Boost logo

Boost Users :

Subject: Re: [Boost-users] proto: BOOST_PROTO_AUTO
From: Eric Niebler (eric_at_[hidden])
Date: 2009-01-05 20:30:46


(Please don't top-post in the future. Thanks.)

Kim Kuen Tang wrote:
> Hi Eric,
>
> thank you for your quick response.
>
> It is interesting to see that expression like "Domain( [0,1] )" is not
> allowed. I think the reason is that such expression cannot be
> transformed in a tree or the transformation is not unique.

No, it is because "Domain( [0,1] )" is disallowed by the grammar for
legal C++ expressions. operator[] is binary. You could do this: "Domain(
_[0,1] )" but ....

> But why does the code "proto::display_expr(Domain[1,2]);" produce the
> output:
>
> subscript(
> terminal(domain_tag)
> ,terminal(2)
> )
>
> and not
>
> subscript(
> terminal(domain_tag)
> ,terminal(1)
> ,terminal(2)
> ) ?

... the expression "0,1" already has a meaning in C++. It means,
evaluate "0", discard it, evaluate "1". You can only overload operators
on user-defined types.

HTH,

> Eric Niebler schrieb:
>> Kim Kuen Tang wrote:
>>> Hi all,
>>>
>>> i have a question and a problem.
>>>
>>> The problem is that i would like to write code like:
>>> domain=Domain([0,1]*[0,1]);
>>> where Domain([0,1]*[0,1]) is an expression created with proto.
>>>
>>> But of course this line of code is illegal, because i have to provide
>>> the type of this expression Domain([0,1]*[0,1]).
>>
>> That line of code is illegal because there is no way to make
>> "Domain([0,1]*[0,1])" a valid C++ expression. You need to work on your
>> syntax a little.
>>
>>> So i have to write something like this:
>>> Domain([0,1]*[0,1])::type domain=Domain([0,1]*[0,1]);
>>> Is this correct?
>>>
>>> Now the question:
>>> To avoid code like above is this what the macro BOOST_PROTO_AUTO is for?
>>> Can i write
>>> BOOST_PROTO_AUTO( domain,Domain([0,1]*[0,1]) ); .
>>
>> Yes, that is what BOOST_PROTO_AUTO is for.
>>

-- 
Eric Niebler
BoostPro Computing
www.boostpro.com

Boost-users list run by williamkempf at hotmail.com, kalb at libertysoft.com, bjorn.karlsson at readsoft.com, gregod at cs.rpi.edu, wekempf at cox.net