"init" and other class properties are kind of shadow-keywords.

class MyClass{
  void init(){
  }
}

The class is fine and can be used.
Now someone attempts to insert this class into a container....

template Templ(T){
  T t = T.init; // Compile error
}

The compiler should forbid the use of the method name "init", if the signature does not make any sense for an "init" property. It must be a static method, with a return value and no arguments.

"Frank Benoit" <keinfarbton@googlemail.com> wrote in message news:g052dt$qt4$1@digitalmars.com...
> "init" and other class properties are kind of shadow-keywords.
>
> class MyClass{
>   void init(){
>   }
> }
>
> The class is fine and can be used.
> Now someone attempts to insert this class into a container....
>
> template Templ(T){
>   T t = T.init; // Compile error
> }
>
> The compiler should forbid the use of the method name "init", if the signature does not make any sense for an "init" property. It must be a static method, with a return value and no arguments.

This should extend to any semi-reserved name.  The compiler used to allow redefinition of 'sizeof' and 'offsetof' but checks were added to disallow that.  That mechanism should just be extended.

在 Sun, 11 May 2008 05:46:57 +0800，Jarrett Billingsley <kb3ctd2@yahoo.com> 写道:

> "Frank Benoit" <keinfarbton@googlemail.com> wrote in message
> news:g052dt$qt4$1@digitalmars.com...
>> "init" and other class properties are kind of shadow-keywords.
>>
>> class MyClass{
>>   void init(){
>>   }
>> }
>>
>> The class is fine and can be used.
>> Now someone attempts to insert this class into a container....
>>
>> template Templ(T){
>>   T t = T.init; // Compile error
>> }
>>
>> The compiler should forbid the use of the method name "init", if the
>> signature does not make any sense for an "init" property. It must be a
>> static method, with a return value and no arguments.
>
> This should extend to any semi-reserved name.  The compiler used to allow
> redefinition of 'sizeof' and 'offsetof' but checks were added to disallow
> that.  That mechanism should just be extended.
>
>

I filed a bug of forbiding the use of stringof method. But it's markd as invalid :(
http://d.puremagic.com/issues/show_bug.cgi?id=1412

-- 
使用 Opera 革命性的电子邮件客户程序: http://www.opera.com/mail/

davidl@126.com wrote:
> ÔÚ Sun, 11 May 2008 05:46:57 +0800£¬Jarrett Billingsley <kb3ctd2@yahoo.com> дµÀ:
> 
>> "Frank Benoit" <keinfarbton@googlemail.com> wrote in message
>> news:g052dt$qt4$1@digitalmars.com...
>>> "init" and other class properties are kind of shadow-keywords.
>>>
>>> class MyClass{
>>>   void init(){
>>>   }
>>> }
>>>
>>> The class is fine and can be used.
>>> Now someone attempts to insert this class into a container....
>>>
>>> template Templ(T){
>>>   T t = T.init; // Compile error
>>> }
>>>
>>> The compiler should forbid the use of the method name "init", if the
>>> signature does not make any sense for an "init" property. It must be a
>>> static method, with a return value and no arguments.
>>
>> This should extend to any semi-reserved name.  The compiler used to allow
>> redefinition of 'sizeof' and 'offsetof' but checks were added to disallow
>> that.  That mechanism should just be extended.
>>
>>
> 
> I filed a bug of forbiding the use of stringof method. But it's markd as invalid :(
> http://d.puremagic.com/issues/show_bug.cgi?id=1412
> 

I've added my reasoning to that bug report. I guess one could make a case for redefining 'init', but as for 'stringof', it should be the same as 'sizeof' and 'offsetof': unextendable, like Jarret mentioned.

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