Re: How to force attribute propagation in x3 when using semantic actions

Seth writes:

> On 15-12-17 16:32, Alberto Luaces wrote:
>> I tried before posting, but it makes no difference for me:
>>
>> http://coliru.stacked-crooked.com/a/b58b992c2a5df373
>
>         // Let us pretend that this parses a list and not only two entities.
>
> Pretending doesn't make compilers happy. You need something that is
> actually compatible with your attribute type.
>

I though it was: at least on the Qi documentation[1] it is said that:

A >> A → vector<A>

I will check if for x3 things are different.

>
> I'd simply parse the list...:
>
>         auto const complist_def = mycomp % ',';
>
> Then you definitely did *not* want aumatic propagation here:
>
>             x3::rule<class _, comptype/*, true*/>{"my comp parser"} =
>
> For the simple reason that the attributes are not compatible.
>

Great!  Now I understand that part: a tuple or a vector would be
returned instead of a "comptype" —this is why automatic propagation is
disabled with semantic actions.

Now I would question why cannot I rely on attribute automatic detection
for the list, and have to write instead

x3::rule<class _, std::vector<comptype> > const complist("complist");

It turns out that if I remove that line and its matching
BOOST_SPIRIT_DEFINE call, and then insert this parser into bigger
compositions:

    auto const complist = mycomp % ',';  //List of complex
    auto calculator = complist % ';';    //List of lists of complex

...everything works nicely, but maybe at the expense of a bigger load
for the compiler (Coliru refuses to build this version).  So maybe the
use of BOOST_SPIRIT_DEFINE has a significant build performance advantage
over "auto" parser composition.

>
> Fixed working version: http://coliru.stacked-crooked.com/a/9926cdf49ee8215b
>
> A much simpler, more idiomatic version using
> `BOOST_FUSION_ADAPT_STRUCT`:
> http://coliru.stacked-crooked.com/a/57d6b3fcd7834644

Sorry if I did not make myself clear; this is what I did from the
beginning, and indeed it worked, but I moved on to simulate what would
happen if for some reason I would have to resort to semantic actions in
a small part of the parser.


Footnotes: 
[1]  http://www.boost.org/doc/libs/1_65_1/libs/spirit/doc/html/spirit/qi/quick_reference/compound_attribute_rules.html

-- 
Alberto


------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Spirit-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/spirit-general