Is is ok that this compiles without errors?

class Foo {
	alias errorProne this;
}

On Fri, Jun 26, 2009 at 5:54 PM, Ary Borenszweig<ary@esperanto.org.ar> wrote:
> Is is ok that this compiles without errors?
>
> class Foo {
>        alias errorProne this;
> }
>

If I had to guess, it's only OK because the compiler probably doesn't actually look for the aliased symbol until you actually try to access something from it.  But that does seem .. wrong.

Jarrett Billingsley wrote:
> On Fri, Jun 26, 2009 at 5:54 PM, Ary Borenszweig<ary@esperanto.org.ar> wrote:
>> Is is ok that this compiles without errors?
>>
>> class Foo {
>>        alias errorProne this;
>> }

Errors as expected on 1.042; compiles on 2.029.

> If I had to guess, it's only OK because the compiler probably doesn't
> actually look for the aliased symbol until you actually try to access
> something from it.  But that does seem .. wrong.

It is wrong.  Since this isn't in a template, the compiler should perform semantic analysis on it in any case, and thus report an error.

But the really strange thing is that it doesn't complain about the attempt to redefine 'this', which is a keyword.

This ought to be in Bugzilla if it isn't already and you can't find anything in the spec that gives this form a meaning of its own.


Strangely, if I change 'this' to 'that', I get the expected

errorprone.d(2): Error: identifier 'errorProne' is not defined


Stewart.

Stewart Gordon escribió:
> Jarrett Billingsley wrote:
>> On Fri, Jun 26, 2009 at 5:54 PM, Ary Borenszweig<ary@esperanto.org.ar> wrote:
>>> Is is ok that this compiles without errors?
>>>
>>> class Foo {
>>>        alias errorProne this;
>>> }
> 
> Errors as expected on 1.042; compiles on 2.029.
> 
>> If I had to guess, it's only OK because the compiler probably doesn't
>> actually look for the aliased symbol until you actually try to access
>> something from it.  But that does seem .. wrong.
> 
> It is wrong.  Since this isn't in a template, the compiler should perform semantic analysis on it in any case, and thus report an error.
> 
> But the really strange thing is that it doesn't complain about the attempt to redefine 'this', which is a keyword.

It's called "alias this", that's not the problem:

http://digitalmars.com/d/2.0/class.html#AliasThis

I also think it should complain, because it makes no sense to just leave it there, unresolved.