|
Boost : |
From: Christian Mazakas (christian.mazakas_at_[hidden])
Date: 2022-05-23 15:33:22
> The use of NULL as a null pointer is perfectly valid, in the sense of
both
> formal C++ compliance and code readability.
In spirit, I wholeheartedly agree with this.
Unfortunately, this issue seems to come from a perfect storm of elements.
We've
had to solve something like this for Unordered as well.
https://github.com/boostorg/unordered/issues/39
It seems to be users including Boost as a non-system header and also using
`-Wzero-as-null-pointer-constant` with clang. We don't control how users
ultimately choose to add Boost to their projects and it should be a goal to
remove any warnings we generate in their builds, within reason.
In this case, using a simple macro `BOOST_NULLPTR` is easy and
straight-forward
for both maintainers and potential contributors.
- Christian
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk