Hi Richard, The gcc devs have apparently declined to do anything about this bug (it's marked as "RESOLVED WONTFIX") and don't care much about plugins I guess: <https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105171>, so I won't be sending a v2. However,
On Thu, Apr 7, 2022 at 8:30 AM Richard Biener <richard.guent...@gmail.com> wrote: > For the issue of overloading -1 I'd > instead suggest to do > > > struct timeval tv; > > gettimeofday (&tv, NULL); > > local_tick = (unsigned) tv.tv_sec * 1000 + tv.tv_usec / 1000; > /* Avoid aliasing with the special-meaning -1. */ > if (local_tick == -1) > local_tick = 1; > > because even with uint64_t the result could be -1, no? >>> (2**64 - 1 - 1000 - 1649334804) / 1000 / (365 * 24 * 60 * 60) 584942417.3027719 If we're still using this code in 584 million years, God help us all... But, anyway, if the devs in that bug report do want to fix this ultimately but don't want to change the type, your `if (local_tick == -1) local_tick = 1` thing there and in the next stanza seems like it'd fix the problem, and also might make things easier on plugins straddling versions. Jason
