> From: bastien ROUCARIES <roucaries.bast...@gmail.com> > Date: Fri, 30 Dec 2011 12:36:07 +0100 > Cc: bonz...@gnu.org, > bug-gnulib@gnu.org, > bug-g...@gnu.org > > > > In fact isatty under windows is equivalent to test if a file is a char > > > device. > > > > That is true. But unless a Windows user goes out of their way, the > > only character device they will ever see in routine operations is the > > null device. > Not true LPT is also a char device, and I use it for jtag stuff.
And the code I submitted says LPT is not a TTY. In fact, it works with every character device I tried: PRN, CLOCK$, and AUX. So I see no problem with the code I proposed. Which character devices succeed to dupe it? > The right fix is : > DWORD st; > /* note use _ version for porting to vc */ > if(!_isattty(fd)) > return 0; > if (!GetConsoleMode(_get_osfhandle(fd), &st)) > return 0; > return 1; Thanks. > This is the idea, it is a little bit more complicated due to INVALID_HANDLE > exception raising in _get_osfhandle(fd) but the idea > is here. And moreover erno handling. Further complications are the need to include windows.h, which could interact in nasty ways with standard C headers or even macros such as DATADIR. So if the code I proposed doesn't break in practice, I think its simplicity is a win.
