Takayuki Matsuoka

For some reason, GH actions for MinGW fails at the following line of the `Makefile` https://github.com/Cyan4973/xxHash/blob/fb93aa4c414ed40de580c7140e9dad16a56749b4/Makefile#L207 Error log ``` OK. (passes 49948 tests) 5ffb01494ce73724 stdin Error: unable to open input...

I can't reproduce above error in my local msys environment. But here is a different issue from my local environment: ``` # CFLAGS=-Werror make -C tests/collisions check Testing xxh3 algorithm...

@29039 While a manifest file is convenient, my goal is to provide broader compatibility. For example, I’d like to support “bad” Win32 filenames such as `my_file.` (filename with a trailing...

WIP. But now this PR supports long-path, UNC path and trailing period. e.g. `my_trailing_period_file.` `tests/windows/00-test-all.bat` builds and tests all. TODO: - [ ] Ask Cyan: We can allow to use...

Added cmake option `XXHASH_WIN_TARGET_WIN10` for MinGW / cross-compile environment. - cmake + MSVC user doesn't have to care about this option. - MinGW user who need long-path support, they should...

`PATHCCH_DO_NOT_NORMALIZE_SEGMENTS` (support trailing period and space) and `PATHCCH_ENSURE_IS_EXTENDED_LENGTH_PATH` (ensure `\\?\` prefix) are implemented since Windows 10 RS2 1703 as operation system DLL and its SDK. Unfortunately, earlier version of Windows...

@Cyan4973 , thanks. I forgot to add `@` to the question 😅

I think this warning (as an error) is simply `gcc-4.8`'s bug. Please refer "C99 §6.3.1.8 Usual arithmetic conversions". - By §6.3.1.8 p2, rank of unsigned and signed int is same....

I was wrong. I think we should recommend to use `MOREFLAGS` for compatibility issue. In this case, `MOREFLAGS=-std=c99` should be fine. Setup gcc-4.8 in Ubuntu 24.04 ----------------------------- I've followed [this...

Hi @Dayuxiaoshui Thanks for your proposal. Quick question: Does this proposal aim to implement RV64V support for LZ4 as a static library and `lz4`-CLI with compile-time configuration? Or aims further?...