The optimizing compiler is smart. In a release build, under /EHs, if it
doesn't see a throw, it may just optimize away the catch clause.

The word exception is overloaded. C++ exceptions are different from
structured exceptions used by the o/s to signal things like access
violations and zero divided. My friend Doug explains the difference here

http://members.cox.net/doug_web/eh.htm

Note that often benign conditions are signalled by C++ typed exceptions and
it is often OK to catch one and go on one's way. SE's are almost always a
sign of a programming error. Often the best strategy is to log the error and
quit.

Still, if you find a need to catch an SE and continue, say in the
development phase of a project, I have an example that shows how to do that
at the bottom of this page on my website:

http://www.ivrforbeginners.com/downloads/downloads.htm

Regards,
Will

You provide your own type.  See _set_se_translator in the documentation.