fast_io
fast_io copied to clipboard
cppfastio
Significantly faster input/output for C++20
The original design is that all the formatting APIs should support all categories of iterators. However, with time, we discover that most iterators' categories are completely useless. Even if that...
it is just wrong to assume sizeof(wchar_t) would necessarily be 2 on windows. GCC even has toggle to change that. They should all use char16_t instead. char16_t must be the...
https://github.com/bminor/glibc/commit/e6547d635b991651600fab31f788ed5facd77610 As glibc's commit points out, loongarch is the first architecture that would no longer be for both fstat and stat. The reason is quite simple, statx is an objectively...
C++ operator new can always throw exceptions even if it is freestanding. Just remove this nonsense. Death to C++ new
Since C++23, more headers and features (like std::addressof) are available now. Therefore, we need to remove all the workaround. GCC provides extensions like making `` and `` be also in...
Windows Console requires decorators aware. Our new api should pass both the pointers from internal buffer and external buffer to the api so api can make a decision.
Re-locations with dlls will create copies across processes and might be very slow.
The write logic has been completely dogshit. The entire logic is just horribly wrong.
The code compiles for loongarch64-linux-gnu-g++. However, it does not work for clang for unknown reasons. I guess it is probably too new for clang (the target added to clang in...