christophe henry wrote:

"Eric Niebler" schrieb im Newsbeitrag

...

Steven Watanabe wrote:

...

christophe henry wrote:

...

I am trying to build a mpl::vector from an expression like

(string(),ClassA(),ClassB())

To get a mpl::vector

Is there a way to do this or I should give up the idea?

Does fold_tree help?

It does.

*snip*

Ah It see. Thanks a lot, I think I got the idea. I just have a concern with:

template<class T> struct is_my : mpl::false_ {}; template<> struct is_mystd::string : mpl::true_ {}; template<> struct is_my<ClassA> : mpl::true_ {}; template<> struct is_my<ClassB> : mpl::true_ {};

As it forces extra declarations which not everybody is willing to make. But based on the same fold idea I can allow something like: attributes_ << int() << std::map() << bool()

Where attributes_ is made a proto terminal with proto::extends

Sure, that's a reasonable approach.

and the following grammar (which I shamelessly stole from the doc):

struct BuildAttributes : proto::or_< // Don't add the flag terminal to the list proto::when< proto::terminal< my_tag > , proto::_state

...

// Put all other terminals at the head of the // list that we're building in the "state" parameter , proto::when< proto::terminalproto::_ , fusion::cons(proto::_value, proto::_state)

...

// For left-shift operations, first fold the right // child to a list using the current state. Use // the result as the state parameter when folding // the left child to a list. , proto::when< proto::shift_left , proto::fold< proto::_ , fusion::nil() , BuildAttributes>()

...

{};

I then just have to invert the order again as it gets reversed (ok a few seconds more to compile...):

boost::mpl::copy::type,boost::mpl::front_inserter

...

...

::type

Correct?

You could use reverse_fold instead of reversing the whole list after the transform. Or you could use reverse_fold_tree to get the vector and then pop the attribute_ off the front. That's probably fastest and simplest. -- Eric Niebler BoostPro Computing http://www.boostpro.com