"Fossies" - the Fresh Open Source Software Archive

Member "Time-HiRes-1.9764/Changes" (10 Aug 2020, 49830 Bytes) of package /linux/privat/Time-HiRes-1.9764.tar.gz:


As a special service "Fossies" has tried to format the requested text file into HTML format (style: standard) with prefixed line numbers. Alternatively you can here view or download the uninterpreted source code file. See also the last Fossies "Diffs" side-by-side code changes report for "Changes": 1.9758_vs_1.9760.

    1 Revision history for the Perl extension Time::HiRes.
    2 
    3 {{NEXT}}
    4 
    5 1.9764 [2020-08-10]
    6  - Fix a bunch of repeated-word typos
    7  - Fix compilation with Visual C++ 2013 and older
    8  - Implement clock_gettime() and clock_getres() for win32
    9  - Remove workaround for an ancient Devel::PPPort bug
   10  - Remove the remains of MacOS classic support
   11  - Remove code for perls older than 5.6
   12  - Remove vestiges of mpeix support (removed in 5.17.x)
   13  - Remove trailing blanks
   14  - Add links to pod
   15 
   16 1.9760 [2019-02-18]
   17  - make utime() available only if we have both fd and name setting
   18    [RT #133030]
   19  - Adjutst Makefile.PL for windows: the DEFINE() in Makefile.PL
   20    can't be in sub init() because that sub isn't called on windows
   21  - t/itimer.t: avoid race condition.
   22  - don't truncate nanosec utime
   23  - fallback/const-c.inc: Avoid compiler warning showing up on darwin.
   24 
   25 1.9759 [2018-03-23]
   26  - This version reverts the changes on CPAN that were included in 1.9754 - 1.9758.
   27    Several of these changes caused instability on some windows platforms. We can
   28    revisit these fixes if necessary in Perl 5.29.0
   29 
   30 1.9758 [2018-03-21]
   31  - fix build in Win32 with Visual C by introducing a fake struct timezone
   32    [rt.cpan.org #124844]
   33  - in utime.t detect better being run in a noatime filesystem,
   34    and if so, skip the test (for the HAMMER filesystem of DragonflyBSD)
   35  - also for the HAMMER fs (if not in noatime), lower the expected subsecond
   36    timestamp granularity to microseconds
   37  - fix the version number typo in Changes: 1.9577 -> 1.9757
   38 
   39 1.9757 [2018-03-15]
   40  - fix win32 (and cygwin?) builds which have been broken since 1.9755:
   41    problem was that gettimeofday() is a macro and needs to stay such
   42  - regenerate ppport.h with Perl 5.26.1 (was previously generated with
   43    5.25.6) and Devel::PPPort 3.35 (previously 3.36) -- this doesn't
   44    change the ppport.h, though
   45 
   46 1.9756 [2018-03-14]
   47  - division by zero in new test tv_interval.t [rt.cpan.org #124775]:
   48    made the test pass even if the difference is zero, but also
   49    made it practically impossible to ever be zero by adding
   50    a sub-second sleep.
   51  - remove comments from inside qw [rt.cpan.org #124777],
   52    only seemed to be caught by Perl 5.18.4
   53 
   54 1.9755 [2018-03-14]
   55  - adjust the error messages of tv_interval()
   56  - the NV_1EX constants now cast to be NV so that
   57    tv_interval() with long double builds does not
   58    produce ugly results [rt.cpan.org #106456]
   59  - add tests for tv_interval()
   60  - centralize the mygettimeofday() logic
   61  - make the mygettimeofday() a function instead of macro
   62  - add the OS X 10.12+ clock_gettime() constants
   63    CLOCK_MONOTONIC_RAW_APPROX CLOCK_UPTIME_RAW_APPROX
   64  - regenerate the fallback files
   65 
   66 1.9754 [2018-02-16]
   67  - unify the gettimeofday() and time() forked code
   68    that had near identical code paths for MacOS Classic
   69    (which has two unusual features: unsigned time_t,
   70    and a special tz struct member), and for non-MacOS Classic
   71  - tv_interval should be implemented in XS
   72    [rt.cpan.org #106456]
   73    thanks to Sergey Aleynikov (suggested implementation simplified
   74    by the above-mentioned MacOS Classic simplification)
   75 
   76 1.9753 [2018-01-11]
   77  - in t/clock.t in a fast system we need to burn more CPU,
   78    reported and fix suggested by Joel C. Maslak <jmaslak@antelope.net>
   79    [rt.cpan.org #124025]
   80  - for t/utime.t in netbsd we need /sbin in PATH to find the mount command,
   81    this is needed for the fix in 1.9751 to actually help netbsd with noatime
   82    mounts.
   83    (thanks to Nigel Horne)
   84 
   85 1.9752 [2018-01-04]
   86  - fix an error in the error message of utimensat() not available:
   87    it said futimens() not available
   88  - add --force alias for Makefile.PL --configure
   89 
   90 1.9751 [2018-01-02]
   91  - in macOS/OSX/Darwin, use __has_builtin() check also for utimensat(),
   92    can cause errors like
   93    "HiRes.xs:1474:16: error: unrecognized platform name macOS"
   94    [rt.cpan.org #123994]
   95    (oversight from 1.9749)
   96  - do not define TIME_HIRES_STAT/d_hires_stat if none was found, instead
   97    of defining it to be zero, which case has no implementation in hrstatns()
   98    (thanks to Nigel Horne)
   99  - in t/utime.t try to divine if the filesystem of the tempfiles has been
  100    mounted with the 'noatime' option, which can prohibit updating the
  101    access time timestamp.  Also document this in HiRes.pm.
  102    (thanks to Nigel Horne, original analysis by Slaven Rezic)
  103  - synchronize the constant lists in HiRes.pm:@EXPORT_OK
  104    and Makefile.PL:doConstants and regenerate fallback/const-c.inc
  105    and fallback/const-xs.inc, this fixes Perl 5.6.2 issue with
  106    d_futimens not allegedly being a valid macro in t/utime.t
  107    (using Perl 5.26.1 for the regenerating, not 5.6.2)
  108    (thanks to Nigel Horne)
  109  - in t/utime.t define a nop sub done_testing for ancient Perls
  110    (like Perl 5.6.2)
  111  - in Perl 5.6.2 a bogus warning
  112    "Use of uninitialized value in subroutine entry"
  113    is issued from t/alarm.t: add a comment documenting that
  114 
  115 1.9750 [2017-12-22]
  116  - update META.yml and META.json for XSLoader instead of DynaLoader
  117    [rt.cpan.org #123933]
  118 
  119 1.9749 [2017-12-21]
  120  - CPAN release
  121  - Quieten Time-HiRes test noise: blead 0f009486
  122  - Switch some core modules to XSLoader: blead b9a5a78f [rt.cpan.org #129789]
  123  - disable a flapping test for CI: blead cd2a978d
  124  - Avoid newGVgen in blead-upstream modules: blead 732d3893
  125  - Declaration after statement in typemap: blead d269f586
  126  - semicolon-friendly diagnostic control: blead 7347ee54
  127  - in macOS (OS X) check properly in runtime whether futimens()
  128    and utimensat() are available
  129 
  130 1.9748 [2017-12-16]
  131  - bleadperl only
  132  - semicolon-friendly diagnostic control: blead 7347ee54
  133 
  134 1.9747 [2017-09-13]
  135  - bleadperl only
  136  - Switch some core modules to XSLoader: blead b9a5a78f [rt.cpan.org #129789]
  137 
  138 1.9746 [2017-08-17]
  139  - Unreliable t/usleep.t and t/utime.t tests [rt.cpan.org #122819]
  140    Avoid testing for $dt = $t2 - $t1 and assuming $dt is less than
  141    some value since a heavily loaded machine can delay the $t2.
  142    Testing for greater than is fine.
  143  - Tweak the configuring messages about subsecond stat/utime.
  144 
  145 1.9745 [2017-08-14]
  146   - Properly check for futimens/utimensat actually doing something.
  147     This should handle 'gnukfreebsd' (which has only stubs, so far
  148     we have been skipping the test) and as a new thing 'gnu' (Hurd)
  149     (also only stubs).  Thanks to Nigel Horne.
  150   - Scan in t/utime.t whether the filesystem (wherever File::Temp
  151     puts it tempfiles) supports subsecond timestamps.  This removes
  152     the fragile Linux/ext3 specific hack.  As a side effect, the
  153     setting of subsecond timestamps is tested only if reading of
  154     them is supported.  Thanks to Carsten Gaebler for the test idea,
  155     and Ryan Voots for testing.
  156 
  157 1.9744 [2017-07-27]
  158   - add more potential clock constants, like CLOCK_MONOTONIC_FAST
  159     (available in FreeBSD), and not all potentially found clock
  160     constants were properly exported to be available from Perl,
  161     see your system's clock_gettime() documentation for the available ones
  162 
  163 1.9743 [2017-07-20]
  164   - correct declared minimum Perl version (should be 5.6, was declared
  165     as 5.8 since 1.9727_03): blead af94b3ac
  166   - fix the fix for 'do file' to load hints in Makefile.PL: blead 3172fdbc
  167 
  168 1.9742 [2017-04-16]
  169   - prefer 3-argument open: blead 1ae6ead9
  170   - fix dist/Time-HiRes/t/*.t that assumed '.' in @INC: blead 465db51d
  171   - fix cases where 'do file' should be 'do ./file'.: blead 8b69401c
  172 
  173 1.9741 [2016-11-20]
  174   - C++11 compatibility: blead a914236c
  175   - El Capitan compatibility: blead 45bbc013
  176   - use CLOCK_REALTIME for clock_nanosleep scan: blead 56ed632c
  177   - include file consistency in scans: blead 605b4dbe
  178   - use clockid_t consistently: blead 2979fdf3
  179   - use hv_fetchs(): blead 07e4dd7a
  180   - scan for clockid_t (needed for macos Sierra): blead 69d1f2c2
  181   - darwin lacks clockid_t: blead d1e7b5aa
  182     [rt.cpan.org #129789]
  183   - provide template for tempfile: blead 91dca83e
  184   - explicit cast for clockid_t in C++11: blead 1c566547
  185   - upgrade to a newer ppport (3.35)
  186 
  187 1.9740 [2016-09-25]
  188   - the ext3/ext2 filesystems do not have subsecond resolution,
  189     therefore skip the t/utime.t test
  190     [rt.cpan.org #116127]
  191 
  192 1.9739 [2016-06-28]
  193   - the upcoming macOS 10.12 (Sierra, the operating system formerly
  194     known as OS X, or Darwin) has implemented the clock_gettime()
  195     and clock_getres() interfaces but not the clock_nanosleep()
  196 
  197 1.9738 [2016-06-26]
  198   - fix the dates in the Changes, it's not July yet
  199     [rt.cpan.org #115553]
  200   - in cygwin, drop the utime.t expected timestamp resolution
  201   - documented for utime that the actual timestamp resolution depends
  202     on the operating system and the filesystem, just like for stat
  203   - looks like in gnukfreebsd 10.1 neither futimens or utimensat work
  204     even though they exist, skipping utime tests there for now
  205   - demodernize the Test::More use in utime.t (no subtest) to
  206     better support older Perl installations
  207   - for all the tests, drop the Test::More 0.82 requirement,
  208     the only "modern" thing used is the note routine which
  209     can be replaced with print("# ...\n") (or printf),
  210     with this works back at least with Test::More 0.54 (perl 5.8.7) 
  211 
  212 1.9737 [2016-06-22]
  213   - utime in HiRes.xs had dXSARGS which was unnecessary, leading
  214     to two markstack pops, panic, and two-minute wait for the watchdog
  215     (thanks to Dave Mitchell for the eagle-eye)
  216   - even though 5.10.x should have the Perl_ck_warner, many CPAN
  217     smokes proved otherwise, so requiring now at least 5.12 for
  218     Perl_ck_warner use in IS_SAFE_PATHNAME() 
  219 
  220 1.9736 [2016-06-22]
  221   - the IS_SAFE_PATHNAME() emulation was too modern for Perl 5.10
  222   - make it so that only one value is set for -DTIME_HIRES_STAT
  223     even on systems that support many options
  224   - the 1.9734 (only released in blead, and as part of the development
  225     release 5.25.2) fix for Darwin's clock_gettime was wrong under
  226     threaded builds, leading into mutex panics in clock.t 
  227   - the d_hires_utime determination was made more precise
  228   - the utime.t was failing in platforms where only one of
  229     futimens/utimensat was present (neither or both was okay)
  230 
  231 1.9735 [2016-06-20]
  232   - Time::HiRes should override `utime` to allow setting hires
  233     (futimens and utimensat support)
  234     [rt.perl.org #114809]
  235   - the utime patch used IS_SAFE_PATHNAME() which isn't available in
  236     too old Perls, so emulate (in case the Devel::PPPort is too old)
  237 
  238 1.9734 [2016-06-17]
  239   - fix Darwins with clock_gettime: blead 2d41a263
  240     [rt.perl.org #128427]
  241 
  242 1.9733 [2016-04-23]
  243   - C90 declaration-after-statement error with darwin threads: blead de1003b4
  244     [rt.cpan.org #113856]
  245 
  246 1.9732 [2016-03-13]
  247   - MUTEX_LOCK, not PERL_MUTEX_LOCK: blead e5b02b5d
  248   - also hrt_ualarm_itimer() is unused: 1cb6cce3
  249   - the mutex needs init: 2d639e20
  250   - version bump to 1.9732: 730d7fdc
  251 
  252 1.9731 [2016-03-13]
  253   - mark unused variable as such: blead a914236c
  254   - OS X emulation mutex accidentally unused: da7a6455
  255   - remove hrt_ualarm: 6da77c36
  256   - pod error fixes: 919ca095
  257   - nanosleep and clock_nanosleep cleanups: c8ea02b8..e3ff671b
  258   - static funcs in HiRes.xs: 4e922c26
  259   - Remove unused variable: 52ffb1b5
  260 
  261 1.9730 [2016-02-17]
  262   - TIME_HIRES_DONT_RUN_PROBES=1 to build the probes but not run them
  263     [rt.cpan.org #111391]
  264 
  265 1.9729 [2016-02-13]
  266   - new constants
  267     [rt.cpan.org #111170]
  268   - clock_getres(), clock_gettime(), clock_nanosleep() emulation for OS X
  269     [rt.cpan.org #78566]
  270 
  271 1.9728 [2015-11-13]
  272   - No changes since 1.9727_03
  273 
  274 1.9727_03 [2015-11-06]
  275   - use PERL_UNUSED_ARG() in the not-here-croak branches
  276   - Makefile.PL/metadata improvements
  277 
  278 1.9727_02 [2015-08-29]
  279   - correct documentation to note that Time::NVtime is NV, not double
  280     [perl #123879]
  281   - avoid a warning about clock_t on OS X / clang 3.6 - 3.6
  282     [rt.cpan.org #102718]
  283 
  284 1.9727_01 [2015-08-17]
  285   - Remove usage of 5.005 thread APIs in favor of 5.6 and up thread APIs
  286     [rt.cpan.org #102799].
  287   - Silence a C++ compiler warning [rt.cpan.org #101497].
  288   - respect $Config{run} for cross-compiling [rt.cpan.org #92607]
  289 
  290 1.9726 [2013-08-17]
  291 	- Correct s/us splitting of usleep(1000000) [rt.cpan.org #78266].
  292 	- Avoid integer overflow in itimer-based alarm() with large
  293 	  argument [rt.cpan.org #87160].
  294 	- Define PERL_NO_INLINE_FUNCTIONS during probe compilations, to
  295 	  avoid false negatives caused by not linking with the perl core.
  296 	- Be more careful about context in stat().
  297 	- Install into "site" directories by default on perl 5.11+
  298 	  [rt.cpan.org #79797].
  299 	- Fix a couple of doc typos [rt.cpan.org #85365].
  300 	- Fix function name in a doc example [rt.cpan.org #86318].
  301 	- Provide lstat() that yields high-res timestamps, alongside
  302 	  the existing high-res stat() [rt.cpan.org #78732].
  303 
  304 1.9725 [2012-02-01]
  305 	- Correct stack discipline in stat(), which was screwing up list
  306 	  operations in expressions containing calls to it [rt.cpan.org
  307 	  #72926].
  308 	- Add missing OUTPUT sections to the XS code [rt.cpan.org #70930].
  309 	- Skip itimer tests on GNU/Hurd, which has the API but lacks
  310 	  the implementation [rt.cpan.org #72754].
  311 	- Fix a doubled word in the documentation [rt.cpan.org #72763].
  312 
  313 1.9724 [2011-06-09]
  314 	- Correct XS parameter list, and therefore prototype, for
  315 	  unimplemented-on-this-platform version of clock_nanosleep()
  316 	  [rt.cpan.org #68700].
  317 	- Declare package variables with "our" rather than "use vars".
  318 	- Corresponding to "our" usage, check for minimum Perl version
  319 	  5.006.
  320 	- Declare module dependencies.
  321 
  322 1.9723 [2011-06-07]
  323 	- Remove $ENV{PERL_CORE} logic from test suite, which is no
  324 	  longer desired in the core.
  325 	- Convert test suite to use Test::More.
  326 	- Factor out watchdog code from test suite.
  327 	- In test suite, be consistent about using fully-qualified form
  328 	  of function names.
  329 	- Divide test suite into feature-specific scripts.
  330 	- Make ualarm timing test less vulnerable to delay-induced false
  331 	  failure, from Dave Mitchell.
  332 
  333 1.9722 [2011-05-18]
  334 	- Update for changes in build process in the core, patches
  335 	  from BinGOs [rt.cpan.org #58858] and Craig Berry [rt.cpan.org
  336 	  #63363].
  337 	- Fix broken linkage on Windows with gcc 3.4 seen with ActivePerl,
  338 	  report from Christian Walde [rt.cpan.org #61648], fix derived
  339 	  from Vincent Pit.
  340 	- Jump through hoops to avoid compiler warnings.
  341 
  342 1.9721 [2010-03-17]
  343 	- Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
  344 	  additional fix from Gisle Aas
  345 	- Address [rt.cpan.org #55665] "Bad plan" on Windows,
  346 	  report and fix from Jan Dubois
  347 
  348 1.9720 [2010-02-14]
  349 	- Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
  350 	  report and fix from Nicholas Clark
  351 
  352 1.9719	[2009-01-04]
  353 	- As with QNX, Haiku has the API of interval timers but not
  354 	  the implementation (bleadperl change #34630), hence skip
  355 	  the tests, via David Mitchell.
  356 
  357 1.9718	[2008-12-31]
  358 	- .xs code cleanup from Albert Dvornik
  359 	- in the #39 and #40 do not do us I did, mixing alarm() and
  360 	  sleep().  Now instead spin until enough time has passed.
  361 
  362 1.9717	[2008-12-30]
  363 	- Skip the tests added in 1.9716 (#39, #40) if there's no subsecond
  364 	  alarm capability, like with the older subsecond alarm tests
  365 
  366 1.9716	[2008-12-26]
  367 	- Change documentation to agree with reality: there are
  368 	  no interval timers in Win32.
  369 	- Address [rt.cpan.org #35899] (problem in subsecond sleeps),
  370           add two tests to guard against this problem
  371 	- Address [rt.cpan.org #36600] 'Division by zero' failure in test suite
  372 	- Address [rt.cpan.org #37340] [PATCH] Address timer process in test
  373 	- Address [rt.cpan.org#40311 ] bad implementation of hrt_usleep
  374           with TIME_HIRES_NANOSLEEP
  375 
  376 1.9715	[2008-04-08]
  377 	- Silly me: Makefile.PL does need to accept arguments other than mine.
  378 	  Some testing frameworks obviously do this.
  379 	- Add retrying for tests 34..37, which are the most commonly
  380 	  failing tests.  If this helps, consider extending the retry
  381 	  framework to all the tests.  [Inspired by Slaven Rezic,
  382 	  [rt.cpan.org #34711] Occasional failures of test 35 or 36 (FreeBSD)]
  383 
  384 1.9714	[2008-04-07]
  385 	- Under Perl 5.6.* NVgf needs to be "g", reported by Zefram,
  386 	  it seems that ppport.h 3.13 gets this wrong.
  387 	- remove the check in Makefile.PL for 5.7.2, shouldn't be
  388 	  (a) necessary (b) relevant
  389 	- add logic to Makefile.PL to skip configure/write Makefile
  390 	  step if the "xdefine" file already exists, indicating that
  391 	  the configure step has already been done, one can still
  392 	  force (re)configure by "perl Makefile.PL configure",
  393 	  or of course by "make clean && perl Makefile.PL".
  394 
  395 1.9713	[2008-04-04]
  396 	- for alarm() and ualarm() [Perl] prefer setitimer() [C]
  397 	  instead of ualarm() [C] since ualarm() [C] cannot portably
  398 	  (and standards-compliantly) be used for more than 999_999
  399 	  microseconds (rt.cpan.org #34655)
  400 	- it seems that HP-UX has started (at least in 11.31 ia64)
  401 	  #defining the CLOCK_REALTIME et alia (instead of having
  402 	  them just as enums)
  403 	- document all the diagnostics
  404 
  405 1.9712	[2008-02-09]
  406 	- move the sub tick in the test file back to where it used to be
  407 	- in the "consider upgrading" message recommend at least Perl 5.8.8
  408 	  and make the message to appear only for 5.8.0 since 5.8.1 and
  409 	  later have the problem fixed
  410 	- VOS tweak for Makefile (core perl change #33259)
  411 	- since the test #17 seems to fail often, relax its limits a bit
  412 
  413 1.9711	[2007-11-29]
  414 	- lost VMS test skippage from Craig Berry
  415 	- reformat the test code a little
  416 
  417 1.9710	[2007-11-29]
  418 	- I got the sense of the QNX test the wrong way in an attempt
  419 	  to generalize it for future
  420 
  421 1.9709	[2007-11-28]
  422 	- casting fixes from Robin Barker for g++ and 64bitint
  423 	- in QNX skip the itimer tests because though the API
  424 	  is there, the implementation isn't, from Matt Kraai
  425 	- raise the dead man timer to 180 seconds for really
  426 	  slow/busy systems
  427 	- elaborate the UTF-8 locale warning from Makefile.PL
  428 
  429 1.9708	[2007-10-05]
  430 	- [rt.cpan.org #29747]: Build failure with perl 5.005_05
  431 	  Fixed by regenerating the ppport.h using Devel::PPPort 3.13.
  432 
  433 1.9707	[2007-02-27]
  434 	- t/HiRes.t failed in Perl 5.6.2,
  435 	  "action is not of type POSIX::SigAction at t/HiRes.t line 318",
  436 	  reported and fixed by Anton Berezin, the reason was faulty
  437 	  use of sigaction() when restoring the old action.
  438 
  439 1.9706	[2007-02-25]
  440 	- with bleadperl in VMS the HiRes.t overrun the maximum number
  441 	  of deferred signals because the libc SIGALRM was not strong
  442 	  enough to interrupt select(), and select() got restarted every
  443 	  time, solution is to use POSIX::SigAction if available.
  444 	  A fix from Craig Berry (not 100% there, but helps).
  445 	- allow for more measuring noise for ualarm() tests 35..37
  446 
  447 1.9705	[2007-02-06]
  448 	- nanosleep() and clock_nanosleep() detection and use were
  449 	  quite broken; in Linux -lrt needed; fixes from Zefram
  450 	- [internal] slightly cleaner building of $DEFINE in Makefile.PL,
  451 	  should avoid double/conflicting -D flags
  452 
  453 1.9704	[2007-01-01]
  454 	- allow 10% of slop in test #14 (testing difference between
  455 	  CORE::time() and Time::HiRes::time()), there seem to be often
  456 	  transient failures from Perl smoke builds on this test
  457 	- small pod tweaks
  458 
  459 1.9703	[2006-12-08]
  460 	- use int main(int argc, char **argv) consistently in Makefile.PL,
  461 	  should help with
  462 	  [rt.cpan.org #23868] nanosleep not detected under Mac OS 10.3.9 starting with Time::HiRes 1.96
  463 	- if someone still has the locale-broken Perl 5.8.0,
  464 	  suggest that they upgrade their Perl
  465 
  466 1.9702	[2006-12-06]
  467 	- restore the -DATLEASTFIVEOHOHFIVE, Win32 needed it still
  468 
  469 1.9701	[2006-12-04]
  470 	- upgrade to ppport.h 3.10_02
  471 	- remove the -DATLEASTFIVEOHOHFIVE
  472 	- use the ppport.h PL_ppaddr, PL_statcache, PL_laststatval
  473 	- use the ppport.h aTHXR for calling Perl stat()
  474 	- switch into four-digit version since 2.0 is coming up
  475 	  awfully fast but not feeling like a major rewrite
  476 
  477 1.97	[2006-11-30]
  478 	- 1.95 broke building in Win32 (since pp_stat is not exported),
  479 	  figured out how to call an op directly in 5.005 (use Perl_ppaddr
  480 	  instead of PL_ppaddr)
  481 	- backport to Perl 5.004_05  (requires using statcache
  482 	  and laststatval instead of PL_statcache and PL_laststatval)
  483 	  (also checked to work in 5.005_04, 5.6.1, and 5.8.8 with threads)
  484 
  485 1.96	[2006-11-30]
  486 	- 1.95 broke builds for threaded Perls, rt.cpan.org tickets:
  487 	  [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
  488 	  [rt.cpan.org #23712] Time-HiRes 1.95 Fails make on AIX 5.2 with Perl 5.8.8
  489 	  [rt.cpan.org #23730] Time::HiRes 1.95 fails make on MacOS X 10.3.9/perl 5.8.8
  490 	- use main() prototype consistently in Makefile.PL
  491 
  492 1.95	[2006-11-29]
  493 	- integrate core change #29180: Silence VC++ compiler warnings
  494 	  from Steve Hay
  495 	- do not use PL_ppaddr in stat() because that is not available
  496 	  in Perl 5.005_04
  497 	- regenerate fallback/*.inc for older Perls without
  498           ExtUtils::Constant because of d_hires_stat, resolves
  499 	  [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
  500 	- Make Makefile.PL more defensive against false PERL_CORE
  501 
  502 1.94	[2006-10-16]
  503 	- file timestamps oddities seen: the atime and mtime
  504 	  can be out of sync (modify first and read second can leave
  505 	  atime < mtime) and mtime can be subsecond while atime is not.
  506 	  So make the test more forgiving.
  507 
  508 1.93	[2006-10-15]
  509 	- the ualarm() tests (34-37) assumed that ualarm(N)
  510 	  could never alarm in less than N seconds, widened
  511 	  the acceptable relative range to 0.9..1.5.  Addresses
  512 	  [rt.cpan.org #22090] and [rt.cpan.org #22091].
  513 
  514 	- skip the stat() tests in cygwin and win32, because
  515 	  if run on FAT the timestamp granularity is only 2 seconds.
  516 	  Any good way to detect (cygwin or win32) whether we are
  517 	  being run on NTFS or anywhere with better timestamps?
  518 	  Addresses [rt.cpan.org #22089] and [rt.cpan.org #22098].
  519 
  520 1.92	[2006-10-13]
  521 	- scan for subsecond resolution timestamps in struct stat,
  522 	  some known possibilities:
  523 
  524 	  (1)  struct  timespec st_atimespec;
  525                                 st_atimespec.tv_nsec;
  526 	  (2)  time_t  st_atime;
  527 	       long    st_atimensec;
  528           (3)  time_t  st_atime;
  529                int     st_atime_n;
  530 	  (4)  timestruc_t st_atim;
  531                            st_atim.tv_nsec
  532 	  (5)  time_t  st_atime;
  533                int     st_uatime;
  534 
  535           If something like this is found, one can do
  536 
  537 	    use Time::HiRes;
  538 	    my @stat = Time::HiRes::stat();
  539 
  540 	  or even override the standard stat():
  541 
  542 	    use Time::HiRes qw(stat);
  543 
  544           to get the stat() timestamps
  545 
  546 	    my ($atime, $mtime, $ctime) = @stat[8, 9, 10];
  547 
  548 	  with subsecond resolution (assuming both the operating
  549 	  system and the filesystem support that kind of thing).
  550 
  551 	  Contributions for more systems (especially non-UNIX,
  552 	  e.g. but not limited to: Win32, VMS, OS/2) gladly accepted.
  553 	  (also more UNIX variants welcome: HP-UX? IRIX?)
  554 
  555 	  Thanks to H.Merijn Brand, John Peacock, and Craig
  556 	  Berry for brave beta testing.
  557 
  558 1.91	[2006-09-29]
  559 	- ualarm() in SuSE 10.1 was overflowing after ~4.2 seconds,
  560 	  possibly due to a glibc bug/feature (suspected overflow at
  561 	  2**32 microseconds?), workaround by using the setitimer()
  562 	  implementation of ualarm() if either useconds or
  563 	  interval > 999_999 (this case seems to vary between systems:
  564 	  are useconds more than 999_999 for ualarm() defined or not)
  565 	  Added more ualarm() tests to catch various overflow points,
  566 	  hopefully no problems in various platforms.
  567 	  (The problem report by Mark Seger and Jon Paul Sullivan of HP.)
  568 
  569 1.90	[2006-08-22]
  570 	- tweak still needed for Const64(), from Jerry Hedden
  571 	- get a freshly generated ppport.h
  572 	- update Copyright years
  573 
  574 1.89	[2006-08-22]
  575 	- Const64() already appends an 'LL' (or i64), so provide LL and i64
  576 	  forms for the IV_1E[679] (effects Win32 and Cygwin), reported by
  577 	  Jerry Hedden.
  578 	- the Changes entry for 1.88 talked about [IN]V_1[679],
  579 	  missing the 'E'.
  580 
  581 1.88	[2006-08-21]
  582 	- clean up the g++ warnings in HiRes.xs, all of them
  583 	  about mixing integer and floating point, introduce
  584 	  constants IV_1E[679] and NV_1E[679]
  585 
  586 1.87	[2006-02-13]
  587 	- [rt.cpan.org #17442] 'make test' frequently fails under
  588 	  Cygwin Perl v5.8.8, reported and patched by J. R. Hedden
  589 	  (two race condition bugs in the END block in the case the
  590 	   main process dies before the timer process, unearthed
  591 	   by a bug in Cygwin ualarm)
  592 
  593 1.86	[2005-12-17]
  594 	- HiRes.t:s/ok 32/ok 33/, from Dominic Dunlop
  595 	- tighten up the clock() test marginally by requiring non-negative
  596 	- clock_nanosleep() and clock() doc tweaks
  597 
  598 1.85	[2005-12-16]
  599 	- the interface to clock_nanosleep() is more natural
  600 	  when it is like (hires) time() (instead of like nanosleep),
  601 	  and the .xs implementation of clock_nanosleep() in 1.84
  602 	  was broken anyway
  603 	- the semantics of clock() are not quite so silly as I thought,
  604 	  but still somewhat odd, documented as such
  605 	- additional enhancements to the clock() documentation
  606 	- add test for clock_nanosleep() (I cannot test this
  607 	  since none of my systems have the function)
  608 	- add test for clock()
  609 
  610 1.84	[2005-12-16]
  611 	- add clock() which returns the processor time in
  612 	  (floating point) seconds since an arbitrary era
  613 	- add clock_nanosleep() which suspends the current
  614 	  thread until either absolute time or for relative time
  615 	- [rt.cpan.org #16486] printf missing value in HiRes.t
  616 	- add constants CLOCKS_PER_SEC, CLOCK_SOFTTIME, TIMER_ABSTIME
  617 	- tiny typo fixes
  618 
  619 1.83	[2005-11-19]
  620 	- has_symbol() was wrong since e.g. ITIMER_VIRTUAL is exported
  621 	  via @EXPORT_OK even when it is not available.  This is heinous.
  622 	  @EXPORT_OK should be determined at Makefile.PL time.
  623 	- be more lenient is testing clock_gettime(): allow more slop,
  624 	  and retry up to three times, sleeping a random nap between
  625 	  the retries
  626 	- human months are one-based (noticed by Anton Berezin)
  627 
  628 1.82	[2005-10-06]
  629 	- CLOCK_REALTIME is an enum value (of the clockid_t enum)
  630 	  in HP-UX (and might be so elsewhere, too), debugged by
  631 	  H. Merijn Brand
  632 	- include const-c.inc as late as possible (from Randy Kobes,
  633 	  [rt.cpan.org #15552] to avoid undefined usleep() on Win32
  634 
  635 1.81	[2005-11-05]
  636 	- try to be more robust and consistent in the detection of
  637           CLOCK_REALTIME and ITIMER_VIRTUAL in HiRes.t: the proper
  638 	  way is
  639 
  640 		sub has_symbol {
  641 		    my $symbol = shift;
  642 		    eval 'import Time::HiRes qw($symbol)';
  643 		    return 0 unless $@ eq '';
  644 		    return exists ${"Time::HiRes::$symbol"};
  645 		}
  646 
  647 	  and then use
  648 
  649 		&FOO_BAR
  650 
  651 	  in the test.  All these moves are needed because
  652 
  653 	  1) one cannot directly do eval 'Time::HiRes::FOO_BAR'
  654 	     because FOO_BAR might have a true value of zero
  655 	     (or in the general case an empty string or even undef)
  656 
  657 	  2) In case FOO_BAR is not available in this platform,
  658 	     &FOO_BAR avoids the bareword warning
  659 
  660 	- wait more (1.5 seconds instead of 0.1) for the CLOCK_REALTIME test
  661 	  but expect the 'customary' slop of 0.20 instead of 0.25
  662 	- fixed inside a comment HAS_POLL -> TIME_HIRES_NANOSLEEP
  663 	- at the end of HiRest.t tell how close we were to termination
  664 
  665 1.80	[2005-11-04]
  666 	- Gisle noticed a mistake (using HAS_NANOSLEEP) in 1.79
  667 
  668 1.79	[2005-11-03]
  669 	- try nanosleep for emulating usleep -- may help in some weird
  670 	  embedded realtime places which have nanosleep but neither usleep
  671 	  nor select nor poll (doesn't have to be weird embedded realtime
  672 	  place, though -- in many places usleep is nanosleep anyway)
  673 	- try poll for emulating usleep -- this may help some obscure/old
  674 	  SVR4 places that have neither usleep nor select
  675 	- a redundant test guard in HiRes.t
  676 
  677 1.78	[2005-11-03]
  678 	- ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
  679 	  in the general case fail already at 'use' phase is suboptimal)
  680 	- fixes to the documentation of clock_gettime() and clock_getres()
  681 
  682 1.77	[2005-11-03]
  683 	- add support for the POSIX clock_gettime() and clock_getres(),
  684 	  if available, either as library calls or as syscalls
  685 	- be more defensive about missing functionality: break out
  686 	  early (during 'use') if no e.g. clock_getres() is available,
  687 	  and protect our back by trapping those cases also in HiRes.xs
  688 	- the test added in 1.76 could cause an endless loop e.g. in Solaris,
  689 	  due to mixing of sleep() and alarm() (bad programmer, no cookie!)
  690 
  691 1.76	[2005-10-22]
  692 	- testing for nanosleep had wrong logic which caused nanosleep
  693 	  to become undefined for e.g. Mac OS X
  694 	- added a test for a core dump that was introduced by Perl 5.8.0
  695 	  safe signals and was fixed for the time of 5.8.1 (one report of
  696 	  the core dump was [perl #20920]), the test skipped pre-5.8.1.
  697 	- *cough* s/unanosleep/nanosleep/g; *cough*
  698 
  699 1.75	[2005-10-18]
  700 	- installation patch from Gisle Aas: in Perls 5.8.x and later
  701 	  use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
  702 
  703 1.74	[2005-09-19]
  704 	- [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
  705 	  (the workaround is not to use rel2abs, should not be necessary)
  706 	- [cpan #14642] U2time wrongly exported on the C API
  707 	  (patch supplied by the reporter, SALVA@cpan.org)
  708 	- add release dates to Changes
  709 
  710 1.73	[2005-08-16]
  711 	- Time::HiRes::nanosleep support for Solaris [PATCH]
  712 	  (POSIX::uname() not available if building with core perl,
  713 	   from Gisle Aas, via perl5-porters, perl change #25295)
  714 
  715 1.72	[2005-07-01]
  716 	- going back to the 1.68 loader setup (using DynaLoader)
  717 	  since too many weird things starting breaking
  718 	- fix a typo in José Auguste-Etienne's name
  719 
  720 1.71	[2005-06-28]
  721 	- a thinko in the nanosleep() detection
  722 	- move more changes stuff from the README to Changes
  723 	- add -w to the Makefile.PL
  724 
  725 1.70	[2005-06-26]
  726 	- oops in 1.69 about @ISA (not affecting anything but silly)
  727 	- add copyright 2005 to HiRes.pm
  728 	- add copyright and license to HiRes.xs
  729 	- add copyrights 2003, 2004, 2005 to README
  730 
  731 1.69	[2005-06-25]
  732 	- actually run a test for nanosleep
  733 	  (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
  734 	  it seems that one can link in nanosleep() but then calling
  735 	  it fails instantly and sets errno to ENOSYS (Not implemented).
  736 	  This may be fixable in the AIX case by figuring out the right
  737 	  (realtime POSIX?) libs and whatnot, but in the general case
  738 	  running a real test case is better.  (Of course, this change
  739 	  will no doubt run into portability problems because of the
  740 	  execution step...)  Note that because of hysterical raisins
  741 	  most Perls do NOT have $Config{d_nanosleep} (scanning for
  742 	  it by Configure would in many platforms require linking in
  743 	  things like -lrt, which would in many platforms be a bad idea
  744 	  for Perl itself).
  745 	  (from José Auguste-Etienne)
  746 	- support XSLoader also since it's much faster
  747 	  (from Alexey Tourbin)
  748 	- add SEE ALSO (BSD::Resource and Time::TAI64)
  749 
  750 1.68	[2005-05-14]
  751 	- somehow 1.67 had a lot of doubled lines (a major cut-and-paste
  752 	  error suspected), but miraculously it still worked since the
  753 	  doubling took place below the __END__ token
  754 	- undef Pause() before defining it to avoid redefinition warnings
  755 	  during compilation in case perl.h had already defined Pause()
  756 	  (part of perl change #24271)
  757 	- minor doc tweaks
  758 
  759 1.67	[2005-05-04]
  760 	- (internal) don't ignore the return value of gettimeofday()
  761 	- (external) return undef or an empty if the C gettimeofday() fails
  762 	  (affects Time::HiRes gettimeofday() and the hires time())
  763 
  764 1.66	[2004-12-19]
  765 	- add nanosleep()
  766 	- fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
  767 	- should now build in Solaris [rt.cpan.org #7165] (since 1.64)
  768 	- should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
  769 	- close also [rt.cpan.org #5933] "Time::HiRes::time does not
  770 	  pick up time adjustments like ntp" since ever reproducing it
  771 	  (and therefore verifying a possible fix) in the same environment
  772 	  has become rather unlikely
  773 
  774 1.65	[2004-09-18]
  775 	- one should not mix u?alarm and sleep (the tests modified
  776 	  by 1.65, #12 and #13, hung in Solaris), now we just busy
  777 	  loop executing an empty block
  778 	- in the documentation underline the unspecificity of mixing
  779 	  sleeps and alarms
  780 	- small spelling fixes
  781 
  782 1.64	[2004-09-16]
  783 	- regenerate ppport.h with Devel::PPPort 3.03,
  784 	  now the MY_CXT_CLONE is defined in ppport.h,
  785 	  we no more need to do that.
  786 
  787 	- the test #12 would often hang in sigsuspend() (at least that's
  788 	  where Mac OS X' ktrace shows it hanging).  With the sleep()s
  789 	  changed to sleep(1)s, the tests still pass but no hang after
  790 	  a few hundred repeats.
  791 
  792 1.63	[2004-09-01]
  793 	- Win32 and any ithread build: ppport.h didn't define
  794 	  MY_CXT_CLONE, which seems to be a Time-HiRes-ism.
  795 
  796 1.62	[2004-08-31]
  797 	- Skip testing if under PERL_CORE and Time::HiRes has not
  798 	  been Configured (from Marcus Holland-Moritz, core change
  799 	  #23246)
  800 	- Use ppport.h generated by Devel::PPPort 3.01,
  801 	  allowing cutting away our own portability code.
  802 	- Don't use $ENV{PERL_CORE} for < 5.6.0.
  803 	- Don't use "for my $i" for <= 5.003.
  804 	- Don't use Pause() for <= 5.003.
  805 	- Can't use newSVpvf for <= 5.003.
  806 	(most of the changes from Marcus)
  807 
  808 1.61	[2004-08-21]
  809 	- Win32: reset reading from the performance counters every
  810 	  five minutes to better track wall clock time (thanks to
  811 	  PC timers being often quite bad), should help long-running
  812 	  programs.
  813 
  814 1.60	[2004-08-15]
  815 	- Win32: Patch from Steve Hay
  816 	  [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
  817 	  to [perl #30755] reported by Nigel Sandever
  818 
  819 	- Cygwin: Use the Win32 recalibration code also in Cygwin if the
  820 	  <w32api/windows.h> APIs are available.  Cygwin testing by
  821 	  Yitzchak Scott-Thoennes.
  822 
  823 	- Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
  824 	  after that keep using -lrt, patch from Alan Burlison,
  825 	  bug reported in [cpan #7165]
  826 
  827 1.59	[2004-04-08]
  828 	- Change the Win32 recalibration limit to 0.5 seconds and tweak
  829 	  the documentation to blather less about the gory details of the
  830 	  Win32 implementation and more about the complications in general
  831 	  of meddling with the system clock.
  832 
  833 1.58	[2004-04-08]
  834 	- Document the 1.57 change better.
  835 
  836 1.57	[2004-07-04]
  837 	- Win32/Cygwin/MinGW: if the performance counter drifts by more
  838 	  than two seconds from the system clock (due to ntp adjustments,
  839 	  for example), recalibrate our internal counter: from Jan Dubois,
  840 	  based on [cpan #5933] by Jerry D. Hedden.
  841 
  842 1.56	[2004-29-02]
  843 	- Give a clearer message if the tests timeout (perl change #22253)
  844 	- Don't use /tmp or its moral equivalents (perl bug #15036,
  845 	  perl change #22258)
  846 
  847 1.55	[2004-01-14]
  848 	- Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
  849 	  instead of VC-specific i64 suffix)
  850 
  851 1.54	[2003-12-31]
  852 	- Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
  853 
  854 1.53	[2003-12-30]
  855 	- Windows: higher resolution time() by using the Windows
  856 	  performance counter API, from Jan Dubois and Anton Shcherbinin.
  857 	  The exact new higher resolution depends on the hardware,
  858 	  but it should be quite a bit better than using the basic
  859 	  Windows timers.
  860 
  861 1.52	[2003-10-28]
  862 	- In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
  863 	  the subtest 18.  No known analysis nor fix, but added
  864 	  an alarm (that requires fork() and alarm()) to the test.
  865 
  866 1.51	[2003-09-22]
  867 	- doc tweaks from mjd (perl change #20456)
  868 	- NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
  869 
  870 1.50	[2003-08-02]
  871 	- add a message (for non-core builds) to Makefile.PL about
  872 	  the LC_ALL=C workaround
  873 	- &Time::HiRes::d_nanosleep was broken (perl change #20131)
  874 	- the nanosleep() probe was broken (perl change #20061)
  875 	- use existence instead of definedness for feature probes
  876 	  (perl change #20043)
  877 	- MPE/iX tweak (perl change #20042)
  878 	- do not use HAS_NANOSLEEP (perl change #19898)
  879 
  880 1.49	[2003-06-23]
  881 	- UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
  882 	- OS/2 can always mix subsecond sleeps with signals
  883 	  (part of perl change #19789)
  884 
  885 1.48	[2003-06-04]
  886 	- workaround for buggy gcc 2.95.3 in openbsd/sparc64
  887 	  (perl change #19592)
  888 
  889 1.47	[2003-05-03]
  890 	- do not use -lrt in Linux (from March Lehmann, perl change #19449)
  891 		- unnecessary (nanosleep is in libc anyway)
  892 		- harmful (-lrt slows down execution)
  893 		- incompatible (with many distributions' pthreads)
  894 
  895 1.46	[2003-04-25]
  896 	- do not create files in blib directories under core
  897 	  (perl change #19160, from rgs)
  898 	- detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
  899 
  900 1.45	[2003-04-01]
  901 	- guarantee that $xdefine in HiRes.t is always defined
  902 	  (perl change #19109, from IlyaZ)
  903 	- a cleaner way to detect PERL_CORE (perl change #19111,
  904 	  from IlyaZ)
  905 
  906 1.44	[2003-03-30]
  907 	- add hints/irix.pl to turn off overly POSIX flags that
  908 	  cause hide struct timespec to be hidden (and compilation
  909 	  to fail) (bleadperl change #19085)
  910 	- documentation tweaks
  911 
  912 1.43	[2003-03-11]
  913 	- add c:/temp to the list of temp directories to probe
  914 	  so that cygwin (and win*?) builds are happy.  This was
  915 	  needed at least in my cygwin 1.3.20/w2k setup.
  916 
  917 1.42	[2003-01-07]
  918 	- modernize the constants code (from Nicholas Clark)
  919 
  920 1.41	[2003-01-03]
  921 	- At some point the ability to figure our the correct incdir
  922 	  for EXTERN.h (either a core perl build, or an installed perl)
  923 	  had broken (which lead into all test compiles failing with
  924 	  a core perl build, but thanks to the robustness of Makefile.PL
  925 	  nothing of this was visible).  The brokenness seemed to be
  926 	  caused by $ENV{PERL_CORE} not being on for core builds?
  927 	  Now stole a trick from the Encode that sets $ENV{PERL_CORE}
  928 	  right, and both styles of build should work again.
  929 
  930 1.40	[2003-01-03]
  931 	- Nicholas Clark noticed that the my_catdir() emulation function
  932 	  was broken (which means that we didn't really work for Perls
  933 	  5.002 and 5.003)
  934 	- inspired by fixing the above made the whole Makefile.PL -w
  935 	  and strict clean
  936 	- tightened up the Makefile.PL output, less whitespace
  937 
  938 1.39	[2003-10-20]
  939 	- fix from Craig Berry for better building in VMS with PERL_CORE
  940 
  941 1.38	[2003-10-13]
  942 	- no functional changes
  943 	- move lib/Time/HiRes.pm as Hires.pm
  944 	- libraries scanning was slightly broken (always scanned
  945 	  for a library even when $Config{libs} already had it)
  946 
  947 1.37	[2003-09-23]
  948 	- Ray Zimmerman ran into a race condition in Mac OS X.
  949 	  A 0.01-second alarm fired before the test expected.
  950 	  The test first slept indefinitely (blocking for signals)
  951 	  and only after that tested for the signal having been sent.
  952 	  Since the signal had already been sent, the test #12 never
  953 	  completed.  The solution: test first, then block.
  954 	- default to being silent on all probing attempts, set the
  955 	  environment variable VERBOSE to a true value to see the
  956 	  details (the probing command and the possible errors)
  957 
  958 1.36	[2003-09-12]
  959 	- do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
  960 	- INSTALLDIRS => 'perl' missing which means that Time::HiRes
  961 	  cannot be upgraded from CPAN to override the 5.8.0 version
  962 	  (Guido A. Ostkamp)
  963 	- Time::HiRes 1.35 could not be dropped as-is to bleadperl
  964 	  because the include directories did not adjust themselves
  965 	  if $ENV{PERL_CORE} (Hugo van der Sanden)
  966 	- add documentation about the restart of select() under alarm()
  967 
  968 1.35	[2003-08-24]
  969 	- small documentation tweaks
  970 
  971 
  972 1.34	[2003-08-22]
  973 	- better VMS operation (Craig Berry)
  974 
  975 1.33	[2003-08-20]
  976 	- our time machine is accelerating: now works with Perl 5.004_01
  977 	  (tried with 5.003_07 and 5.002 but I get segmentation faults
  978 	   from running the Makefile.PL with those in Tru64 4.0D)
  979 
  980 1.32	[2003-08-20]
  981 	- backward compatibility (pre-5.6.0) tweaks:
  982 	  - no XSLoader in 5.00503, use DynaLoader instead
  983 	  - no SvPV_nolen, either
  984 	  - no PerlProc_pause(), either
  985 	  - now tested with 5.00404 and 5.00503
  986 	  - Makefile.PL requires 5.00404 (no more 5.002)
  987 	- use nanosleep instead of usleep, if it is available (Wilson Snyder)
  988 	  (this means that one can mix subsecond sleeps with alarms)
  989 	- because of nanosleep we probe for -lrt and -lposix4
  990 	- the existence of getitimer/nanosleep/setitimer/ualarm/usleep
  991 	  is available by exportable constants Time::HiRes::d_func
  992 	  (since older Perl do not have them in %Config, and even
  993 	   5.8.0 does not probe for nanosleep)
  994 
  995 1.31	[2003-08-19]
  996 	- backward compatibility (pre-5.6.1) tweaks:
  997 	  - define NV if no NVTYPE
  998 	  - define IVdf if needed (note: the Devel::PPPort
  999 	    in 5.8.0 does not try hard enough since
 1000 	    the IVSIZE might not be defined)
 1001 	  - define NVgf if needed
 1002 	  - grab the typemap from 5.8.0 for the NV stuff
 1003 
 1004 	1.31 and 1.32 add more backward compatibility (now all the way
 1005 	back to Perl 5.00404), and using nanosleep() (if available) for
 1006 	subsecond sleeps.
 1007 
 1008 1.30	[2003-08-16]
 1009 
 1010 	- release 1.29_02 as 1.30
 1011 
 1012 	1.30 adds all the changes made during the Perl 5.6->5.7->5.8
 1013 	development cycle.  Most notably portability across platforms has been
 1014 	enhanced, and the interval timers (setitimer, getitimer) have been
 1015 	added.  Note that the version of Time::HiRes that is included in Perl
 1016 	5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
 1017 	version.  Note also that in 1.30 Wegscheid turns over the maintenance
 1018 	to Jarkko Hietaniemi.
 1019 
 1020 1.29_02	[2003-08-16]
 1021 
 1022 	- fix a silly unclosed comment typo in HiRes.xs
 1023 	- document and export REALTIME_REALPROF (Solaris)
 1024 
 1025 1.29_01	[2003-08-16]
 1026 
 1027 	- only getitimer(ITIMER_REAL) available in Cygwin and Win32
 1028 	  (need to patch this also in Perl 5.[89])
 1029 	- remove CVS revision log from HiRes.xs
 1030 
 1031 1.29_00	[2003-08-14]
 1032 
 1033 	The following numbered patches refer to the Perl 5.7 changes,
 1034 	you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
 1035 
 1036 	- 17558: Add #!./perl to the .t
 1037 	- 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
 1038 	- 16198: political correctness, from Simon Cozens
 1039 	- 15857: doc tweaks, from Jarkko Hietaniemi
 1040 	- 15593: optimization in .xs, from Paul Green
 1041 	- 14892: pod fixes, from Robin Barker
 1042 	- 14100: VOS fixes, from Paul Green
 1043 	- 13422: XS segfault, from Marc Lehmann
 1044 	- 13378: whether select() gets restarted on signals, depends
 1045 	- 13354: timing constraints, again, from Andy Dougherty
 1046 	- 13278: can't do subsecond alarms with ualarm;
 1047 		 break out early if alarms do not seem to be working
 1048 	- 13266: test relaxation (cygwin gets lower hires
 1049 		 times than lores ones)
 1050 	- 12846: protect against high load, from Jarkko Hietaniemi
 1051 	- 12837: HiRes.t VMS tweak, from Craig A. Berry
 1052 	- 12797: HiRes.t VMS tweak, from Charles Lane
 1053 	- 12769: HiRes.t VMS tweak, from Craig A. Berry
 1054 	- 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
 1055 	- 12722: VMS ualarm for VMS without ualarm, from Charles Lane
 1056 	- 12692: alarm() ain't gonna work if ualarm() ain't,
 1057 		 from Gurusamy Sarathy
 1058 	- 12680: minor VMS tweak, from Charles Lane
 1059 	- 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
 1060 	- 12609: croak on negative time, from Jarkko Hietaniemi
 1061 	- 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
 1062 	- 12594: MacOS Classic timeofday, from Chris Nandor
 1063 	- 12473: allow for more than one second for sleep() and usleep()
 1064 	- 12458: test tuning, relax timing constraints,
 1065 		 from Jarkko Hietaniemi
 1066 	- 12449: make sleep() and usleep() to return the number
 1067 		 of seconds and microseconds actually slept (analogously
 1068 		 with the builtin sleep()), also make usleep() croak if
 1069 		 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
 1070 	- 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
 1071 	- 12199: do not use ftime on Win32, from Gurusamy Sarathy
 1072 	- 12196: use ftime() on Win32, from Artur Bergman
 1073 	- 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
 1074 	- 12105: use GetSystemTime() on Win32, from Artur Bergman
 1075 	- 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
 1076 	- 11901: UNICOS sloppy division, from Jarkko Hietaniemi
 1077 	- 11797: problem in HiRes.t, from John P. Linderman
 1078 	- 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
 1079 	- 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
 1080 	- 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
 1081 	- 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
 1082 	- 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
 1083 	- 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
 1084 		 from Jonathan Stowe
 1085 	- 10942: MPE/IX test tweaks, from Mark Bixby
 1086 	- 10784: unnecessary pod2man calls, from Andy Dougherty
 1087 	- 10354: ext/ + -Wall, from Doug MacEachern
 1088 	- 10320: fix the BOOT section to call myU2time correctly
 1089 	- 10317: correct casting for AIX< from H. Merijn Brand
 1090 	- 10119: document that the core time() may be rounding, not truncating
 1091 	- 10118: test fix, from John Peacock
 1092 	-  9988: long =item, from Robin Barker
 1093 	-  9714: correct test output
 1094 	-  9708: test also the scalar aspect of getitimer()
 1095 	-  9705: Add interval timers (setitimer, getitimer)
 1096 	-  9692: do not require at least 5.005 using XS
 1097 		
 1098 	The following changes were made on top of the changes
 1099 	made for Time::HiRes during the Perl 5.7 development
 1100 	cycle that culminated in the release of Perl 5.8.0.
 1101 
 1102 	- add "require 5.005" to the Makefile.PL
 1103 	- remove the REVISION section (CVS log) from HiRes.pm
 1104 	- add jhi's copyright alongside Douglas'
 1105 	- move HiRes.pm to lib/Time/
 1106 	- move HiRes.t to t/
 1107 	- modify HiRes.t to use $ENV{PERL_CORE}
 1108 	- modify the original Time::HiRes version 1.20 Makefile.PL
 1109 	  to work both with Perl 5.8.0 and the new code with pre-5.8.0
 1110 	  Perls (tried with 5.6.1)
 1111 	- tiny tweaks and updates in README and TODO
 1112 	- bump the VERSION to 1.29
 1113 
 1114 1.20  Wed Feb 24 21:30 1999
 1115 	- make our usleep and ualarm substitutes into hrt_usleep
 1116 	  and hrt_ualarm. This helps static links of Perl with other
 1117 	  packages that also have usleep, etc. From
 1118 	  Ilya Zakharevich <ilya@math.ohio-state.edu>
 1119 	- add C API stuff. From Joshua Pritikin
 1120 	  <joshua.pritikin@db.com>
 1121 	- VMS Makefile.PL fun.	From pvhp@forte.com (Peter Prymmer)
 1122 	- hopefully correct "-lc" fix for SCO.
 1123 	- add PPD stuff
 1124 
 1125 	1.20 adds a platform neutral set of C accessible routines if you are
 1126 	running 5.005+.  All other changes are packaging changes and build
 1127 	fixes(?) for statically linked Perl, SCO, and VMS.
 1128 
 1129 1.19  Tue Sep 29 22:30 1998
 1130 	- put VMS gettimeofday() in. Patch is from Sebastian Bazley
 1131 	  <seb@stian.demon.co.uk>
 1132 	- change GIMME_V to GIMME to help people with older versions of
 1133 	  Perl.
 1134 	- fix Win32 version of gettimeofday(). It didn't affect anything,
 1135 	  but it confuses people reading the code when the return value
 1136 	  is backwards (0 is success).
 1137 	- fix Makefile.PL (more) so that detection of gettimeofday is
 1138 	  more correct.
 1139 
 1140 	1.19 has better VMS support.
 1141 
 1142 1.18  Mon Jul 6 22:40 1998
 1143 	- add usleep() for Win32.
 1144 	- fix Makefile.PL to fix reported HP/UX feature where unresolved
 1145 	  externals still cause an executable to be generated (though no
 1146 	  x bit set). Thanks to David Kozinn for report and explanation.
 1147 	  Problems with the fix are mine :)
 1148 
 1149 	1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
 1150 	Probably buggy. I'm sure I'll hear.
 1151 
 1152 1.17  Wed Jul 1 20:10 1998
 1153 	- fix setitimer calls so microseconds is not more than 1000000.
 1154 	  Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
 1155 	- make Win32. We only get gettimeofday (the select hack doesn't
 1156 	  seem to work on my Win95 system).
 1157 	- fix test 4 on 01test.t. add test to see if time() and
 1158 	  Time::HiRes::time() are close.
 1159 
 1160 1.16  Wed Nov 12 21:05 1997
 1161 	- add missing EXTEND in new gettimeofday scalar code.
 1162 
 1163 	1.16+ should be closer to building out of the box on Linux. Thanks
 1164 	to Gisle Aas for patches, and the ualarm equivalent using setitimer.
 1165 
 1166 	If your underlying operating system doesn't implement ualarm(), then
 1167 	a fake using setitimer() will be made.  If the OS is missing usleep(),
 1168 	a fake one using select() will be made. If a fake can't be made for
 1169 	either ualarm() or usleep(), then the corresponding Perl function will
 1170 	not be available.  If the OS is missing gettimeofday(), you will get
 1171 	unresolved externals, either at link- or run-time.
 1172 
 1173 	This is an improvement; the package used to not even build if
 1174 	you were missing any of these bits. Roderick Schertler
 1175 
 1176 	<roderick@argon.org> did all the conditional compilation stuff,
 1177 	look at HiRes.pm and the test suites; it's good educational reading.
 1178 
 1179 1.15  Mon Nov 10 21:30 1997
 1180 	- HiRes.pm: update pod. Provided by Gisle Aas.
 1181 	- HiRes.xs: if gettimeofday() called in scalar context, do
 1182 	  something more useful than before. Provided by Gisle Aas.
 1183 	- README: tell of xsubpp '-nolinenumber' woes. thanks to
 1184 	  Edward Henigin <ed@texas.net> for pointing out the problem.
 1185 
 1186 1.14  Wed Nov 5 9:40 1997
 1187 	- Makefile.PL: look for setitimer
 1188 	- HiRes.xs: if missing ualarm, but we have setitimer, make up
 1189 	  our own setitimer. These were provided by Gisle Aas.
 1190 
 1191 1.13  Tue Nov 4 23:30 1997
 1192 	- Makefile.PL: fix autodetect mechanism to do try linking in addition
 1193 	  to just compiling; should fix Linux build problem. Fix was provided
 1194 	  by Gisle Aas.
 1195 
 1196 1.12  Sun Oct 12 12:00:00 1997
 1197 	- Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
 1198 	  you may need to comment this back out if you have an older xsubpp.
 1199 	- HiRes.xs: set PROTOTYPES: DISABLE
 1200 
 1201 1.11  Fri Sep 05 16:00:00 1997
 1202 	- Makefile.PL:
 1203 	  Had some line commented out that shouldn't have been (testing
 1204 	  remnants)
 1205 	- README:
 1206 	  Previous version was corrupted.
 1207 
 1208 1.10  Thu May 22 20:20:00 1997
 1209 	- HiRes.xs, HiRes.pm, t/*:
 1210 	      -	only compile what we have OS support for (or can
 1211 		fake with select())
 1212 	      - only test what we compiled
 1213 	      - gross improvement to the test suite
 1214 	      - fix EXPORT_FAIL.
 1215 	  This work was all done by Roderick Schertler
 1216 	  <roderick@argon.org>. If you run Linux or
 1217 	  one of the other ualarm-less platforms, and you like this
 1218 	  module, let Roderick know; without him, it still wouldn't
 1219 	  be working on those boxes...
 1220 	- Makefile.PL: figure out what routines the OS has and
 1221 	  only build what we need. These bits were written by Jarkko
 1222 	  Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
 1223 
 1224 1.02  Mon Dec 30 08:00:00 1996
 1225 	- HiRes.pm: update documentation to say what to do when missing
 1226 	  ualarm() and friends.
 1227 	- README: update to warn that ualarm() and friends need to exist
 1228 
 1229 1.01  Fri Oct 17 08:00:00 1996
 1230 	- Makefile.PL: make XSPROTOARGS => '-noprototyopes'
 1231 	- HiRes.pm: put blank line between __END__ and =head1 so that
 1232 	  pod2man works.
 1233 
 1234 1.00  Tue Sep 03 13:00:00 1996
 1235 	- original version; created by h2xs 1.16