I would like to suggest the following variable type added to D as a supplement to pointers.  Please review the following code and suggest modifications and additions as you see.

Thanks!

-S.

Oh yes, and please assume all the following asserts PASS.


long x = 2;
long& y = &x;

assert(y == 2);

y = 4;

assert(x == 4);

assert(&y == &x);  //Notice the change in the operation of &.  Not a handle to x!

long z = 5;

&y = &z;  //& Allows l-value assignment,

assert( y == 5 );

assert( &y == &z );

assert(y.sizeof == z.sizeof);  //Sizeof gives information on referenced type, not reference.  Ditto for other properties

S. Chancellor wrote:
> I would like to suggest the following variable type added to D as a supplement to pointers.  Please review the following code and suggest modifications and additions as you see.
> 
> Thanks!
> 
> -S.
> 
> Oh yes, and please assume all the following asserts PASS.
> 
> 
> long x = 2;
> long& y = &x;
> 
> assert(y == 2);
> 
> y = 4;
> 
> assert(x == 4);
> 
> assert(&y == &x);  //Notice the change in the operation of &.  Not a handle to x!
> 
> long z = 5;
> 
> &y = &z;  //& Allows l-value assignment,
> 
> assert( y == 5 );
> 
> assert( &y == &z );
> 
> assert(y.sizeof == z.sizeof);  //Sizeof gives information on referenced type, not reference.  Ditto for other properties
> 

While I like the idea of references and refernce semantics for things like structs and whatnot, I think the C++ & operator is the *wrong* way to do it.

Hasan Aljudy wrote:
> S. Chancellor wrote:
>> I would like to suggest the following variable type added to D as a supplement to pointers.  Please review the following code and suggest modifications and additions as you see.
>>
>> Thanks!
>>
>> -S.
>>
>> Oh yes, and please assume all the following asserts PASS.
>>
>>
>> long x = 2;
>> long& y = &x;
>>
>> assert(y == 2);
>>
>> y = 4;
>>
>> assert(x == 4);
>>
>> assert(&y == &x);  //Notice the change in the operation of &.  Not a handle to x!
>>
>> long z = 5;
>>
>> &y = &z;  //& Allows l-value assignment,
>>
>> assert( y == 5 );
>>
>> assert( &y == &z );
>>
>> assert(y.sizeof == z.sizeof);  //Sizeof gives information on referenced type, not reference.  Ditto for other properties
>>
> 
> While I like the idea of references and refernce semantics for things like structs and whatnot, I think the C++ & operator is the *wrong* way to do it.

What do you mean "reference semantics"? "References" (as in "reference variables, "alias references", or "c++ references") do not give "reference semantics", (as in "reference types").

-- 
Bruno Medeiros - CS/E student
http://www.prowiki.org/wiki4d/wiki.cgi?BrunoMedeiros#D

S. Chancellor wrote:
> I would like to suggest the following variable type added to D as a supplement to pointers.  Please review the following code and suggest modifications and additions as you see.
> 
> Thanks!
> 
> -S.
> 
> Oh yes, and please assume all the following asserts PASS.
> 
> 
> long x = 2;
> long& y = &x;
> 
> assert(y == 2);
> 
> y = 4;
> 
> assert(x == 4);
> 
> assert(&y == &x);  //Notice the change in the operation of &.  Not a handle to x!
> 
> long z = 5;
> 
> &y = &z;  //& Allows l-value assignment,
> 
> assert( y == 5 );
> 
> assert( &y == &z );
> 
> assert(y.sizeof == z.sizeof);  //Sizeof gives information on referenced type, not reference.  Ditto for other properties
> 

I would moderately prefer a syntax with "ref" instead of "&", as:
  long x = 1;
  ref long z = &x;
Also, "ref" (or "&") would supersede the functionality of "inout", since they do the same, only that "inout" only works for parameter variables.

-- 
Bruno Medeiros - CS/E student
http://www.prowiki.org/wiki4d/wiki.cgi?BrunoMedeiros#D

Bruno Medeiros wrote:
> Hasan Aljudy wrote:
> 
>> S. Chancellor wrote:
>>
>>> I would like to suggest the following variable type added to D as a supplement to pointers.  Please review the following code and suggest modifications and additions as you see.
>>>
>>> Thanks!
>>>
>>> -S.
>>>
>>> Oh yes, and please assume all the following asserts PASS.
>>>
>>>
>>> long x = 2;
>>> long& y = &x;
>>>
>>> assert(y == 2);
>>>
>>> y = 4;
>>>
>>> assert(x == 4);
>>>
>>> assert(&y == &x);  //Notice the change in the operation of &.  Not a handle to x!
>>>
>>> long z = 5;
>>>
>>> &y = &z;  //& Allows l-value assignment,
>>>
>>> assert( y == 5 );
>>>
>>> assert( &y == &z );
>>>
>>> assert(y.sizeof == z.sizeof);  //Sizeof gives information on referenced type, not reference.  Ditto for other properties
>>>
>>
>> While I like the idea of references and refernce semantics for things like structs and whatnot, I think the C++ & operator is the *wrong* way to do it.
> 
> 
> What do you mean "reference semantics"? "References" (as in "reference variables, "alias references", or "c++ references") do not give "reference semantics", (as in "reference types").
> 

Maybe I don't understand the meaning of "semantics"!
What I mean by "Reference semantics" is reference behavior applied to pointers. (if that makes it any clearer ..)

ref/alias/inout are possible choices. It seems impossible to write proper containers without C++ like reference type

>
>I would moderately prefer a syntax with "ref" instead of "&", as:
>   long x = 1;
>   ref long z = &x;
>Also, "ref" (or "&") would supersede the functionality of "inout", since they do the same, only that "inout" only works for parameter variables.
>
>-- 
>Bruno Medeiros - CS/E student http://www.prowiki.org/wiki4d/wiki.cgi?BrunoMedeiros#D