Thread: Is "WIN32" #defined in Cygwin builds?
I started wondering about $subject because we are fairly schizophrenic about whether we believe this. For example, only a few lines apart in dirmod.c, there are #if defined(WIN32) || defined(__CYGWIN__) #if defined(WIN32) && !defined(__CYGWIN__) Presumably, one of these could be simplified, but I'm not sure which. regards, tom lane
On 08/31/2015 02:21 PM, Tom Lane wrote: > I started wondering about $subject because we are fairly schizophrenic > about whether we believe this. For example, only a few lines apart in > dirmod.c, there are > > #if defined(WIN32) || defined(__CYGWIN__) > > #if defined(WIN32) && !defined(__CYGWIN__) > > Presumably, one of these could be simplified, but I'm not sure which. > > No, and we've made sure not to do that ourselves, or at least I hope we have. Here's what the cygwin compiler says: andrew@antonio ~/bf $ touch xx.c andrew@antonio ~/bf $ gcc -E -dM xx.c | grep WIN #define __WINT_MAX__ 4294967295U #define __WINT_MIN__ 0U #define__SIZEOF_WINT_T__ 4 #define __CYGWIN__ 1 #define __WINT_TYPE__ unsigned int #define __CYGWIN32__ 1 andrew@antonio ~/bf cheers andrew
Andrew Dunstan <andrew@dunslane.net> writes: > On 08/31/2015 02:21 PM, Tom Lane wrote: >> I started wondering about $subject because we are fairly schizophrenic >> about whether we believe this. > No, and we've made sure not to do that ourselves, or at least I hope we > have. OK, thanks. I was wondering whether I'd broken the Cygwin build with a65e0864, but it should be okay. Not sure if it's worth trying to clean up the #if tests that are redundant given this knowledge. It looks like there's about a dozen. regards, tom lane