Conversation
anonrig
force-pushed
the
yagiz/add-simdutf-constexpr
branch
from
f2c301a to
f63b351
Compare
anonrig
force-pushed
the
yagiz/add-simdutf-constexpr
branch
from
f63b351 to
d75c48e
Compare
|
This is reasonable. Let us wait for the tests to go green. |
| simdutf_really_inline void utf16fix_block(char16_t *out, const char16_t *in) { | ||
| const char16_t replacement = scalar::utf16::replacement<big_endian>(); | ||
| auto swap_if_needed = [](uint16_t c) -> uint16_t { | ||
| simdutf_constexpr auto swap_if_needed = [](uint16_t c) -> uint16_t { |
There was a problem hiding this comment.
this feels a bit misplaced?
I think the constexpr should go inside the lambda, with the ternary replaced with an if.
There was a problem hiding this comment.
Yes this is kind of unnecessary and can be moved into the correct place.
|
A related question: what would it take for us to require C++17 instead of C++11 as build requirement? Do we have any idea of how large proportion of the simdutf users are stuck at C++11? |
Obviously, I would just skip to C++20 if I could. Let us see what people use: |
|
I am blaming the failed CI tests on Cloudflare. @anonrig's PR should be generally safe. |
3 participants
Fixes #856
Up until the last commit in this PR, we didn't enabled C++17 on macos workflow even though our README said otherwise...