[aspectc-user] calls to members through a pointer to member

Olaf Spinczyk os at aspectc.org
Thu Dec 20 17:35:56 CET 2007

Hi Panu,

it is not the case that we don't want to support this feature (advice 
for calls via function pointers). The only question is in which way we 
are going to support. Our underlying philosophy is that users who don't 
need a specific feature should not have to pay for it.

Best regards,


Panu Bloigu wrote:
> Hi,
> Daniel Lohmann wrote:
>>> If this indeed is possible, and I'm doing something wrong, it would 
>>> be interesting to hear how this feature is implemented. As far as I 
>>> can see, weaver has to somehow gain awareness of the actual value 
>>> (the pointed-to function/method) of a 
>>> pointer-to-function/pointer-to-member to be able to tell the target 
>>> of the call.
>> Right, and this is the main reason ac++ we hesitate to support this. 
>> It could lead to significant extra runtime costs, which is somewhat 
>> against the AspectC++ philosophy.
> Are you saying that you are not even planning to implement this feature?
> I'm asking this because this isn't listed in the paper 'The design and 
> implementation of AspectC++' published in Knowledge Based Systems, 
> volume 20, issue 7, in the section titled 'Intentionally missing 
> features'. This led me to think that the ability to trap calls 
> occurring through a pointer would be either already implemented or at 
> least on the drawing board.
> Of course, I acknowledge the amount of necessary instrumentation that 
> implementing this feature would require. However, I would think that 
> at least, for example, pointcut function call("void %(int)") could 
> catch calls to any function returning void and taking an int as the 
> argument -- even if that call occurs through a pointer. I have a hunch 
> that implementing this wouldn't cause too much run time expense, as 
> the compile time type of a pointer-to-function contains the signature 
> -- minus the function name, of course -- of the function that's to be 
> called through that pointer.
> -- Panu
> _______________________________________________
> aspectc-user mailing list
> aspectc-user at aspectc.org
> http://www.aspectc.org/mailman/listinfo/aspectc-user

More information about the aspectc-user mailing list