I think C++ Perl binding is an exciting issue, like C++, STL and Perl are. C++ is evolving fast, as does Perl. So, in my opinion, it would be great if we have a clear, well documented way to deploy a few C++ library's functions into Perl modules. My XS knowledge is poor, but I know some C++, and am willing to help somehow.
The clearest point we have is: h2xs is _not_ suitable for autogenerating C++ Perl bindings. Assuming that, and after reading the direct and excelent John Keiser's http://www.johnkeiser.com/perl-xs-c++.html tutorial, it seems like: avoid h2xs and C++ parsing and every thing's gonna be all right. Things are walking here. Now I worry about this: Nick Ing-Simmons wrote: > It it really fails at "make test" stage then you probably need > LD=g++ or similar to link in the iostream etc. > Note that not all C++ implementations are dynamic loading friendly. > e.g. if you load a C++ extension and root executable isn't C++ > (and perl isn't) then static constuctors for 'cout' may not have > been called. In CookBookB/CCsimple (CPAN/authors/id/DMR/), mentioned in Keiser's tutorial, there's a note about a way to properly "add" C++ loading capability into the C perl binary. (I confess I didn't fully understand that, though, too much confusion about static/dynamic C/C++ linkage.) Silvio
