Re: [boost] [CallableTraits] The formal review begins today

Subject: Re: [boost] [CallableTraits] The formal review begins today
From: Barrett Adair (barrettellisadair_at_[hidden])
Date: 2017-04-09 04:28:30

On Wed, Apr 5, 2017 at 6:12 PM, Edward Diener via Boost
<boost_at_[hidden]> wrote:
> On 4/3/2017 2:45 AM, Louis Dionne via Boost wrote:
>>
>> [snip]
>
> The documentation is confusing regarding the level of C++ conformance needed
> to use the library. At the beginning it says that the library is a
> C++11/C++14/C++17 library, whatever this means. Later information explains
> that it is only dependent on a C++11 standard library. There is mention of
> compatibility claims, which further confuses the issue. Finally there is
> mention of static asserts or substitution failures for some functionality.
>
> Frankly with all this confusion I would be very loath to use the library as
> the doc explains the issue(s). It would be much clearer if the library
> expressed the minimum level of C++ compliance needed for the majority of the
> functionality and then documented any greater level of C++ conformance
> needed for specific functionality. Also the library should document a
> standard behavior when the level of conformance is not met, either internal
> adjustment to a lower level of conformance, compiler errors, static asserts,
> or run-time exceptions, depending on how the library is designed, and these
> need to be carefully explained if they differed for different functionality.
>
> The number 3) item in comparing callable traits to the current Boost
> function types needs more explanation. Having used Boost function types in
> code ( Boost tti uses it, and I have used it in other programming endeavors
> ) I do not see how the callable traits library author could think that Boost
> function types encourages template specializations, so maybe an example
> showing what is meant by item 3) would be needed to back up this claim.
> Please note: I am certainly not against modernizing the function type
> interface; I just think that claims vis a vis Boost function types need to
> be backed up with actual proof by example.
>

Thanks for the excellent feedback, Edward. I agree with your criticisms.
I will improve the documentation after the review to clarify these issues.

>
> _______________________________________________
> Unsubscribe & other changes:
> http://lists.boost.org/mailman/listinfo.cgi/boost