xref: /openbsd/gnu/usr.bin/perl/dist/Time-HiRes/Changes (revision 3d61058a)
1Revision history for the Perl extension Time::HiRes.
2
3{{NEXT}}
4
5 - Remove obsolete vms code
6 - Use core version compare
7 - Use GIMME_V instead of the deprecated GIMME
8 - t/utime.t: dragonflybsd has only microsecond precision
9 - t/utime.t: dragonflybsd is noatime by default
10 - t/stat.t: skip testing access times on HaikuOS, it doesn't support
11   atime
12 - darwin: make sure the compiler can find the system perl headers
13   https://github.com/Perl/perl5/issues/20362
14 - darwin: make sure PERL_DARWIN is defined on darwin.
15 - don't compare stat and lstat atime if PERL_FILE_ATIME_CHANGES is set in
16   the environment.
17   https://github.com/Perl/perl5/issues/19321
18 - don't use C++ guards around the perl header files, it caused C++
19   build failures with MSVC.
20 - don't try to suppress C++ compatibility warnings in C++ builds, since
21   that warns.
22
231.9764 [2020-08-10]
24 - Fix a bunch of repeated-word typos
25 - Fix compilation with Visual C++ 2013 and older
26 - Implement clock_gettime() and clock_getres() for win32
27 - Remove workaround for an ancient Devel::PPPort bug
28 - Remove the remains of MacOS classic support
29 - Remove code for perls older than 5.6
30 - Remove vestiges of mpeix support (removed in 5.17.x)
31 - Remove trailing blanks
32 - Add links to pod
33
341.9760 [2019-02-18]
35 - make utime() available only if we have both fd and name setting
36   [RT #133030]
37 - Adjutst Makefile.PL for windows: the DEFINE() in Makefile.PL
38   can't be in sub init() because that sub isn't called on windows
39 - t/itimer.t: avoid race condition.
40 - don't truncate nanosec utime
41 - fallback/const-c.inc: Avoid compiler warning showing up on darwin.
42
431.9759 [2018-03-23]
44 - This version reverts the changes on CPAN that were included in 1.9754 - 1.9758.
45   Several of these changes caused instability on some windows platforms. We can
46   revisit these fixes if necessary in Perl 5.29.0
47
481.9758 [2018-03-21]
49 - fix build in Win32 with Visual C by introducing a fake struct timezone
50   [rt.cpan.org #124844]
51 - in utime.t detect better being run in a noatime filesystem,
52   and if so, skip the test (for the HAMMER filesystem of DragonflyBSD)
53 - also for the HAMMER fs (if not in noatime), lower the expected subsecond
54   timestamp granularity to microseconds
55 - fix the version number typo in Changes: 1.9577 -> 1.9757
56
571.9757 [2018-03-15]
58 - fix win32 (and cygwin?) builds which have been broken since 1.9755:
59   problem was that gettimeofday() is a macro and needs to stay such
60 - regenerate ppport.h with Perl 5.26.1 (was previously generated with
61   5.25.6) and Devel::PPPort 3.35 (previously 3.36) -- this doesn't
62   change the ppport.h, though
63
641.9756 [2018-03-14]
65 - division by zero in new test tv_interval.t [rt.cpan.org #124775]:
66   made the test pass even if the difference is zero, but also
67   made it practically impossible to ever be zero by adding
68   a sub-second sleep.
69 - remove comments from inside qw [rt.cpan.org #124777],
70   only seemed to be caught by Perl 5.18.4
71
721.9755 [2018-03-14]
73 - adjust the error messages of tv_interval()
74 - the NV_1EX constants now cast to be NV so that
75   tv_interval() with long double builds does not
76   produce ugly results [rt.cpan.org #106456]
77 - add tests for tv_interval()
78 - centralize the mygettimeofday() logic
79 - make the mygettimeofday() a function instead of macro
80 - add the OS X 10.12+ clock_gettime() constants
81   CLOCK_MONOTONIC_RAW_APPROX CLOCK_UPTIME_RAW_APPROX
82 - regenerate the fallback files
83
841.9754 [2018-02-16]
85 - unify the gettimeofday() and time() forked code
86   that had near identical code paths for MacOS Classic
87   (which has two unusual features: unsigned time_t,
88   and a special tz struct member), and for non-MacOS Classic
89 - tv_interval should be implemented in XS
90   [rt.cpan.org #106456]
91   thanks to Sergey Aleynikov (suggested implementation simplified
92   by the above-mentioned MacOS Classic simplification)
93
941.9753 [2018-01-11]
95 - in t/clock.t in a fast system we need to burn more CPU,
96   reported and fix suggested by Joel C. Maslak <jmaslak@antelope.net>
97   [rt.cpan.org #124025]
98 - for t/utime.t in netbsd we need /sbin in PATH to find the mount command,
99   this is needed for the fix in 1.9751 to actually help netbsd with noatime
100   mounts.
101   (thanks to Nigel Horne)
102
1031.9752 [2018-01-04]
104 - fix an error in the error message of utimensat() not available:
105   it said futimens() not available
106 - add --force alias for Makefile.PL --configure
107
1081.9751 [2018-01-02]
109 - in macOS/OSX/Darwin, use __has_builtin() check also for utimensat(),
110   can cause errors like
111   "HiRes.xs:1474:16: error: unrecognized platform name macOS"
112   [rt.cpan.org #123994]
113   (oversight from 1.9749)
114 - do not define TIME_HIRES_STAT/d_hires_stat if none was found, instead
115   of defining it to be zero, which case has no implementation in hrstatns()
116   (thanks to Nigel Horne)
117 - in t/utime.t try to divine if the filesystem of the tempfiles has been
118   mounted with the 'noatime' option, which can prohibit updating the
119   access time timestamp.  Also document this in HiRes.pm.
120   (thanks to Nigel Horne, original analysis by Slaven Rezic)
121 - synchronize the constant lists in HiRes.pm:@EXPORT_OK
122   and Makefile.PL:doConstants and regenerate fallback/const-c.inc
123   and fallback/const-xs.inc, this fixes Perl 5.6.2 issue with
124   d_futimens not allegedly being a valid macro in t/utime.t
125   (using Perl 5.26.1 for the regenerating, not 5.6.2)
126   (thanks to Nigel Horne)
127 - in t/utime.t define a nop sub done_testing for ancient Perls
128   (like Perl 5.6.2)
129 - in Perl 5.6.2 a bogus warning
130   "Use of uninitialized value in subroutine entry"
131   is issued from t/alarm.t: add a comment documenting that
132
1331.9750 [2017-12-22]
134 - update META.yml and META.json for XSLoader instead of DynaLoader
135   [rt.cpan.org #123933]
136
1371.9749 [2017-12-21]
138 - CPAN release
139 - Quieten Time-HiRes test noise: blead 0f009486
140 - Switch some core modules to XSLoader: blead b9a5a78f [rt.cpan.org #129789]
141 - disable a flapping test for CI: blead cd2a978d
142 - Avoid newGVgen in blead-upstream modules: blead 732d3893
143 - Declaration after statement in typemap: blead d269f586
144 - semicolon-friendly diagnostic control: blead 7347ee54
145 - in macOS (OS X) check properly in runtime whether futimens()
146   and utimensat() are available
147
1481.9748 [2017-12-16]
149 - bleadperl only
150 - semicolon-friendly diagnostic control: blead 7347ee54
151
1521.9747 [2017-09-13]
153 - bleadperl only
154 - Switch some core modules to XSLoader: blead b9a5a78f [rt.cpan.org #129789]
155
1561.9746 [2017-08-17]
157 - Unreliable t/usleep.t and t/utime.t tests [rt.cpan.org #122819]
158   Avoid testing for $dt = $t2 - $t1 and assuming $dt is less than
159   some value since a heavily loaded machine can delay the $t2.
160   Testing for greater than is fine.
161 - Tweak the configuring messages about subsecond stat/utime.
162
1631.9745 [2017-08-14]
164  - Properly check for futimens/utimensat actually doing something.
165    This should handle 'gnukfreebsd' (which has only stubs, so far
166    we have been skipping the test) and as a new thing 'gnu' (Hurd)
167    (also only stubs).  Thanks to Nigel Horne.
168  - Scan in t/utime.t whether the filesystem (wherever File::Temp
169    puts it tempfiles) supports subsecond timestamps.  This removes
170    the fragile Linux/ext3 specific hack.  As a side effect, the
171    setting of subsecond timestamps is tested only if reading of
172    them is supported.  Thanks to Carsten Gaebler for the test idea,
173    and Ryan Voots for testing.
174
1751.9744 [2017-07-27]
176  - add more potential clock constants, like CLOCK_MONOTONIC_FAST
177    (available in FreeBSD), and not all potentially found clock
178    constants were properly exported to be available from Perl,
179    see your system's clock_gettime() documentation for the available ones
180
1811.9743 [2017-07-20]
182  - correct declared minimum Perl version (should be 5.6, was declared
183    as 5.8 since 1.9727_03): blead af94b3ac
184  - fix the fix for 'do file' to load hints in Makefile.PL: blead 3172fdbc
185
1861.9742 [2017-04-16]
187  - prefer 3-argument open: blead 1ae6ead9
188  - fix dist/Time-HiRes/t/*.t that assumed '.' in @INC: blead 465db51d
189  - fix cases where 'do file' should be 'do ./file'.: blead 8b69401c
190
1911.9741 [2016-11-20]
192  - C++11 compatibility: blead a914236c
193  - El Capitan compatibility: blead 45bbc013
194  - use CLOCK_REALTIME for clock_nanosleep scan: blead 56ed632c
195  - include file consistency in scans: blead 605b4dbe
196  - use clockid_t consistently: blead 2979fdf3
197  - use hv_fetchs(): blead 07e4dd7a
198  - scan for clockid_t (needed for macos Sierra): blead 69d1f2c2
199  - darwin lacks clockid_t: blead d1e7b5aa
200    [rt.cpan.org #129789]
201  - provide template for tempfile: blead 91dca83e
202  - explicit cast for clockid_t in C++11: blead 1c566547
203  - upgrade to a newer ppport (3.35)
204
2051.9740 [2016-09-25]
206  - the ext3/ext2 filesystems do not have subsecond resolution,
207    therefore skip the t/utime.t test
208    [rt.cpan.org #116127]
209
2101.9739 [2016-06-28]
211  - the upcoming macOS 10.12 (Sierra, the operating system formerly
212    known as OS X, or Darwin) has implemented the clock_gettime()
213    and clock_getres() interfaces but not the clock_nanosleep()
214
2151.9738 [2016-06-26]
216  - fix the dates in the Changes, it's not July yet
217    [rt.cpan.org #115553]
218  - in cygwin, drop the utime.t expected timestamp resolution
219  - documented for utime that the actual timestamp resolution depends
220    on the operating system and the filesystem, just like for stat
221  - looks like in gnukfreebsd 10.1 neither futimens or utimensat work
222    even though they exist, skipping utime tests there for now
223  - demodernize the Test::More use in utime.t (no subtest) to
224    better support older Perl installations
225  - for all the tests, drop the Test::More 0.82 requirement,
226    the only "modern" thing used is the note routine which
227    can be replaced with print("# ...\n") (or printf),
228    with this works back at least with Test::More 0.54 (perl 5.8.7)
229
2301.9737 [2016-06-22]
231  - utime in HiRes.xs had dXSARGS which was unnecessary, leading
232    to two markstack pops, panic, and two-minute wait for the watchdog
233    (thanks to Dave Mitchell for the eagle-eye)
234  - even though 5.10.x should have the Perl_ck_warner, many CPAN
235    smokes proved otherwise, so requiring now at least 5.12 for
236    Perl_ck_warner use in IS_SAFE_PATHNAME()
237
2381.9736 [2016-06-22]
239  - the IS_SAFE_PATHNAME() emulation was too modern for Perl 5.10
240  - make it so that only one value is set for -DTIME_HIRES_STAT
241    even on systems that support many options
242  - the 1.9734 (only released in blead, and as part of the development
243    release 5.25.2) fix for Darwin's clock_gettime was wrong under
244    threaded builds, leading into mutex panics in clock.t
245  - the d_hires_utime determination was made more precise
246  - the utime.t was failing in platforms where only one of
247    futimens/utimensat was present (neither or both was okay)
248
2491.9735 [2016-06-20]
250  - Time::HiRes should override `utime` to allow setting hires
251    (futimens and utimensat support)
252    [rt.perl.org #114809]
253  - the utime patch used IS_SAFE_PATHNAME() which isn't available in
254    too old Perls, so emulate (in case the Devel::PPPort is too old)
255
2561.9734 [2016-06-17]
257  - fix Darwins with clock_gettime: blead 2d41a263
258    [rt.perl.org #128427]
259
2601.9733 [2016-04-23]
261  - C90 declaration-after-statement error with darwin threads: blead de1003b4
262    [rt.cpan.org #113856]
263
2641.9732 [2016-03-13]
265  - MUTEX_LOCK, not PERL_MUTEX_LOCK: blead e5b02b5d
266  - also hrt_ualarm_itimer() is unused: 1cb6cce3
267  - the mutex needs init: 2d639e20
268  - version bump to 1.9732: 730d7fdc
269
2701.9731 [2016-03-13]
271  - mark unused variable as such: blead a914236c
272  - OS X emulation mutex accidentally unused: da7a6455
273  - remove hrt_ualarm: 6da77c36
274  - pod error fixes: 919ca095
275  - nanosleep and clock_nanosleep cleanups: c8ea02b8..e3ff671b
276  - static funcs in HiRes.xs: 4e922c26
277  - Remove unused variable: 52ffb1b5
278
2791.9730 [2016-02-17]
280  - TIME_HIRES_DONT_RUN_PROBES=1 to build the probes but not run them
281    [rt.cpan.org #111391]
282
2831.9729 [2016-02-13]
284  - new constants
285    [rt.cpan.org #111170]
286  - clock_getres(), clock_gettime(), clock_nanosleep() emulation for OS X
287    [rt.cpan.org #78566]
288
2891.9728 [2015-11-13]
290  - No changes since 1.9727_03
291
2921.9727_03 [2015-11-06]
293  - use PERL_UNUSED_ARG() in the not-here-croak branches
294  - Makefile.PL/metadata improvements
295
2961.9727_02 [2015-08-29]
297  - correct documentation to note that Time::NVtime is NV, not double
298    [perl #123879]
299  - avoid a warning about clock_t on OS X / clang 3.6 - 3.6
300    [rt.cpan.org #102718]
301
3021.9727_01 [2015-08-17]
303  - Remove usage of 5.005 thread APIs in favor of 5.6 and up thread APIs
304    [rt.cpan.org #102799].
305  - Silence a C++ compiler warning [rt.cpan.org #101497].
306  - respect $Config{run} for cross-compiling [rt.cpan.org #92607]
307
3081.9726 [2013-08-17]
309	- Correct s/us splitting of usleep(1000000) [rt.cpan.org #78266].
310	- Avoid integer overflow in itimer-based alarm() with large
311	  argument [rt.cpan.org #87160].
312	- Define PERL_NO_INLINE_FUNCTIONS during probe compilations, to
313	  avoid false negatives caused by not linking with the perl core.
314	- Be more careful about context in stat().
315	- Install into "site" directories by default on perl 5.11+
316	  [rt.cpan.org #79797].
317	- Fix a couple of doc typos [rt.cpan.org #85365].
318	- Fix function name in a doc example [rt.cpan.org #86318].
319	- Provide lstat() that yields high-res timestamps, alongside
320	  the existing high-res stat() [rt.cpan.org #78732].
321
3221.9725 [2012-02-01]
323	- Correct stack discipline in stat(), which was screwing up list
324	  operations in expressions containing calls to it [rt.cpan.org
325	  #72926].
326	- Add missing OUTPUT sections to the XS code [rt.cpan.org #70930].
327	- Skip itimer tests on GNU/Hurd, which has the API but lacks
328	  the implementation [rt.cpan.org #72754].
329	- Fix a doubled word in the documentation [rt.cpan.org #72763].
330
3311.9724 [2011-06-09]
332	- Correct XS parameter list, and therefore prototype, for
333	  unimplemented-on-this-platform version of clock_nanosleep()
334	  [rt.cpan.org #68700].
335	- Declare package variables with "our" rather than "use vars".
336	- Corresponding to "our" usage, check for minimum Perl version
337	  5.006.
338	- Declare module dependencies.
339
3401.9723 [2011-06-07]
341	- Remove $ENV{PERL_CORE} logic from test suite, which is no
342	  longer desired in the core.
343	- Convert test suite to use Test::More.
344	- Factor out watchdog code from test suite.
345	- In test suite, be consistent about using fully-qualified form
346	  of function names.
347	- Divide test suite into feature-specific scripts.
348	- Make ualarm timing test less vulnerable to delay-induced false
349	  failure, from Dave Mitchell.
350
3511.9722 [2011-05-18]
352	- Update for changes in build process in the core, patches
353	  from BinGOs [rt.cpan.org #58858] and Craig Berry [rt.cpan.org
354	  #63363].
355	- Fix broken linkage on Windows with gcc 3.4 seen with ActivePerl,
356	  report from Christian Walde [rt.cpan.org #61648], fix derived
357	  from Vincent Pit.
358	- Jump through hoops to avoid compiler warnings.
359
3601.9721 [2010-03-17]
361	- Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
362	  additional fix from Gisle Aas
363	- Address [rt.cpan.org #55665] "Bad plan" on Windows,
364	  report and fix from Jan Dubois
365
3661.9720 [2010-02-14]
367	- Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
368	  report and fix from Nicholas Clark
369
3701.9719	[2009-01-04]
371	- As with QNX, Haiku has the API of interval timers but not
372	  the implementation (bleadperl change #34630), hence skip
373	  the tests, via David Mitchell.
374
3751.9718	[2008-12-31]
376	- .xs code cleanup from Albert Dvornik
377	- in the #39 and #40 do not do us I did, mixing alarm() and
378	  sleep().  Now instead spin until enough time has passed.
379
3801.9717	[2008-12-30]
381	- Skip the tests added in 1.9716 (#39, #40) if there's no subsecond
382	  alarm capability, like with the older subsecond alarm tests
383
3841.9716	[2008-12-26]
385	- Change documentation to agree with reality: there are
386	  no interval timers in Win32.
387	- Address [rt.cpan.org #35899] (problem in subsecond sleeps),
388          add two tests to guard against this problem
389	- Address [rt.cpan.org #36600] 'Division by zero' failure in test suite
390	- Address [rt.cpan.org #37340] [PATCH] Address timer process in test
391	- Address [rt.cpan.org#40311 ] bad implementation of hrt_usleep
392          with TIME_HIRES_NANOSLEEP
393
3941.9715	[2008-04-08]
395	- Silly me: Makefile.PL does need to accept arguments other than mine.
396	  Some testing frameworks obviously do this.
397	- Add retrying for tests 34..37, which are the most commonly
398	  failing tests.  If this helps, consider extending the retry
399	  framework to all the tests.  [Inspired by Slaven Rezic,
400	  [rt.cpan.org #34711] Occasional failures of test 35 or 36 (FreeBSD)]
401
4021.9714	[2008-04-07]
403	- Under Perl 5.6.* NVgf needs to be "g", reported by Zefram,
404	  it seems that ppport.h 3.13 gets this wrong.
405	- remove the check in Makefile.PL for 5.7.2, shouldn't be
406	  (a) necessary (b) relevant
407	- add logic to Makefile.PL to skip configure/write Makefile
408	  step if the "xdefine" file already exists, indicating that
409	  the configure step has already been done, one can still
410	  force (re)configure by "perl Makefile.PL configure",
411	  or of course by "make clean && perl Makefile.PL".
412
4131.9713	[2008-04-04]
414	- for alarm() and ualarm() [Perl] prefer setitimer() [C]
415	  instead of ualarm() [C] since ualarm() [C] cannot portably
416	  (and standards-compliantly) be used for more than 999_999
417	  microseconds (rt.cpan.org #34655)
418	- it seems that HP-UX has started (at least in 11.31 ia64)
419	  #defining the CLOCK_REALTIME et alia (instead of having
420	  them just as enums)
421	- document all the diagnostics
422
4231.9712	[2008-02-09]
424	- move the sub tick in the test file back to where it used to be
425	- in the "consider upgrading" message recommend at least Perl 5.8.8
426	  and make the message to appear only for 5.8.0 since 5.8.1 and
427	  later have the problem fixed
428	- VOS tweak for Makefile (core perl change #33259)
429	- since the test #17 seems to fail often, relax its limits a bit
430
4311.9711	[2007-11-29]
432	- lost VMS test skippage from Craig Berry
433	- reformat the test code a little
434
4351.9710	[2007-11-29]
436	- I got the sense of the QNX test the wrong way in an attempt
437	  to generalize it for future
438
4391.9709	[2007-11-28]
440	- casting fixes from Robin Barker for g++ and 64bitint
441	- in QNX skip the itimer tests because though the API
442	  is there, the implementation isn't, from Matt Kraai
443	- raise the dead man timer to 180 seconds for really
444	  slow/busy systems
445	- elaborate the UTF-8 locale warning from Makefile.PL
446
4471.9708	[2007-10-05]
448	- [rt.cpan.org #29747]: Build failure with perl 5.005_05
449	  Fixed by regenerating the ppport.h using Devel::PPPort 3.13.
450
4511.9707	[2007-02-27]
452	- t/HiRes.t failed in Perl 5.6.2,
453	  "action is not of type POSIX::SigAction at t/HiRes.t line 318",
454	  reported and fixed by Anton Berezin, the reason was faulty
455	  use of sigaction() when restoring the old action.
456
4571.9706	[2007-02-25]
458	- with bleadperl in VMS the HiRes.t overrun the maximum number
459	  of deferred signals because the libc SIGALRM was not strong
460	  enough to interrupt select(), and select() got restarted every
461	  time, solution is to use POSIX::SigAction if available.
462	  A fix from Craig Berry (not 100% there, but helps).
463	- allow for more measuring noise for ualarm() tests 35..37
464
4651.9705	[2007-02-06]
466	- nanosleep() and clock_nanosleep() detection and use were
467	  quite broken; in Linux -lrt needed; fixes from Zefram
468	- [internal] slightly cleaner building of $DEFINE in Makefile.PL,
469	  should avoid double/conflicting -D flags
470
4711.9704	[2007-01-01]
472	- allow 10% of slop in test #14 (testing difference between
473	  CORE::time() and Time::HiRes::time()), there seem to be often
474	  transient failures from Perl smoke builds on this test
475	- small pod tweaks
476
4771.9703	[2006-12-08]
478	- use int main(int argc, char **argv) consistently in Makefile.PL,
479	  should help with
480	  [rt.cpan.org #23868] nanosleep not detected under Mac OS 10.3.9 starting with Time::HiRes 1.96
481	- if someone still has the locale-broken Perl 5.8.0,
482	  suggest that they upgrade their Perl
483
4841.9702	[2006-12-06]
485	- restore the -DATLEASTFIVEOHOHFIVE, Win32 needed it still
486
4871.9701	[2006-12-04]
488	- upgrade to ppport.h 3.10_02
489	- remove the -DATLEASTFIVEOHOHFIVE
490	- use the ppport.h PL_ppaddr, PL_statcache, PL_laststatval
491	- use the ppport.h aTHXR for calling Perl stat()
492	- switch into four-digit version since 2.0 is coming up
493	  awfully fast but not feeling like a major rewrite
494
4951.97	[2006-11-30]
496	- 1.95 broke building in Win32 (since pp_stat is not exported),
497	  figured out how to call an op directly in 5.005 (use Perl_ppaddr
498	  instead of PL_ppaddr)
499	- backport to Perl 5.004_05  (requires using statcache
500	  and laststatval instead of PL_statcache and PL_laststatval)
501	  (also checked to work in 5.005_04, 5.6.1, and 5.8.8 with threads)
502
5031.96	[2006-11-30]
504	- 1.95 broke builds for threaded Perls, rt.cpan.org tickets:
505	  [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
506	  [rt.cpan.org #23712] Time-HiRes 1.95 Fails make on AIX 5.2 with Perl 5.8.8
507	  [rt.cpan.org #23730] Time::HiRes 1.95 fails make on MacOS X 10.3.9/perl 5.8.8
508	- use main() prototype consistently in Makefile.PL
509
5101.95	[2006-11-29]
511	- integrate core change #29180: Silence VC++ compiler warnings
512	  from Steve Hay
513	- do not use PL_ppaddr in stat() because that is not available
514	  in Perl 5.005_04
515	- regenerate fallback/*.inc for older Perls without
516          ExtUtils::Constant because of d_hires_stat, resolves
517	  [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
518	- Make Makefile.PL more defensive against false PERL_CORE
519
5201.94	[2006-10-16]
521	- file timestamps oddities seen: the atime and mtime
522	  can be out of sync (modify first and read second can leave
523	  atime < mtime) and mtime can be subsecond while atime is not.
524	  So make the test more forgiving.
525
5261.93	[2006-10-15]
527	- the ualarm() tests (34-37) assumed that ualarm(N)
528	  could never alarm in less than N seconds, widened
529	  the acceptable relative range to 0.9..1.5.  Addresses
530	  [rt.cpan.org #22090] and [rt.cpan.org #22091].
531
532	- skip the stat() tests in cygwin and win32, because
533	  if run on FAT the timestamp granularity is only 2 seconds.
534	  Any good way to detect (cygwin or win32) whether we are
535	  being run on NTFS or anywhere with better timestamps?
536	  Addresses [rt.cpan.org #22089] and [rt.cpan.org #22098].
537
5381.92	[2006-10-13]
539	- scan for subsecond resolution timestamps in struct stat,
540	  some known possibilities:
541
542	  (1)  struct  timespec st_atimespec;
543                                st_atimespec.tv_nsec;
544	  (2)  time_t  st_atime;
545	       long    st_atimensec;
546          (3)  time_t  st_atime;
547               int     st_atime_n;
548	  (4)  timestruc_t st_atim;
549                           st_atim.tv_nsec
550	  (5)  time_t  st_atime;
551               int     st_uatime;
552
553          If something like this is found, one can do
554
555	    use Time::HiRes;
556	    my @stat = Time::HiRes::stat();
557
558	  or even override the standard stat():
559
560	    use Time::HiRes qw(stat);
561
562          to get the stat() timestamps
563
564	    my ($atime, $mtime, $ctime) = @stat[8, 9, 10];
565
566	  with subsecond resolution (assuming both the operating
567	  system and the filesystem support that kind of thing).
568
569	  Contributions for more systems (especially non-UNIX,
570	  e.g. but not limited to: Win32, VMS, OS/2) gladly accepted.
571	  (also more UNIX variants welcome: HP-UX? IRIX?)
572
573	  Thanks to H.Merijn Brand, John Peacock, and Craig
574	  Berry for brave beta testing.
575
5761.91	[2006-09-29]
577	- ualarm() in SuSE 10.1 was overflowing after ~4.2 seconds,
578	  possibly due to a glibc bug/feature (suspected overflow at
579	  2**32 microseconds?), workaround by using the setitimer()
580	  implementation of ualarm() if either useconds or
581	  interval > 999_999 (this case seems to vary between systems:
582	  are useconds more than 999_999 for ualarm() defined or not)
583	  Added more ualarm() tests to catch various overflow points,
584	  hopefully no problems in various platforms.
585	  (The problem report by Mark Seger and Jon Paul Sullivan of HP.)
586
5871.90	[2006-08-22]
588	- tweak still needed for Const64(), from Jerry Hedden
589	- get a freshly generated ppport.h
590	- update Copyright years
591
5921.89	[2006-08-22]
593	- Const64() already appends an 'LL' (or i64), so provide LL and i64
594	  forms for the IV_1E[679] (effects Win32 and Cygwin), reported by
595	  Jerry Hedden.
596	- the Changes entry for 1.88 talked about [IN]V_1[679],
597	  missing the 'E'.
598
5991.88	[2006-08-21]
600	- clean up the g++ warnings in HiRes.xs, all of them
601	  about mixing integer and floating point, introduce
602	  constants IV_1E[679] and NV_1E[679]
603
6041.87	[2006-02-13]
605	- [rt.cpan.org #17442] 'make test' frequently fails under
606	  Cygwin Perl v5.8.8, reported and patched by J. R. Hedden
607	  (two race condition bugs in the END block in the case the
608	   main process dies before the timer process, unearthed
609	   by a bug in Cygwin ualarm)
610
6111.86	[2005-12-17]
612	- HiRes.t:s/ok 32/ok 33/, from Dominic Dunlop
613	- tighten up the clock() test marginally by requiring non-negative
614	- clock_nanosleep() and clock() doc tweaks
615
6161.85	[2005-12-16]
617	- the interface to clock_nanosleep() is more natural
618	  when it is like (hires) time() (instead of like nanosleep),
619	  and the .xs implementation of clock_nanosleep() in 1.84
620	  was broken anyway
621	- the semantics of clock() are not quite so silly as I thought,
622	  but still somewhat odd, documented as such
623	- additional enhancements to the clock() documentation
624	- add test for clock_nanosleep() (I cannot test this
625	  since none of my systems have the function)
626	- add test for clock()
627
6281.84	[2005-12-16]
629	- add clock() which returns the processor time in
630	  (floating point) seconds since an arbitrary era
631	- add clock_nanosleep() which suspends the current
632	  thread until either absolute time or for relative time
633	- [rt.cpan.org #16486] printf missing value in HiRes.t
634	- add constants CLOCKS_PER_SEC, CLOCK_SOFTTIME, TIMER_ABSTIME
635	- tiny typo fixes
636
6371.83	[2005-11-19]
638	- has_symbol() was wrong since e.g. ITIMER_VIRTUAL is exported
639	  via @EXPORT_OK even when it is not available.  This is heinous.
640	  @EXPORT_OK should be determined at Makefile.PL time.
641	- be more lenient is testing clock_gettime(): allow more slop,
642	  and retry up to three times, sleeping a random nap between
643	  the retries
644	- human months are one-based (noticed by Anton Berezin)
645
6461.82	[2005-10-06]
647	- CLOCK_REALTIME is an enum value (of the clockid_t enum)
648	  in HP-UX (and might be so elsewhere, too), debugged by
649	  H. Merijn Brand
650	- include const-c.inc as late as possible (from Randy Kobes,
651	  [rt.cpan.org #15552] to avoid undefined usleep() on Win32
652
6531.81	[2005-11-05]
654	- try to be more robust and consistent in the detection of
655          CLOCK_REALTIME and ITIMER_VIRTUAL in HiRes.t: the proper
656	  way is
657
658		sub has_symbol {
659		    my $symbol = shift;
660		    eval 'import Time::HiRes qw($symbol)';
661		    return 0 unless $@ eq '';
662		    return exists ${"Time::HiRes::$symbol"};
663		}
664
665	  and then use
666
667		&FOO_BAR
668
669	  in the test.  All these moves are needed because
670
671	  1) one cannot directly do eval 'Time::HiRes::FOO_BAR'
672	     because FOO_BAR might have a true value of zero
673	     (or in the general case an empty string or even undef)
674
675	  2) In case FOO_BAR is not available in this platform,
676	     &FOO_BAR avoids the bareword warning
677
678	- wait more (1.5 seconds instead of 0.1) for the CLOCK_REALTIME test
679	  but expect the 'customary' slop of 0.20 instead of 0.25
680	- fixed inside a comment HAS_POLL -> TIME_HIRES_NANOSLEEP
681	- at the end of HiRest.t tell how close we were to termination
682
6831.80	[2005-11-04]
684	- Gisle noticed a mistake (using HAS_NANOSLEEP) in 1.79
685
6861.79	[2005-11-03]
687	- try nanosleep for emulating usleep -- may help in some weird
688	  embedded realtime places which have nanosleep but neither usleep
689	  nor select nor poll (doesn't have to be weird embedded realtime
690	  place, though -- in many places usleep is nanosleep anyway)
691	- try poll for emulating usleep -- this may help some obscure/old
692	  SVR4 places that have neither usleep nor select
693	- a redundant test guard in HiRes.t
694
6951.78	[2005-11-03]
696	- ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
697	  in the general case fail already at 'use' phase is suboptimal)
698	- fixes to the documentation of clock_gettime() and clock_getres()
699
7001.77	[2005-11-03]
701	- add support for the POSIX clock_gettime() and clock_getres(),
702	  if available, either as library calls or as syscalls
703	- be more defensive about missing functionality: break out
704	  early (during 'use') if no e.g. clock_getres() is available,
705	  and protect our back by trapping those cases also in HiRes.xs
706	- the test added in 1.76 could cause an endless loop e.g. in Solaris,
707	  due to mixing of sleep() and alarm() (bad programmer, no cookie!)
708
7091.76	[2005-10-22]
710	- testing for nanosleep had wrong logic which caused nanosleep
711	  to become undefined for e.g. Mac OS X
712	- added a test for a core dump that was introduced by Perl 5.8.0
713	  safe signals and was fixed for the time of 5.8.1 (one report of
714	  the core dump was [perl #20920]), the test skipped pre-5.8.1.
715	- *cough* s/unanosleep/nanosleep/g; *cough*
716
7171.75	[2005-10-18]
718	- installation patch from Gisle Aas: in Perls 5.8.x and later
719	  use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
720
7211.74	[2005-09-19]
722	- [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
723	  (the workaround is not to use rel2abs, should not be necessary)
724	- [cpan #14642] U2time wrongly exported on the C API
725	  (patch supplied by the reporter, SALVA@cpan.org)
726	- add release dates to Changes
727
7281.73	[2005-08-16]
729	- Time::HiRes::nanosleep support for Solaris [PATCH]
730	  (POSIX::uname() not available if building with core perl,
731	   from Gisle Aas, via perl5-porters, perl change #25295)
732
7331.72	[2005-07-01]
734	- going back to the 1.68 loader setup (using DynaLoader)
735	  since too many weird things starting breaking
736	- fix a typo in José Auguste-Etienne's name
737
7381.71	[2005-06-28]
739	- a thinko in the nanosleep() detection
740	- move more changes stuff from the README to Changes
741	- add -w to the Makefile.PL
742
7431.70	[2005-06-26]
744	- oops in 1.69 about @ISA (not affecting anything but silly)
745	- add copyright 2005 to HiRes.pm
746	- add copyright and license to HiRes.xs
747	- add copyrights 2003, 2004, 2005 to README
748
7491.69	[2005-06-25]
750	- actually run a test for nanosleep
751	  (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
752	  it seems that one can link in nanosleep() but then calling
753	  it fails instantly and sets errno to ENOSYS (Not implemented).
754	  This may be fixable in the AIX case by figuring out the right
755	  (realtime POSIX?) libs and whatnot, but in the general case
756	  running a real test case is better.  (Of course, this change
757	  will no doubt run into portability problems because of the
758	  execution step...)  Note that because of hysterical raisins
759	  most Perls do NOT have $Config{d_nanosleep} (scanning for
760	  it by Configure would in many platforms require linking in
761	  things like -lrt, which would in many platforms be a bad idea
762	  for Perl itself).
763	  (from José Auguste-Etienne)
764	- support XSLoader also since it's much faster
765	  (from Alexey Tourbin)
766	- add SEE ALSO (BSD::Resource and Time::TAI64)
767
7681.68	[2005-05-14]
769	- somehow 1.67 had a lot of doubled lines (a major cut-and-paste
770	  error suspected), but miraculously it still worked since the
771	  doubling took place below the __END__ token
772	- undef Pause() before defining it to avoid redefinition warnings
773	  during compilation in case perl.h had already defined Pause()
774	  (part of perl change #24271)
775	- minor doc tweaks
776
7771.67	[2005-05-04]
778	- (internal) don't ignore the return value of gettimeofday()
779	- (external) return undef or an empty if the C gettimeofday() fails
780	  (affects Time::HiRes gettimeofday() and the hires time())
781
7821.66	[2004-12-19]
783	- add nanosleep()
784	- fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
785	- should now build in Solaris [rt.cpan.org #7165] (since 1.64)
786	- should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
787	- close also [rt.cpan.org #5933] "Time::HiRes::time does not
788	  pick up time adjustments like ntp" since ever reproducing it
789	  (and therefore verifying a possible fix) in the same environment
790	  has become rather unlikely
791
7921.65	[2004-09-18]
793	- one should not mix u?alarm and sleep (the tests modified
794	  by 1.65, #12 and #13, hung in Solaris), now we just busy
795	  loop executing an empty block
796	- in the documentation underline the unspecificity of mixing
797	  sleeps and alarms
798	- small spelling fixes
799
8001.64	[2004-09-16]
801	- regenerate ppport.h with Devel::PPPort 3.03,
802	  now the MY_CXT_CLONE is defined in ppport.h,
803	  we no more need to do that.
804
805	- the test #12 would often hang in sigsuspend() (at least that's
806	  where Mac OS X' ktrace shows it hanging).  With the sleep()s
807	  changed to sleep(1)s, the tests still pass but no hang after
808	  a few hundred repeats.
809
8101.63	[2004-09-01]
811	- Win32 and any ithread build: ppport.h didn't define
812	  MY_CXT_CLONE, which seems to be a Time-HiRes-ism.
813
8141.62	[2004-08-31]
815	- Skip testing if under PERL_CORE and Time::HiRes has not
816	  been Configured (from Marcus Holland-Moritz, core change
817	  #23246)
818	- Use ppport.h generated by Devel::PPPort 3.01,
819	  allowing cutting away our own portability code.
820	- Don't use $ENV{PERL_CORE} for < 5.6.0.
821	- Don't use "for my $i" for <= 5.003.
822	- Don't use Pause() for <= 5.003.
823	- Can't use newSVpvf for <= 5.003.
824	(most of the changes from Marcus)
825
8261.61	[2004-08-21]
827	- Win32: reset reading from the performance counters every
828	  five minutes to better track wall clock time (thanks to
829	  PC timers being often quite bad), should help long-running
830	  programs.
831
8321.60	[2004-08-15]
833	- Win32: Patch from Steve Hay
834	  [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
835	  to [perl #30755] reported by Nigel Sandever
836
837	- Cygwin: Use the Win32 recalibration code also in Cygwin if the
838	  <w32api/windows.h> APIs are available.  Cygwin testing by
839	  Yitzchak Scott-Thoennes.
840
841	- Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
842	  after that keep using -lrt, patch from Alan Burlison,
843	  bug reported in [cpan #7165]
844
8451.59	[2004-04-08]
846	- Change the Win32 recalibration limit to 0.5 seconds and tweak
847	  the documentation to blather less about the gory details of the
848	  Win32 implementation and more about the complications in general
849	  of meddling with the system clock.
850
8511.58	[2004-04-08]
852	- Document the 1.57 change better.
853
8541.57	[2004-07-04]
855	- Win32/Cygwin/MinGW: if the performance counter drifts by more
856	  than two seconds from the system clock (due to ntp adjustments,
857	  for example), recalibrate our internal counter: from Jan Dubois,
858	  based on [cpan #5933] by Jerry D. Hedden.
859
8601.56	[2004-29-02]
861	- Give a clearer message if the tests timeout (perl change #22253)
862	- Don't use /tmp or its moral equivalents (perl bug #15036,
863	  perl change #22258)
864
8651.55	[2004-01-14]
866	- Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
867	  instead of VC-specific i64 suffix)
868
8691.54	[2003-12-31]
870	- Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
871
8721.53	[2003-12-30]
873	- Windows: higher resolution time() by using the Windows
874	  performance counter API, from Jan Dubois and Anton Shcherbinin.
875	  The exact new higher resolution depends on the hardware,
876	  but it should be quite a bit better than using the basic
877	  Windows timers.
878
8791.52	[2003-10-28]
880	- In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
881	  the subtest 18.  No known analysis nor fix, but added
882	  an alarm (that requires fork() and alarm()) to the test.
883
8841.51	[2003-09-22]
885	- doc tweaks from mjd (perl change #20456)
886	- NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
887
8881.50	[2003-08-02]
889	- add a message (for non-core builds) to Makefile.PL about
890	  the LC_ALL=C workaround
891	- &Time::HiRes::d_nanosleep was broken (perl change #20131)
892	- the nanosleep() probe was broken (perl change #20061)
893	- use existence instead of definedness for feature probes
894	  (perl change #20043)
895	- MPE/iX tweak (perl change #20042)
896	- do not use HAS_NANOSLEEP (perl change #19898)
897
8981.49	[2003-06-23]
899	- UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
900	- OS/2 can always mix subsecond sleeps with signals
901	  (part of perl change #19789)
902
9031.48	[2003-06-04]
904	- workaround for buggy gcc 2.95.3 in openbsd/sparc64
905	  (perl change #19592)
906
9071.47	[2003-05-03]
908	- do not use -lrt in Linux (from March Lehmann, perl change #19449)
909		- unnecessary (nanosleep is in libc anyway)
910		- harmful (-lrt slows down execution)
911		- incompatible (with many distributions' pthreads)
912
9131.46	[2003-04-25]
914	- do not create files in blib directories under core
915	  (perl change #19160, from rgs)
916	- detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
917
9181.45	[2003-04-01]
919	- guarantee that $xdefine in HiRes.t is always defined
920	  (perl change #19109, from IlyaZ)
921	- a cleaner way to detect PERL_CORE (perl change #19111,
922	  from IlyaZ)
923
9241.44	[2003-03-30]
925	- add hints/irix.pl to turn off overly POSIX flags that
926	  cause hide struct timespec to be hidden (and compilation
927	  to fail) (bleadperl change #19085)
928	- documentation tweaks
929
9301.43	[2003-03-11]
931	- add c:/temp to the list of temp directories to probe
932	  so that cygwin (and win*?) builds are happy.  This was
933	  needed at least in my cygwin 1.3.20/w2k setup.
934
9351.42	[2003-01-07]
936	- modernize the constants code (from Nicholas Clark)
937
9381.41	[2003-01-03]
939	- At some point the ability to figure our the correct incdir
940	  for EXTERN.h (either a core perl build, or an installed perl)
941	  had broken (which lead into all test compiles failing with
942	  a core perl build, but thanks to the robustness of Makefile.PL
943	  nothing of this was visible).  The brokenness seemed to be
944	  caused by $ENV{PERL_CORE} not being on for core builds?
945	  Now stole a trick from the Encode that sets $ENV{PERL_CORE}
946	  right, and both styles of build should work again.
947
9481.40	[2003-01-03]
949	- Nicholas Clark noticed that the my_catdir() emulation function
950	  was broken (which means that we didn't really work for Perls
951	  5.002 and 5.003)
952	- inspired by fixing the above made the whole Makefile.PL -w
953	  and strict clean
954	- tightened up the Makefile.PL output, less whitespace
955
9561.39	[2003-10-20]
957	- fix from Craig Berry for better building in VMS with PERL_CORE
958
9591.38	[2003-10-13]
960	- no functional changes
961	- move lib/Time/HiRes.pm as Hires.pm
962	- libraries scanning was slightly broken (always scanned
963	  for a library even when $Config{libs} already had it)
964
9651.37	[2003-09-23]
966	- Ray Zimmerman ran into a race condition in Mac OS X.
967	  A 0.01-second alarm fired before the test expected.
968	  The test first slept indefinitely (blocking for signals)
969	  and only after that tested for the signal having been sent.
970	  Since the signal had already been sent, the test #12 never
971	  completed.  The solution: test first, then block.
972	- default to being silent on all probing attempts, set the
973	  environment variable VERBOSE to a true value to see the
974	  details (the probing command and the possible errors)
975
9761.36	[2003-09-12]
977	- do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
978	- INSTALLDIRS => 'perl' missing which means that Time::HiRes
979	  cannot be upgraded from CPAN to override the 5.8.0 version
980	  (Guido A. Ostkamp)
981	- Time::HiRes 1.35 could not be dropped as-is to bleadperl
982	  because the include directories did not adjust themselves
983	  if $ENV{PERL_CORE} (Hugo van der Sanden)
984	- add documentation about the restart of select() under alarm()
985
9861.35	[2003-08-24]
987	- small documentation tweaks
988
989
9901.34	[2003-08-22]
991	- better VMS operation (Craig Berry)
992
9931.33	[2003-08-20]
994	- our time machine is accelerating: now works with Perl 5.004_01
995	  (tried with 5.003_07 and 5.002 but I get segmentation faults
996	   from running the Makefile.PL with those in Tru64 4.0D)
997
9981.32	[2003-08-20]
999	- backward compatibility (pre-5.6.0) tweaks:
1000	  - no XSLoader in 5.00503, use DynaLoader instead
1001	  - no SvPV_nolen, either
1002	  - no PerlProc_pause(), either
1003	  - now tested with 5.00404 and 5.00503
1004	  - Makefile.PL requires 5.00404 (no more 5.002)
1005	- use nanosleep instead of usleep, if it is available (Wilson Snyder)
1006	  (this means that one can mix subsecond sleeps with alarms)
1007	- because of nanosleep we probe for -lrt and -lposix4
1008	- the existence of getitimer/nanosleep/setitimer/ualarm/usleep
1009	  is available by exportable constants Time::HiRes::d_func
1010	  (since older Perl do not have them in %Config, and even
1011	   5.8.0 does not probe for nanosleep)
1012
10131.31	[2003-08-19]
1014	- backward compatibility (pre-5.6.1) tweaks:
1015	  - define NV if no NVTYPE
1016	  - define IVdf if needed (note: the Devel::PPPort
1017	    in 5.8.0 does not try hard enough since
1018	    the IVSIZE might not be defined)
1019	  - define NVgf if needed
1020	  - grab the typemap from 5.8.0 for the NV stuff
1021
1022	1.31 and 1.32 add more backward compatibility (now all the way
1023	back to Perl 5.00404), and using nanosleep() (if available) for
1024	subsecond sleeps.
1025
10261.30	[2003-08-16]
1027
1028	- release 1.29_02 as 1.30
1029
1030	1.30 adds all the changes made during the Perl 5.6->5.7->5.8
1031	development cycle.  Most notably portability across platforms has been
1032	enhanced, and the interval timers (setitimer, getitimer) have been
1033	added.  Note that the version of Time::HiRes that is included in Perl
1034	5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
1035	version.  Note also that in 1.30 Wegscheid turns over the maintenance
1036	to Jarkko Hietaniemi.
1037
10381.29_02	[2003-08-16]
1039
1040	- fix a silly unclosed comment typo in HiRes.xs
1041	- document and export REALTIME_REALPROF (Solaris)
1042
10431.29_01	[2003-08-16]
1044
1045	- only getitimer(ITIMER_REAL) available in Cygwin and Win32
1046	  (need to patch this also in Perl 5.[89])
1047	- remove CVS revision log from HiRes.xs
1048
10491.29_00	[2003-08-14]
1050
1051	The following numbered patches refer to the Perl 5.7 changes,
1052	you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
1053
1054	- 17558: Add #!./perl to the .t
1055	- 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
1056	- 16198: political correctness, from Simon Cozens
1057	- 15857: doc tweaks, from Jarkko Hietaniemi
1058	- 15593: optimization in .xs, from Paul Green
1059	- 14892: pod fixes, from Robin Barker
1060	- 14100: VOS fixes, from Paul Green
1061	- 13422: XS segfault, from Marc Lehmann
1062	- 13378: whether select() gets restarted on signals, depends
1063	- 13354: timing constraints, again, from Andy Dougherty
1064	- 13278: can't do subsecond alarms with ualarm;
1065		 break out early if alarms do not seem to be working
1066	- 13266: test relaxation (cygwin gets lower hires
1067		 times than lores ones)
1068	- 12846: protect against high load, from Jarkko Hietaniemi
1069	- 12837: HiRes.t VMS tweak, from Craig A. Berry
1070	- 12797: HiRes.t VMS tweak, from Charles Lane
1071	- 12769: HiRes.t VMS tweak, from Craig A. Berry
1072	- 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
1073	- 12722: VMS ualarm for VMS without ualarm, from Charles Lane
1074	- 12692: alarm() ain't gonna work if ualarm() ain't,
1075		 from Gurusamy Sarathy
1076	- 12680: minor VMS tweak, from Charles Lane
1077	- 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
1078	- 12609: croak on negative time, from Jarkko Hietaniemi
1079	- 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
1080	- 12594: MacOS Classic timeofday, from Chris Nandor
1081	- 12473: allow for more than one second for sleep() and usleep()
1082	- 12458: test tuning, relax timing constraints,
1083		 from Jarkko Hietaniemi
1084	- 12449: make sleep() and usleep() to return the number
1085		 of seconds and microseconds actually slept (analogously
1086		 with the builtin sleep()), also make usleep() croak if
1087		 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
1088	- 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
1089	- 12199: do not use ftime on Win32, from Gurusamy Sarathy
1090	- 12196: use ftime() on Win32, from Artur Bergman
1091	- 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
1092	- 12105: use GetSystemTime() on Win32, from Artur Bergman
1093	- 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
1094	- 11901: UNICOS sloppy division, from Jarkko Hietaniemi
1095	- 11797: problem in HiRes.t, from John P. Linderman
1096	- 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
1097	- 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
1098	- 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
1099	- 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
1100	- 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
1101	- 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
1102		 from Jonathan Stowe
1103	- 10942: MPE/IX test tweaks, from Mark Bixby
1104	- 10784: unnecessary pod2man calls, from Andy Dougherty
1105	- 10354: ext/ + -Wall, from Doug MacEachern
1106	- 10320: fix the BOOT section to call myU2time correctly
1107	- 10317: correct casting for AIX< from H. Merijn Brand
1108	- 10119: document that the core time() may be rounding, not truncating
1109	- 10118: test fix, from John Peacock
1110	-  9988: long =item, from Robin Barker
1111	-  9714: correct test output
1112	-  9708: test also the scalar aspect of getitimer()
1113	-  9705: Add interval timers (setitimer, getitimer)
1114	-  9692: do not require at least 5.005 using XS
1115
1116	The following changes were made on top of the changes
1117	made for Time::HiRes during the Perl 5.7 development
1118	cycle that culminated in the release of Perl 5.8.0.
1119
1120	- add "require 5.005" to the Makefile.PL
1121	- remove the REVISION section (CVS log) from HiRes.pm
1122	- add jhi's copyright alongside Douglas'
1123	- move HiRes.pm to lib/Time/
1124	- move HiRes.t to t/
1125	- modify HiRes.t to use $ENV{PERL_CORE}
1126	- modify the original Time::HiRes version 1.20 Makefile.PL
1127	  to work both with Perl 5.8.0 and the new code with pre-5.8.0
1128	  Perls (tried with 5.6.1)
1129	- tiny tweaks and updates in README and TODO
1130	- bump the VERSION to 1.29
1131
11321.20  Wed Feb 24 21:30 1999
1133	- make our usleep and ualarm substitutes into hrt_usleep
1134	  and hrt_ualarm. This helps static links of Perl with other
1135	  packages that also have usleep, etc. From
1136	  Ilya Zakharevich <ilya@math.ohio-state.edu>
1137	- add C API stuff. From Joshua Pritikin
1138	  <joshua.pritikin@db.com>
1139	- VMS Makefile.PL fun.	From pvhp@forte.com (Peter Prymmer)
1140	- hopefully correct "-lc" fix for SCO.
1141	- add PPD stuff
1142
1143	1.20 adds a platform neutral set of C accessible routines if you are
1144	running 5.005+.  All other changes are packaging changes and build
1145	fixes(?) for statically linked Perl, SCO, and VMS.
1146
11471.19  Tue Sep 29 22:30 1998
1148	- put VMS gettimeofday() in. Patch is from Sebastian Bazley
1149	  <seb@stian.demon.co.uk>
1150	- change GIMME_V to GIMME to help people with older versions of
1151	  Perl.
1152	- fix Win32 version of gettimeofday(). It didn't affect anything,
1153	  but it confuses people reading the code when the return value
1154	  is backwards (0 is success).
1155	- fix Makefile.PL (more) so that detection of gettimeofday is
1156	  more correct.
1157
1158	1.19 has better VMS support.
1159
11601.18  Mon Jul 6 22:40 1998
1161	- add usleep() for Win32.
1162	- fix Makefile.PL to fix reported HP/UX feature where unresolved
1163	  externals still cause an executable to be generated (though no
1164	  x bit set). Thanks to David Kozinn for report and explanation.
1165	  Problems with the fix are mine :)
1166
1167	1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
1168	Probably buggy. I'm sure I'll hear.
1169
11701.17  Wed Jul 1 20:10 1998
1171	- fix setitimer calls so microseconds is not more than 1000000.
1172	  Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
1173	- make Win32. We only get gettimeofday (the select hack doesn't
1174	  seem to work on my Win95 system).
1175	- fix test 4 on 01test.t. add test to see if time() and
1176	  Time::HiRes::time() are close.
1177
11781.16  Wed Nov 12 21:05 1997
1179	- add missing EXTEND in new gettimeofday scalar code.
1180
1181	1.16+ should be closer to building out of the box on Linux. Thanks
1182	to Gisle Aas for patches, and the ualarm equivalent using setitimer.
1183
1184	If your underlying operating system doesn't implement ualarm(), then
1185	a fake using setitimer() will be made.  If the OS is missing usleep(),
1186	a fake one using select() will be made. If a fake can't be made for
1187	either ualarm() or usleep(), then the corresponding Perl function will
1188	not be available.  If the OS is missing gettimeofday(), you will get
1189	unresolved externals, either at link- or run-time.
1190
1191	This is an improvement; the package used to not even build if
1192	you were missing any of these bits. Roderick Schertler
1193
1194	<roderick@argon.org> did all the conditional compilation stuff,
1195	look at HiRes.pm and the test suites; it's good educational reading.
1196
11971.15  Mon Nov 10 21:30 1997
1198	- HiRes.pm: update pod. Provided by Gisle Aas.
1199	- HiRes.xs: if gettimeofday() called in scalar context, do
1200	  something more useful than before. Provided by Gisle Aas.
1201	- README: tell of xsubpp '-nolinenumber' woes. thanks to
1202	  Edward Henigin <ed@texas.net> for pointing out the problem.
1203
12041.14  Wed Nov 5 9:40 1997
1205	- Makefile.PL: look for setitimer
1206	- HiRes.xs: if missing ualarm, but we have setitimer, make up
1207	  our own setitimer. These were provided by Gisle Aas.
1208
12091.13  Tue Nov 4 23:30 1997
1210	- Makefile.PL: fix autodetect mechanism to do try linking in addition
1211	  to just compiling; should fix Linux build problem. Fix was provided
1212	  by Gisle Aas.
1213
12141.12  Sun Oct 12 12:00:00 1997
1215	- Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
1216	  you may need to comment this back out if you have an older xsubpp.
1217	- HiRes.xs: set PROTOTYPES: DISABLE
1218
12191.11  Fri Sep 05 16:00:00 1997
1220	- Makefile.PL:
1221	  Had some line commented out that shouldn't have been (testing
1222	  remnants)
1223	- README:
1224	  Previous version was corrupted.
1225
12261.10  Thu May 22 20:20:00 1997
1227	- HiRes.xs, HiRes.pm, t/*:
1228	      -	only compile what we have OS support for (or can
1229		fake with select())
1230	      - only test what we compiled
1231	      - gross improvement to the test suite
1232	      - fix EXPORT_FAIL.
1233	  This work was all done by Roderick Schertler
1234	  <roderick@argon.org>. If you run Linux or
1235	  one of the other ualarm-less platforms, and you like this
1236	  module, let Roderick know; without him, it still wouldn't
1237	  be working on those boxes...
1238	- Makefile.PL: figure out what routines the OS has and
1239	  only build what we need. These bits were written by Jarkko
1240	  Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
1241
12421.02  Mon Dec 30 08:00:00 1996
1243	- HiRes.pm: update documentation to say what to do when missing
1244	  ualarm() and friends.
1245	- README: update to warn that ualarm() and friends need to exist
1246
12471.01  Fri Oct 17 08:00:00 1996
1248	- Makefile.PL: make XSPROTOARGS => '-noprototyopes'
1249	- HiRes.pm: put blank line between __END__ and =head1 so that
1250	  pod2man works.
1251
12521.00  Tue Sep 03 13:00:00 1996
1253	- original version; created by h2xs 1.16
1254