Zitat von "Stewart, Robert" <Robert.Stewart@sig.com>:
_1_1 is more readable and succinct than arg<1,1>:
auto const_ = lam<1>( lam<1>( arg<2,1>() ) );
auto const_ = lam<1>( lam<1>( _2_1() ) );
and:
auto flip = lam<1>( lam<2>( app( arg<2,1>() , arg<1,2>() , arg<1,1>() ) ) );
auto flip = lam<1>(lam<2>(app(_2_1(), _1_2(), _1_1())));
or: auto flip=lam(lam(app(_2_1,_1,_2))); david, why do you need to number the lambdas? doesn't the "nested index" of the args refer to the level of nesting instead of the number given to the lam<> template? _1, _2 in the example above would refer to the placeholders e.g. of Boost.Bind and be equivalent to _1_1 and _1_2. thomas (author of phoenix v3) has mentioned an effort to unify all the placeholders in boost (Bind, Phoenix, ...)