Re: dynamic_cast problem

[Earl Purple]

Bernd Strieder wrote:
> Hello,
>
> before we get deeper into guesswork. Does the following help?
>
> http://gcc.gnu.org/faq.html#dso
>
> One should always try to find better answers by others first.
> Google with the request "site:gcc.gnu.org RTTI dynamic" did it.
>
> Bernd Strieder

It helped (thank you) and I have fixed the problem by opening (i.e.
dlopen) with mode RTLD_LAZY | RTLD_GLOBAL instead of just RTLD_LAZY.

Did not need to modify any of the compile-time switches. (My executable
does not export global symbols). Performance loss is totally not an
issue here - this is all done at startup-initialisation. There is no
RTTI done at all after that unless an exception is subsequently thrown.

I did google by the way and didn't find the answer to my problem. Must
have googled the wrong thing (came up with loads of hits when I tried
the keywords but all the posts on the subject were over 2 years old and
none of them specifically addressed this particular problem nor came up
with the solution).

Hopefully this particular thread will be useful for someone else who
later has a similar problem.