Re: [sv-ec] RE: id 185, svdb 2342 static constructor

Now I am really concerned.

So are you implying that the use of "this" in a constructor
is illegal?  That would be a very bad place to end up
as far as I am concerned.

I absolutely believe that "this" is valid during
construction.  Given that, it is indistinguishable
from *inside* the constructor whether the constructor
is a static method or not.  I absolutely agree that
from the *caller* side it looks like a static method.
But it really isn't and I believe that it would be incredibly
misleading to describe it as such.

Gord.



jonathan.bromley@doulos.com wrote:
> [constructors]
>> are not invoked using a handle because their invocation allocates
>> the object that they use.  But regardless of how they are invoked, they
>> are inherently not static.  Their purpose is to construct the object,
>> which they cannot do without access to the object.
> 
> Agreed, but...
> 
>>  They have to be passed
>> a 'this' handle (which is the handle for the allocated object) so that
>> they can initialize the nonstatic members of the object.
> 
> I think there's another way of thinking about it that makes 
> rather more sense to me.
> 
> When I call new() explicitly, I most certainly don't think of
> it being passed "this" because "this" doesn't exist at the
> time of the call.  Instead, imagine that the constructor
> truly *is* static, but has "this" as an implicit local
> variable that is implicitly used as a prefix to any mention
> of class members within the body of new().
> 
> I'm very aware that this description is torturing the 
> meaning of "this", but I still maintain that any call
> to new() looks like a static method call; there is no
> meaningful "this" at the time of the call, and so the
> implicit "this" argument that is passed to non-static
> methods can't happen.  And, indeed, this is perhaps
> the root of some of the problems with super.new();
> at the time of the call to super.new(), "super" has
> already been allocated - making that call to new()
> very different from a user's call to the base class's
> constructor.
> 
> Whatever, we have to accept that new() is strange and
> special, so we can make strange and special rules for it.

-- 
--------------------------------------------------------------------
Gordon Vreugdenhil                                503-685-0808
Model Technology (Mentor Graphics)                gordonv@model.com


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.