On Wed, Mar 03, 2021 at 07:23:50AM +0100, Christian Eggers wrote: > On Tuesday, 2 March 2021, 20:58:22 CET, John Kacur wrote: > > > > On Fri, 26 Feb 2021, Christian Eggers wrote: > > > > > time_t is "long" or "long long" (depending on the system). So it should > > > be safe to convert it to "long long", in order to get a consistent data > > > type on all systems. > > > > > > Signed-off-by: Christian Eggers <ceggers@xxxxxxx> > > > --- > > > fprintf(f, " }%s\n", i == num_threads - 1 ? "" : ","); > > > [...} > > Are you seeing a compile time warning on some systems? > > Yes, I do. I have an ARM32 system with 64 bit time_t. > > I think that some of the printf format strings are simply wrong (e.g. using > PRIu64 for a "long int"). For time_t, some systems require "%ld" while others > require "%lld". Casting the values to "long long" seems to be easier compared > to have individual format strings for different platforms. Indeed, that was my fault. I've also send out patches for this a couple of weeks ago IIRC.
