|
1 | 1 | Revision history for the Perl extension Time::HiRes. |
2 | 2 |
|
| 3 | +1.9758_01 2018-09-06 rurban |
| 4 | + - merge with our old cperl fixes |
| 5 | + |
| 6 | +1.9758 [2018-03-21] |
| 7 | + - fix build in Win32 with Visual C by introducing a fake struct timezone |
| 8 | + [rt.cpan.org #124844] |
| 9 | + - in utime.t detect better being run in a noatime filesystem, |
| 10 | + and if so, skip the test (for the HAMMER filesystem of DragonflyBSD) |
| 11 | + - also for the HAMMER fs (if not in noatime), lower the expected subsecond |
| 12 | + timestamp granularity to microseconds |
| 13 | + - fix the version number typo in Changes: 1.9577 -> 1.9757 |
| 14 | + |
| 15 | +1.9757 [2018-03-15] |
| 16 | + - fix win32 (and cygwin?) builds which have been broken since 1.9755: |
| 17 | + problem was that gettimeofday() is a macro and needs to stay such |
| 18 | + - regenerate ppport.h with Perl 5.26.1 (was previously generated with |
| 19 | + 5.25.6) and Devel::PPPort 3.35 (previously 3.36) -- this doesn't |
| 20 | + change the ppport.h, though |
| 21 | + |
| 22 | +1.9756 [2018-03-14] |
| 23 | + - division by zero in new test tv_interval.t [rt.cpan.org #124775]: |
| 24 | + made the test pass even if the difference is zero, but also |
| 25 | + made it practically impossible to ever be zero by adding |
| 26 | + a sub-second sleep. |
| 27 | + - remove comments from inside qw [rt.cpan.org #124777], |
| 28 | + only seemed to be caught by Perl 5.18.4 |
| 29 | + |
| 30 | +1.9755 [2018-03-14] |
| 31 | + - adjust the error messages of tv_interval() |
| 32 | + - the NV_1EX constants now cast to be NV so that |
| 33 | + tv_interval() with long double builds does not |
| 34 | + produce ugly results [rt.cpan.org #106456] |
| 35 | + - add tests for tv_interval() |
| 36 | + - centralize the mygettimeofday() logic |
| 37 | + - make the mygettimeofday() a function instead of macro |
| 38 | + - add the OS X 10.12+ clock_gettime() constants |
| 39 | + CLOCK_MONOTONIC_RAW_APPROX CLOCK_UPTIME_RAW_APPROX |
| 40 | + - regenerate the fallback files |
| 41 | + |
| 42 | +1.9754 [2018-02-16] |
| 43 | + - unify the gettimeofday() and time() forked code |
| 44 | + that had near identical code paths for MacOS Classic |
| 45 | + (which has two unusual features: unsigned time_t, |
| 46 | + and a special tz struct member), and for non-MacOS Classic |
| 47 | + - tv_interval should be implemented in XS |
| 48 | + [rt.cpan.org #106456] |
| 49 | + thanks to Sergey Aleynikov (suggested implementation simplified |
| 50 | + by the above-mentioned MacOS Classic simplification) |
| 51 | + |
| 52 | +1.9753 [2018-01-11] |
| 53 | + - in t/clock.t in a fast system we need to burn more CPU, |
| 54 | + reported and fix suggested by Joel C. Maslak <jmaslak@antelope.net> |
| 55 | + [rt.cpan.org #124025] |
| 56 | + - for t/utime.t in netbsd we need /sbin in PATH to find the mount command, |
| 57 | + this is needed for the fix in 1.9751 to actually help netbsd with noatime |
| 58 | + mounts. |
| 59 | + (thanks to Nigel Horne) |
| 60 | + |
| 61 | +1.9752 [2018-01-04] |
| 62 | + - fix an error in the error message of utimensat() not available: |
| 63 | + it said futimens() not available |
| 64 | + - add --force alias for Makefile.PL --configure |
| 65 | + |
3 | 66 | 1.9751 [2018-01-02] |
4 | 67 | - in macOS/OSX/Darwin, use __has_builtin() check also for utimensat(), |
5 | 68 | can cause errors like |
|
0 commit comments