> Date: Thu, 14 Feb 2019 12:50:09 +0000 > From: Colin Watson <cjwat...@debian.org> > > On Wed, Feb 13, 2019 at 06:58:18PM +0200, Eli Zaretskii wrote: > > Just to make what Keith says (and I concur) crystal clear: there's a > > need to distinguish between C99 compliance of the compiler and the > > C99/Posix compliance of the C runtime. We can assume the former, > > certainly when using MinGW GCC, but we cannot assume the latter when > > building a native MS-Windows port (as opposed to Cygwin port) of > > Groff. > > If it's just the runtime, then Gnulib should be able to paper over a > pretty fair number of the differences, and groff already uses that.
Up to a degree. There's no fork for Windows, for example, and many other functions are missing. > (It's possible that some of the _WIN32 conditionals can be supplied by > Gnulib these days, but there's also no great urgency to remove them, > IMO.) I doubt that. The vast majority of those I see in the current sources deal with issues that Gnulib cannot fix: . absence of SIGPIPE . backslashes as directory separators . differences in how argv[0] is presented to 'main' . different names of environment variables, like TEMP vs TMPDIR . quoting of command arguments 'like this' that isn't supported on Windows