在 2025-7-9 21:09, Kirill Makurin 写道:
Here's the updated patches. They passed all tests[1].The reason was that when misc/btowc.c and misc/wctob.c were compiled for UCRT, they were using msvcr*.dll's mbstate_t which has different size. I added new misc/ucrt_btowc.c and misc/ucrt_wctob.c for UCRT compilation.I also discovered one case when wcrtomb would return incorrect value. The fix is included.
I think this series of patches now look fine, so I pushed them. There were some typos in the second one ('valie' => 'value') which I fixed. Thanks for the work.
I was a bit uncertain whether we should keep CJK characters like in this:
testcases/t_wcrtomb.c:165: wchar_t DBCS[] = {L'日', L'本', L'語', L'。'};
where I'd prefer
wchar_t DBCS[] = { 0x65E5, 0x672C, 0x8A9E, 0x3002 };
However given that the default input charset of GCC is UTF-8, they are probably
fine.
--
Best regards,
LIU Hao
OpenPGP_signature.asc
Description: OpenPGP digital signature
_______________________________________________ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
