christophe henry wrote:

Hi all,

Here am I already at the 2nd compiler (VC9SP1) crash of the week (not kidding...):

fatal error C1001: An internal error has occurred in the compiler. (compiler file 'f:\dd\vctools\compiler\utc\src\p2\p2symtab.c', line 5809)

Attached is a test file producing the problem. The crash happens when I add BOOST_PROTO_DEFINE_OPERATORS. Am I doing something wrong?

I can't reproduce the error on msvc-9.0 sp1 either with or without the BOOST_PROTO_DEFINE_OPERATORS line.

Sadly, this was a workaround to avoid deriving my class from proto::terminal because of a problem (supposedly) created by the address-of operator. Can someone save my day and tell me if there is a way to deactivate this operator for terminals? Concretely, I have:

struct Empty: public state<>, proto::terminal

Which brings me, here:

&::boost::fusion::at_key<State>(xxx) // xxx being a fusion set containing Empty

the following problem:

1>f:\projects\MSM2\SM\boost/msm/back/state_machine.hpp(1061) : error C2784: '::boost::enable_if::type boost::msm::back::state_machine::execute_entry(StateType *,const EventType &,FsmType &,boost::msm::back::dummy<__formal>)' : could not deduce template argument for 'StateType *' from 'const boost::proto::exprns_::expr'

Looks like the error is happening in msm. That doesn't mean it's not caused by Proto, but it means you're not posting the actual code that's failing. I can't help without the actual code. Anyway, Joel's advice is spot-on. You can disable Proto's operator& by using proto::extends instead of straight inheritance, and specify the grammar of your DSEL in the domain parameter. Your grammar could be as simple as: proto::not_proto::_ >. HTH, -- Eric Niebler BoostPro Computing http://www.boostpro.com