================
@@ -618,7 +618,7 @@ template <class InternT, class ExternT>
void utf8_to_utf16_in_error(const std::codecvt<InternT, ExternT, mbstate_t>&
cvt) {
// UTF-8 string of 1-byte CP, 2-byte CP, 3-byte CP, 4-byte CP
const unsigned char input[] = "b\u0448\uD700\U0010AAAA";
- const char16_t expected[] = {'b', 0x0448, 0xD700, 0xDBEA, 0xDEAA, 0};
+ const InternT expected[] = {0x62, 0x0448, 0xD700, 0xDBEA, 0xDEAA, 0};
----------------
EricWF wrote:
It seems to me that these tests are using the API as it was intended (before it
was removed for being bad)?
As in, they are using something like `std::codecvt_utf8_to_utf16` with the
char16_t type in their code.
And this set of changes removes test coverage for that case, because it
generates the warning about potential narrowing?
https://github.com/llvm/llvm-project/pull/138708
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
