>>>>> "Hrvoje" == Hrvoje Niksic <hniksic(a)srce.hr&gt; writes: Hrvoje> Charles G Waldman <cgw(a)pgt.com&gt; writes: >> On a more general note, why is HAVE_FSYNC being handled the way it >> is (in src/sysfile.h) instead of by the configure script? The >> existing logic seems convoluted - since Linux #defines USG, sysfile >> decides not to #define HAVE_FSYNC, even though Linux has fsync. Am >> I misunderstanding the purpose/meaning of the HAVE_FSYNC symbol? >> What does USG have to do with it? Hrvoje> Only Martin is truly competent to answer this, but if I had to guess, Hrvoje> I'd say it's a remnant of the pre-configure period, when all the Hrvoje> HAVE_* symbols were handles by s/m files. If we want to use Hrvoje> HAVE_FSYNC, we should simply check it via Autoconf. Martin here. There are still s&m files that include usg-*.h. I'd like to nuke them, but am afraid. HAVE_FSYNC looks like something that ought to be autoconfiscated, but I don't know enough about fsync(). In the dark ages before autoconf, everything was hard-coded into the s&m files. At that time, it looked pretty clever to have a usg5-4.h `subroutine'. One could just add fsync to the list of functions on that long AC_CHECK_FUNCS line in configure.in, and add #undef HAVE_FSYNC to src/config.h.