Re: [boost] [contract] syntax redesign

Subject: Re: [boost] [contract] syntax redesign
From: lcaminiti (lorcaminiti_at_[hidden])
Date: 2011-05-31 17:21:08

Matt Calabrese wrote:
> 
> On Tue, May 31, 2011 at 2:49 PM, Lorenzo Caminiti
> <lorcaminiti_at_[hidden]>wrote:
>>
>> For example, I _think_ (but I've not implemented this yet) that I can
>> simplify the syntax like this:
>>
> 
> Looks good, I use pretty much exactly the same syntax in my library
> concerning parameter lists and templates (
> https://github.com/boostcon/2011_presentations/raw/master/thu/Boost.Generic.pdf
> slides
> 79 and 96 are okay examples). The only thing I'm confused about is how you
> 

Thanks, I'll take a look.

> specify the return type of your CONTRACT_FUNCTION -- in particular, you
> did
> not wrap it in parentheses. Are you really able to do that, or is that a
> 

If the return type is a "keyword" known a priori like void then no
parenthesis should be needed. I'd parse it using the pp PP_IS_UNARY to first
check if there are parenthesis. If there are no parenthesis then I'd check
if the token is a known "keyword" type `void`, `int`, `double`, etc using
macros like BOOST_DETAIL_PP_KEYWORD_IS_VOID_FRONT and strip it aways from
the front using macros like BOOST_DETAIL_PP_KEYWORD_VOID_REMOVE_FRONT (see
http://svn.boost.org/svn/boost/sandbox/local/boost/detail/preprocessor/keyword/).

So, this does not need the parenthesis because int is a "keyword" known a
priory (void return type also belongs to this category):

    CONTRACT_FUNCTION( public int (size)() ... )

This instead needs the parenthesis because my_own_size_type is a
user-defined type and it's not known a priory:

    CONTRACT_FUNCTION( public (my_own_size_type) (size)() ... )

Also this needs the parenthesis because int* contains a non-alphanumeric
symbol * that cannot be handled by the BOOST_DETAIL_PP_KEYWORD_XXX macros
(because they use pp concatenation):

    CONTRACT_FUNCTION( public (int*) (size)() ... ) 

The same applies to all specified types including function parameter types
`void (f)( int x, int y )` vs. `void (f)( (int*) x, int y )`.

> mistake? If that syntax is indeed possible, then I'd like to see the
> implementation because I too would love to be able to rid myself of some
> parentheses.
> 

I have done some experimentation that indicates this syntax _should_ be
possible but it's not actually implemented yet so... we'll see :)

--Lorenzo

--
View this message in context: http://boost.2283326.n4.nabble.com/contract-syntax-redesign-tp3563993p3564380.html
Sent from the Boost - Dev mailing list archive at Nabble.com.