Shall we switch to C++17?

From: Mateusz Loskot (mateusz_at_[hidden])
Date: 2020-10-23 19:21:44

Hi,

Recently, there have been some discussions on #boost-gil
at cpplang.slack.com and in comments on GitHub issues
about switching GIL to C++17 [1]

TL;TR: Shall we do the switch and when?

We are a small team with limited time and man-power at our
disposal and we need to spend it wisely and effectively.
The development comfort and convenience has already been
improved by the switch to C++11, but there is still more to gain.

e.g. Samuel is one of users and active contributors (joined the
team lately), with very interesting ideas how to improve GIL.
>From the point of view of plain GIL attractiveness due to
succinct and expressive C++, availability of complete standard
library  and no fuss with maintaining support for compiler
(e.g. five years old GCC 5), I'm very keen in making his and
everyone else's time-limited work easier and more rewarding too.

So, paraphrasing Peter Dimov's sensible suggestion [2] and
according to the Boost release schedule [3]:

We may declare C++11 support deprecated in Boost 1.75 (December 2020)
and drop it in Boost release (1.77?) in August 2021.

I think it is a feasible plan with outcome not too distant in the future,
and GSoC 2021 could allow C++17 too.

I'd like to hear if there are any strong objections.

[1] C++14 is not under consideration
[2] https://pdimov.github.io/articles/phasing_out_cxx03.html
[3] https://github.com/boostorg/wiki/wiki/Releases%3A-Schedule

Best regards,
-- 
Mateusz Loskot, http://mateusz.loskot.net