Re: [boost] New warning (non dll-interface ...) with vc7

From: John Maddock (jm_at_[hidden])
Date: 2002-09-24 06:26:22

> I'm probably about to expose embarrassing ignorance about how C++ really
> works, but isn't it possible that our software could be vulnerable to
> future behavior of the regex++ package which would break our code if
> parts of a base class used by the package aren't accessible at runtime?
> What I mean is this: even if we determine by careful inspection that
> nothing in our code explicitly tries to use those inaccessible parts of
> the base class, it's possible that macros in the regex++ headers might
> make reference to those parts, causing some sort of runtime failure.
> How far off base is this line of thought?

I'm not sure I follow: the non-exported base class has it's full definition
visible in the header (all it's functions are declared inline so there is
nothing to be exported/imported).  BTW the STL classes all have the same
issue, if you catch a std::runtime_error thrown from std::string, you're
catching a non-exported class that was thrown from a dll...

John Maddock
http://ourworld.compuserve.com/homepages/john_maddock/index.htm