return t;
        }
}

f(1.0);
// variable f!(double).f.tmp only parameters or foreach declarations can be ref

g(1.0)
// no problem


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=2460


Lars T. Kyllingstad <bugzilla@kyllingen.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bugzilla@kyllingen.net
           Severity|minor                       |major


--- Comment #1 from Lars T. Kyllingstad <bugzilla@kyllingen.net> 2010-03-27 11:27:51 PDT ---
Raising importance, since operator overloading is now done with templated functions, and this will become a major pain in the long run.

-- 
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=2460


Masahiro Nakagawa <repeatedly@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |repeatedly@gmail.com


--- Comment #2 from Masahiro Nakagawa <repeatedly@gmail.com> 2010-05-24 08:59:21 PDT ---
Today, I hit the similar problem.

In the latest dmd, above code works but variadic version doesn't works.

struct Foo
{
    // works! I use this for workaround
    template foo(Args...) if (true)
    {
        ref Foo foo(Args args)
        {
            return this;
        }
    }

    // doesn't work. I want use this!!
    ref Foo foo(Args...)(Args args) if (true)
    {
        return this;
    }

    // works!
    ref Foo foo(Args...)(Args args)
    {
        return this;
    }
}

-- 
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=2460


kennytm@gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |kennytm@gmail.com
         Resolution|                            |FIXED


--- Comment #3 from kennytm@gmail.com 2011-04-23 09:31:02 PDT ---
Looks like this has been fixed. All programs in this page compiles for me.

--------
struct Foo {
    double x, y;

    ref double get(Args...)(Args args) if (args.length == 4) {
        return x;
    }
}
void main() {
    Foo f;
    f.get(4,5,6,7) = 6;
    assert(f.x == 6);
}
--------

Can we fix this std.complex bug then?

"Some operators, such as opAssign and opOpAssign, should return by ref, but currently don't. This will be implemented as soon as DMD bug 2460 is fixed."

-- 
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=2460



--- Comment #4 from Lars T. Kyllingstad <bugzilla@kyllingen.net> 2011-04-26 02:40:59 PDT ---
Cool! I'll fix std.complex.

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