From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2002-07-25 04:58:54
David Brownstein wrote:
> Why not just call it MetaFunctionClass (or
> make_MetaFunctionClass)? It's at least east to read and understand... ;-)
'Cause it's overly long? :) Consider, for instance, this code snippet from
the paper:
typedef bind<
meta_fun2<logical_or>
, bind< meta_fun2<less>
, bind< meta_fun1<size_of>,_1 >
, int_c<16>
>
, bind< meta_fun2<boost::is_same>,_1,_2 >
> func;
Now, we replace 'meta_fun' with 'metafunction_class':
typedef bind<
metafunction_class2<logical_or>
, bind< metafunction_class2<less>
, bind< metafunction_class1<size_of>,_1 >
, int_c<16>
>
, bind< metafunction_class2<boost::is_same>,_1,_2 >
> func;
Ouch, IMO :). Another question is whether anybody would compose
metafunctions in this way in the first place, considering that lambda
notation is available on all compilers now.
Aleksey