On Thu, 25 Jan 2024 15:11:12 +0000 Jon Turney wrote: > On 24/01/2024 15:22, Takashi Yano via Cygwin-apps wrote: > > +#ifdef __CYGWIN__ > > +#define _GNU_SOURCE 1 > > +#endif > > + > > I'd suggest adding -D_GNU_SOURCE=1 to CPPFLAGS, rather than patching > this in everywhere it's needed. > > (Setting this unconditionally could be added to the build system > somewhere, which is theoretically upstreamable, since it's only not > needed on linux for :reasons:)
Sounds good! Thanks. I'll add it to CMakeLists.txt and Makfile for CYGWIN for the time being. > > +#ifdef __CYGWIN__ > > +typedef uint32_t DWORD > > Is it not possible to just #include <windows.h> here? If not, perhaps a > comment why the definitions are copied would nice. When I included <windows.h> here, many conflicts occured between posix headers and w32api headers. However, now I have found that including w32api headers after all the posix headers and typedef IN6_ADDR like: #ifdef __CYGWIN__ #include <windows.h> #include <pdh.h> #include <tlhelp32.h> typedef struct { union { u_char Byte[16]; u_short Word[8]; } u; } IN6_ADDR; #include <iphlpapi.h> #endif solves the conflicts. Thanks! -- Takashi Yano <takashi.y...@nifty.ne.jp>