http://d.puremagic.com/issues/show_bug.cgi?id=1462

           Summary: Templated constructor not supported
           Product: D
           Version: 2.003
          Platform: PC
        OS/Version: Windows
            Status: NEW
          Severity: normal
          Priority: P2
         Component: DMD
        AssignedTo: bugzilla@digitalmars.com
        ReportedBy: reiner.pope@gmail.com


I know of no way to create a templated constructor for a class. (This is useful, for instance, with variadic template args). The following code fails to parse in compilation:

class Foo
{
    this(T...)(T t) {}
}


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1462


smjg@iname.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |smjg@iname.com
           Severity|normal                      |enhancement
           Keywords|                            |spec




------- Comment #1 from smjg@iname.com  2008-09-09 12:32 -------
The problem seems to be that the current D syntax doesn't support it:

Constructor:
        this Parameters FunctionBody

Parameters:
        ( ParameterList )
        ( )

The alternative would be

class Foo
{
    template(T...) this {
        this(T t) {}
    }
}

except that this doesn't work because 'this' is a keyword, not an identifier.


-- 

On Tue, Sep 9, 2008 at 1:32 PM,  <d-bugmail@puremagic.com> wrote:
> http://d.puremagic.com/issues/show_bug.cgi?id=1462
>
>
> smjg@iname.com changed:
>
>           What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                 CC|                            |smjg@iname.com
>           Severity|normal                      |enhancement
>           Keywords|                            |spec
>
>
>
>
> ------- Comment #1 from smjg@iname.com  2008-09-09 12:32 -------
> The problem seems to be that the current D syntax doesn't support it:
>
> Constructor:
>        this Parameters FunctionBody
>
> Parameters:
>        ( ParameterList )
>        ( )
>
> The alternative would be
>
> class Foo
> {
>    template(T...) this {
>        this(T t) {}
>    }
> }
>
> except that this doesn't work because 'this' is a keyword, not an identifier.
>
>
> --
>
>

I think it's much more that templated methods cannot be virtual, and constructors more or less have to be.

On Tue, 09 Sep 2008 22:48:01 +0400, Jarrett Billingsley <jarrett.billingsley@gmail.com> wrote:

> On Tue, Sep 9, 2008 at 1:32 PM,  <d-bugmail@puremagic.com> wrote:
>> http://d.puremagic.com/issues/show_bug.cgi?id=1462
>>
>>
>> smjg@iname.com changed:
>>
>>           What    |Removed                     |Added
>> ----------------------------------------------------------------------------
>>                 CC|                            |smjg@iname.com
>>           Severity|normal                      |enhancement
>>           Keywords|                            |spec
>>
>>
>>
>>
>> ------- Comment #1 from smjg@iname.com  2008-09-09 12:32 -------
>> The problem seems to be that the current D syntax doesn't support it:
>>
>> Constructor:
>>        this Parameters FunctionBody
>>
>> Parameters:
>>        ( ParameterList )
>>        ( )
>>
>> The alternative would be
>>
>> class Foo
>> {
>>    template(T...) this {
>>        this(T t) {}
>>    }
>> }
>>
>> except that this doesn't work because 'this' is a keyword, not an identifier.
>>
>>
>> --
>>
>>
>
> I think it's much more that templated methods cannot be virtual, and
> constructors more or less have to be.

How about struct constructors?

Jarrett Billingsley wrote:
> 
> I think it's much more that templated methods cannot be virtual, and
> constructors more or less have to be.

How are constructors virtual?


Sean

On Tue, Sep 9, 2008 at 3:55 PM, Sean Kelly <sean@invisibleduck.org> wrote:
> Jarrett Billingsley wrote:
>>
>> I think it's much more that templated methods cannot be virtual, and constructors more or less have to be.
>
> How are constructors virtual?

Uhh umm

Well as it is now, constructors are treated as methods of the class, so they participate in overloading.  But I think you're right - it doesn't really seem to have to be that way.

Jarrett Billingsley wrote:
> On Tue, Sep 9, 2008 at 3:55 PM, Sean Kelly <sean@invisibleduck.org> wrote:
>> Jarrett Billingsley wrote:
>>> I think it's much more that templated methods cannot be virtual, and
>>> constructors more or less have to be.
>> How are constructors virtual?
> 
> Uhh umm
> 
> Well as it is now, constructors are treated as methods of the class,
> so they participate in overloading.  But I think you're right - it
> doesn't really seem to have to be that way.

That makes no sense :p . Not only are they not virtual, they could never be virtual. Virtual method dispatch is used when you have polymorphism and don't know the exact class of the receiving object. But when constructing you always know the exact class being creating, so you would always know which method/constructor/whatever to dispatch to.


-- 
Bruno Medeiros - Software Developer, MSc. in CS/E graduate
http://www.prowiki.org/wiki4d/wiki.cgi?BrunoMedeiros#D

On Thu, Sep 11, 2008 at 5:53 AM, Bruno Medeiros <brunodomedeiros+spam@com.gmail> wrote:
> Jarrett Billingsley wrote:
>>
>> On Tue, Sep 9, 2008 at 3:55 PM, Sean Kelly <sean@invisibleduck.org> wrote:
>>>
>>> Jarrett Billingsley wrote:
>>>>
>>>> I think it's much more that templated methods cannot be virtual, and constructors more or less have to be.
>>>
>>> How are constructors virtual?
>>
>> Uhh umm
>>
>> Well as it is now, constructors are treated as methods of the class, so they participate in overloading.  But I think you're right - it doesn't really seem to have to be that way.
>
> That makes no sense :p . Not only are they not virtual, they could never be virtual. Virtual method dispatch is used when you have polymorphism and don't know the exact class of the receiving object. But when constructing you always know the exact class being creating, so you would always know which method/constructor/whatever to dispatch to.

Then beat Walter over the head!

Jarrett Billingsley wrote:
> On Thu, Sep 11, 2008 at 5:53 AM, Bruno Medeiros
> <brunodomedeiros+spam@com.gmail> wrote:
>> Jarrett Billingsley wrote:
>>> On Tue, Sep 9, 2008 at 3:55 PM, Sean Kelly <sean@invisibleduck.org> wrote:
>>>> Jarrett Billingsley wrote:
>>>>> I think it's much more that templated methods cannot be virtual, and
>>>>> constructors more or less have to be.
>>>> How are constructors virtual?
>>> Uhh umm
>>>
>>> Well as it is now, constructors are treated as methods of the class,
>>> so they participate in overloading.  But I think you're right - it
>>> doesn't really seem to have to be that way.
>> That makes no sense :p . Not only are they not virtual, they could never be
>> virtual. Virtual method dispatch is used when you have polymorphism and
>> don't know the exact class of the receiving object. But when constructing
>> you always know the exact class being creating, so you would always know
>> which method/constructor/whatever to dispatch to.
> 
> Then beat Walter over the head!

Apparently my cluebat doesn't reach all the way to Seattle! -_-'

-- 
Bruno Medeiros - Software Developer, MSc. in CS/E graduate
http://www.prowiki.org/wiki4d/wiki.cgi?BrunoMedeiros#D

http://d.puremagic.com/issues/show_bug.cgi?id=1462


Steven Schveighoffer <schveiguy@yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |schveiguy@yahoo.com
         Resolution|                            |DUPLICATE


--- Comment #2 from Steven Schveighoffer <schveiguy@yahoo.com> 2010-08-05 10:57:21 PDT ---
*** This issue has been marked as a duplicate of issue 435 ***

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