Re: [boost] [PREPROCESSOR] Possible improvement for BOOST_PP_VARIADIC_SIZE

Subject: Re: [boost] [PREPROCESSOR] Possible improvement for BOOST_PP_VARIADIC_SIZE
From: Edward Diener (eldiener_at_[hidden])
Date: 2013-09-09 22:47:59

On 9/9/2013 4:44 PM, Brent Lewis wrote:
> Currently,
>
> BOOST_VARIADIC_SIZE()
>
> evaluates to 1. Based on
> http://gustedt.wordpress.com/2010/06/08/detect-empty-macro-arguments/ it
> can be improved to return 0 for an empty argument list.

The flaw in the detection of empty arguments in the above link has 
already been discussed. I am not putting down that implementation in 
saying so.

It is impossible to perfectly detect an empty argument. IMO the best 
case for detecting an empty argument was an example by Paul Mensonides 
on the web which I adapted for my VMD library at 
https://github.com/eldiener/variadic_macro_data.git.

Furthermore as Paul has pointed out even an "empty" argument is 
considered an argument and still counts as 1 for the variadic size. In 
fact the spec for a variadic macro says that you cannot pass no argument 
when the macro expects variadic data. In other words:

#define AVAR(ARG,...)

AVAR(something)

is an error, but of course

AVAR(something,)

or

#define AVARAG(...)

AVARAG()

is perfectly fine since you are indeed passing an argument even if it is 
an "empty" argument.