[cfe-commits] r125699 - in /cfe/trunk: include/clang/Basic/DiagnosticSemaKinds.td lib/Sema/SemaExprObjC.cpp test/SemaObjC/property-missing.m

[Douglas Gregor]

On Feb 16, 2011, at 6:17 PM, Ted Kremenek wrote:

> On Feb 16, 2011, at 6:12 PM, Douglas Gregor wrote:
> 
>> 
>> On Feb 16, 2011, at 5:59 PM, Ted Kremenek wrote:
>> 
>>> Hi Fariborz,
>>> 
>>> I'm not certain this diagnostic is any better:
>>> 
>>> +	myObject.someOtherObject.someProperty = 0; // expected-error {{property 'someOtherObject' names an object of forward class type in class object 'MyClass *'}}
>>> 
>>> What is a "forward class type"?  I know what you mean; you're talking about a Objective-C class that has only been forward declared, but there is no such thing as a forward class type.
>>> 
>>> How about:
>>> 
>>> property 'someOtherObject' refers to a Objective-C class ('SomeOtherClass') with only a forward declaration (no @interface available)
>>> 
>>> This diagnostic clearly spells out that (a) no @interface is available for (b) the class SomeOtherClass when (c) accessing the property someOtherObject.
>> 
>> In the C/C++ world, we would say that the type MyClass is 'incomplete', then point to the forward declaration.
>> 
>> 	- Doug
> 
> 
> That's true, but a strictly "clinical" diagnostic isn't necessarily helpful to the user.  How about:
> 
>  property 'someOtherObject' refers to an incomplete Objective-C class 'SomeOtherClass' (with no @interface available)
> 
> and then have the note refer to the forward declaration (as it does now)?


WFM.

	- Doug