making functors adaptable

[Fraser Ross]

Functors taking 1 argument for operator() should inherit from
unary_function and those with 2 arguments should inherit from
binary_function. If a functor has zero arguments for its operator()
should it inherit from anything to make it adaptable? I've not heard of
anything before which is surprising. If books don't talk about it then
its an unanswered question to many people. Maybe its adaptable without
inheriting anything, which would explain the absence. It may be an
oversight of the STL.

Fraser.


*** Free account sponsored by SecureIX.com ***
*** Encrypt your Internet usage with a free VPN account from http://www.SecureIX.com ***

 

[Rolf Magnus]

Functors taking 1 argument for operator() should inherit from
unary_function and those with 2 arguments should inherit from
binary_function. If a functor has zero arguments for its operator()
should it inherit from anything to make it adaptable? I've not heard of
anything before which is surprising. If books don't talk about it then
its an unanswered question to many people. Maybe its adaptable without
inheriting anything, which would explain the absence. It may be an
oversight of the STL.

Is there any standard algorithm that uses function objects with no
arguments?

 

[Fraser Ross]

"Rolf Magnus"

Is there any standard algorithm that uses function objects with no
arguments?

Theres generate_n.

Fraser.


*** Free account sponsored by SecureIX.com ***
*** Encrypt your Internet usage with a free VPN account from http://www.SecureIX.com ***

 

[Victor Bazarov]

Is there any standard algorithm that uses function objects with no
arguments?

std::generate is the only one comes to mind.

V

 

[Maxim Yegorushkin]

Functors taking 1 argument for operator() should inherit from
unary_function and those with 2 arguments should inherit from
binary_function.

The only thing that the base classes do is declare typedefs for return
type and argument types.

If a functor has zero arguments for its operator()
should it inherit from anything to make it adaptable?

Just declare result_type typedef in your nullary functor.

I've not heard of anything before which is surprising. If books don't talk about it then
its an unanswered question to many people. Maybe its adaptable without
inheriting anything, which would explain the absence. It may be an
oversight of the STL.

Note that no stl algorithm or functor adaptor work with nullary
functors. boost::bind and boost::function do.