[cfe-dev] type-id/functional cast ambiguity resolution

[Arnaud Bergeron]

This is my take on this:

2008/10/28 Argiris Kirtzidis <akyrtzi at gmail.com>:
> In my post, change "function-style cast" to "type construction".
>
> "T()" is a specific case of "T( expression-list[opt] )" described in
> "5.2.3 Explicit type conversion (functional notation)", and I sometimes
> tend to think the general case as "function-style cast".
>
> -Argiris
>
> Argiris Kirtzidis wrote:
>> Hi all,
>>
>> Here's a test case:
>>
>> void f() {
>>  int x = (int())+1;
>> }

This can be interpreted as a cast where 'int()' is a type-id so this
interpretation is favoured.

>>
>> GCC, MSVC and Comeau, all treat "(int())" as a cast to function type
>> and report a "invalid cast" compile error.
>>
>> The standard has this to say:
>>
>>>   // C++ 8.2p2:
>>>   // The ambiguity arising from the similarity between a
>>> function-style cast and
>>>   // a type-id can occur in different contexts. The ambiguity appears
>>> as a
>>>   // choice between a function-style cast expression and a
>>> declaration of a
>>>   // type. The resolution is that any construct that could possibly
>>> be a type-id
>>>   // in its syntactic context shall be considered a type-id.
>>
>> In the above test case, the construct "int()" cannot be a type-id in
>> its syntactic context, shouldn't it be considered as a function-style
>> cast resulting in no compilation errors ?
>>
>> You might think that all three compilers doesn't consider the
>> type-id's syntactic context beyond the parentheses, but take a look at
>> this:
>>
>> void f() {
>>  (int());
>> }

In this context 'int()' cannot be a type.  Any type at this place
would result in a compilation error so the other interpretation is
taken.

>> All 3 compilers now regard "int()" as a function-style cast and report
>> no errors.
>> That doesn't seem consistent, if the second case is regarded as
>> correct, then the first one should be as well.
>>
>> Any thoughts ?
>>
>> -Argiris
>>
>>
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
>