<ranges>
: Fix regression with ranges::to
for ADL-only begin
/end
by frederick-vs-ja · Pull Request #5173 · microsoft/STL (original) (raw)
In #4539, the ranges::for_each
call in the resolution of LWG-4016 was "manually inlined" to reduce inclusion dependency. However, the inlining was not always correct, because builtin-in range-for
may just stop and fail when a deleted member begin
/end
function is encountered, while ranges
CPOs continue to consider ADL-found begin
/end
.
As a result, we should still use ranges::begin
/ranges::end
.
Fixes #5172.