On June 13, 2015 10:10:56 AM EDT, Edward Diener
On 6/13/2015 5:22 AM, Andrey Semashev wrote:
On 13.06.2015 05:26, Edward Diener wrote:
On 6/12/2015 1:05 PM, Peter Dimov wrote:
All __cpp_lib macros have an associated header, although I don't
know if
we currently have Boost.Config equivalents for any of them.
I can see them but they are all C++14 features. Are we really objecting to including a particular standard library header in order to test for the existence of the equivalent SD-6 macro ?
Yes, that would be the objection from my side, at least. I don't want to include a whole std header to check for a single feature that may be implemented in it. And I want Boost.Config to include most of STL even less.
And the rationale for this is what ? Compile-time computer cycles ?
Build times matter too large projects, CI build machines, etc. Extraneous includes are a real issue. Many Boost libraries are pinged for long build times. Includes have measurable affect on that, though perhaps not large. Ignoring those effects is not helpful and could lead to users abandoning Boost libraries. ___ Rob (Sent from my portable computation engine)