6348631 - request for review (updated)

[Coleen Phillimore]

>> I agree. This sounds like the right approach, and one that is consistent
>> with past practice. This would also be exactly what the
>> make-deps removal accomplished for the existing .inline.hpp 
>> dependencies as
>> well.
>
> So just to be clear here, Volker and Ramki agree with me that the 
> system includes should remain in the header files that need them - 
> right? Whereas Ivan and Coleen want to see them moved into the cpp files.

No, I agree with that too.  I have no argument against the .hpp  or 
.inline.hpp  files containing the include directives that it needs.  In 
this hpi case, we were inlining functions that needed system includes.  
My request was that the functions not be inlined so that the system 
includes can be included in the .cpp file instead, to minimize exposure 
of the entire source base to these system includes.   It was just to 
keep the gnarly implementation details as local as possible.

thanks,
Coleen
>
> David
>