Re: [boost] The future and present of Boost

Subject: Re: [boost] The future and present of Boost
From: Edward Diener (eldiener_at_[hidden])
Date: 2018-10-25 01:50:44

On 10/24/2018 7:49 PM, Peter Dimov via Boost wrote:
> Edward Diener wrote:
> 
>> > If we up the minimum requirement
>>
>> What do you mean by the above ?
> 
> We announce that Boost has a minimum requirement of such and so, and 
> will no longer support platforms not meeting it starting with release 1.X.
> 
>> Suppose you made the minimum requirement C++11, what would you expect 
>> a library to do to meet that requirement ?
> 
> The library doesn't meet the minimum requirement, the compiler does. The 
> library may assume the minimum requirement.
> 
> As an example, if our minimum requirement implies a working 
> std::function, Boost.Test may use std::function in its interface and 
> implementation, unconditionally and without checking cxxstd or 
> BOOST_NO_CXX11_HDR_FUNCTIONAL or whatever else, and its maintainers can 
> in good conscience accept pull requests making use of other C++11 
> features whose availability is guaranteed by the minimum requirement.

Haven't we already basically agreed, going forward with Boost, that a 
library can use C++11 features and the C++11 Standard Library without 
informing end-users in advance that it is going to do so, if previously 
it supported C++03 ? If not, that was what I thought the previous long 
discussion about dropping support for C++03 was largely about. I do not 
think there is anything more that Boost can do to attempt to promote 
C++11 as the minimum standard it supports, at least at the level of each 
individual library. As far as Boost Test is concerned, or any other 
Boost library, I think it should use C++11 if it desires. If C++03 
library depends on a library which has gone from C++03 to C++11 then 
that C++03 library is now a C++11 library as long as it maintains that 
dependency.