[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
Re: Container <-> protocol dependancy
Well, I would definitely go for no-RTTI solution. After
all, this is C++. Programmer should Know What They Are Doing (TM)
Now, do boost::any and friends work without RTTI?
Do we still use any of that in API?
Krzysztof Rzymkowski wrote:
>>Now, last issue with your
>>HttpServletRequest::service(ServletRequest&, ServletResponse&).
>>It uses dynamic_cast, which in turn means we'll have to use RTTI for
>>whole servlet container, and all servlets. Not something I'm too happy
>>about. If we could get around it somehow (while preserving this specific
>>type check preferably), it would be nice.
>>
>>
>>
>
>We can always repleace the dynamic_cast with static_cast
>If you want to have the type check and not use RTTI, you'll have to write
>the code that RTTI would have generated by hand. This wouldn't rather give
>better results. Of course when one is implementing ones own RTTI one can
>choose which classes are under the custom RTTI.
>All in all I'm for fully dropping RTTI (and not implementing similar
>mechanism) or fully using it (like it is now) with dynamic_casts, typeid
>etc.
>
>
>
--
Ilya A. Volynets-Evenbakh
Total Knowledge. CTO
http://www.total-knowledge.com