Hi Dave
The problem you stumbled upon is (or was :-() well known. It was
discussed during review and I even tried to find a workaround for it not
too long ago. The solution is very simple and is even described in the
reference manual under the requirements for the InnerInitial parameter:
<quote>
An mpl::list<> containing models of the SimpleState or State concepts or
instantiations of the shallow_history or deep_history class templates.
If there is only a single inner initial state
_that_is_not_a_template_instantiation_ then it can also be passed
directly, without wrapping it into an mpl::list<>. [...]
</quote>
(_Note_the_underlined_part_). So, the only thing you need to do is to
wrap every templated InnerInitial parameter into an mpl::list and
everything will work as expected (see attached example). The reason why
compilers rightly choke on the code lies in the fact that simple_state
internally checks whether the InnerInitial parameter already is an
mpl::list. This check works perfectly even on forward-declared classes.
However, as soon as InnerInitial is a class template instantiation, for
some reason that template needs to be fully instantiated. The problem is
described in more detail here:
http://article.gmane.org/gmane.comp.lib.boost.devel/128741
I guess it was simply a little too late yesterday, sorry...
#include