This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Time::HiRes version bump.
[perl5.git] / dist / Time-HiRes / Changes
CommitLineData
df003a36
JH
1Revision history for the Perl extension Time::HiRes.
2
1625f1d3
JH
31.9730 [2016-02-17]
4 - TIME_HIRES_DONT_RUN_PROBES=1 to build the probes but not run them
5 [rt.cpan.org #111391]
6
bcd2c0ef
JH
71.9729 [2016-02-13]
8 - new constants
9 [rt.cpan.org #111170]
10 - clock_getres(), clock_gettime(), clock_nanosleep() emulation for OS X
11 [rt.cpan.org #78566]
12
df003a36
JH
131.9728 [2015-11-13]
14 - No changes since 1.9727_03
15
161.9727_03 [2015-11-06]
17 - use PERL_UNUSED_ARG() in the not-here-croak branches
18 - Makefile.PL/metadata improvements
19
201.9727_02 [2015-08-29]
21 - correct documentation to note that Time::NVtime is NV, not double
22 [perl #123879]
23 - avoid a warning about clock_t on OS X / clang 3.6 - 3.6
24 [rt.cpan.org #102718]
25
261.9727_01 [2015-08-17]
27 - Remove usage of 5.005 thread APIs in favor of 5.6 and up thread APIs
28 [rt.cpan.org #102799].
29 - Silence a C++ compiler warning [rt.cpan.org #101497].
30 - respect $Config{run} for cross-compiling [rt.cpan.org #92607]
31
321.9726 [2013-08-17]
33 - Correct s/us splitting of usleep(1000000) [rt.cpan.org #78266].
34 - Avoid integer overflow in itimer-based alarm() with large
35 argument [rt.cpan.org #87160].
36 - Define PERL_NO_INLINE_FUNCTIONS during probe compilations, to
37 avoid false negatives caused by not linking with the perl core.
38 - Be more careful about context in stat().
39 - Install into "site" directories by default on perl 5.11+
40 [rt.cpan.org #79797].
41 - Fix a couple of doc typos [rt.cpan.org #85365].
42 - Fix function name in a doc example [rt.cpan.org #86318].
43 - Provide lstat() that yields high-res timestamps, alongside
44 the existing high-res stat() [rt.cpan.org #78732].
45
461.9725 [2012-02-01]
47 - Correct stack discipline in stat(), which was screwing up list
48 operations in expressions containing calls to it [rt.cpan.org
49 #72926].
50 - Add missing OUTPUT sections to the XS code [rt.cpan.org #70930].
51 - Skip itimer tests on GNU/Hurd, which has the API but lacks
52 the implementation [rt.cpan.org #72754].
53 - Fix a doubled word in the documentation [rt.cpan.org #72763].
54
551.9724 [2011-06-09]
56 - Correct XS parameter list, and therefore prototype, for
57 unimplemented-on-this-platform version of clock_nanosleep()
58 [rt.cpan.org #68700].
59 - Declare package variables with "our" rather than "use vars".
60 - Corresponding to "our" usage, check for minimum Perl version
61 5.006.
62 - Declare module dependencies.
63
641.9723 [2011-06-07]
65 - Remove $ENV{PERL_CORE} logic from test suite, which is no
66 longer desired in the core.
67 - Convert test suite to use Test::More.
68 - Factor out watchdog code from test suite.
69 - In test suite, be consistent about using fully-qualified form
70 of function names.
71 - Divide test suite into feature-specific scripts.
72 - Make ualarm timing test less vulnerable to delay-induced false
73 failure, from Dave Mitchell.
74
751.9722 [2011-05-18]
76 - Update for changes in build process in the core, patches
77 from BinGOs [rt.cpan.org #58858] and Craig Berry [rt.cpan.org
78 #63363].
79 - Fix broken linkage on Windows with gcc 3.4 seen with ActivePerl,
80 report from Christian Walde [rt.cpan.org #61648], fix derived
81 from Vincent Pit.
82 - Jump through hoops to avoid compiler warnings.
83
841.9721 [2010-03-17]
85 - Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
86 additional fix from Gisle Aas
87 - Address [rt.cpan.org #55665] "Bad plan" on Windows,
88 report and fix from Jan Dubois
89
901.9720 [2010-02-14]
91 - Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
92 report and fix from Nicholas Clark
93
941.9719 [2009-01-04]
95 - As with QNX, Haiku has the API of interval timers but not
96 the implementation (bleadperl change #34630), hence skip
97 the tests, via David Mitchell.
98
991.9718 [2008-12-31]
100 - .xs code cleanup from Albert Dvornik
101 - in the #39 and #40 do not do us I did, mixing alarm() and
102 sleep(). Now instead spin until enough time has passed.
103
1041.9717 [2008-12-30]
105 - Skip the tests added in 1.9716 (#39, #40) if there's no subsecond
106 alarm capability, like with the older subsecond alarm tests
107
1081.9716 [2008-12-26]
109 - Change documentation to agree with reality: there are
110 no interval timers in Win32.
111 - Address [rt.cpan.org #35899] (problem in subsecond sleeps),
112 add two tests to guard against this problem
113 - Address [rt.cpan.org #36600] 'Division by zero' failure in test suite
114 - Address [rt.cpan.org #37340] [PATCH] Address timer process in test
115 - Address [rt.cpan.org#40311 ] bad implementation of hrt_usleep
116 with TIME_HIRES_NANOSLEEP
117
1181.9715 [2008-04-08]
119 - Silly me: Makefile.PL does need to accept arguments other than mine.
120 Some testing frameworks obviously do this.
121 - Add retrying for tests 34..37, which are the most commonly
122 failing tests. If this helps, consider extending the retry
123 framework to all the tests. [Inspired by Slaven Rezic,
124 [rt.cpan.org #34711] Occasional failures of test 35 or 36 (FreeBSD)]
125
1261.9714 [2008-04-07]
127 - Under Perl 5.6.* NVgf needs to be "g", reported by Zefram,
128 it seems that ppport.h 3.13 gets this wrong.
129 - remove the check in Makefile.PL for 5.7.2, shouldn't be
130 (a) necessary (b) relevant
131 - add logic to Makefile.PL to skip configure/write Makefile
132 step if the "xdefine" file already exists, indicating that
133 the configure step has already been done, one can still
134 force (re)configure by "perl Makefile.PL configure",
135 or of course by "make clean && perl Makefile.PL".
136
1371.9713 [2008-04-04]
138 - for alarm() and ualarm() [Perl] prefer setitimer() [C]
139 instead of ualarm() [C] since ualarm() [C] cannot portably
140 (and standards-compliantly) be used for more than 999_999
141 microseconds (rt.cpan.org #34655)
142 - it seems that HP-UX has started (at least in 11.31 ia64)
143 #defining the CLOCK_REALTIME et alia (instead of having
144 them just as enums)
b88265bf 145 - document all the diagnostics
df003a36
JH
146
1471.9712 [2008-02-09]
148 - move the sub tick in the test file back to where it used to be
149 - in the "consider upgrading" message recommend at least Perl 5.8.8
150 and make the message to appear only for 5.8.0 since 5.8.1 and
151 later have the problem fixed
152 - VOS tweak for Makefile (core perl change #33259)
153 - since the test #17 seems to fail often, relax its limits a bit
154
1551.9711 [2007-11-29]
156 - lost VMS test skippage from Craig Berry
157 - reformat the test code a little
158
1591.9710 [2007-11-29]
160 - I got the sense of the QNX test the wrong way in an attempt
161 to generalize it for future
162
1631.9709 [2007-11-28]
164 - casting fixes from Robin Barker for g++ and 64bitint
165 - in QNX skip the itimer tests because though the API
166 is there, the implementation isn't, from Matt Kraai
167 - raise the dead man timer to 180 seconds for really
168 slow/busy systems
169 - elaborate the UTF-8 locale warning from Makefile.PL
170
1711.9708 [2007-10-05]
172 - [rt.cpan.org #29747]: Build failure with perl 5.005_05
173 Fixed by regenerating the ppport.h using Devel::PPPort 3.13.
174
1751.9707 [2007-02-27]
176 - t/HiRes.t failed in Perl 5.6.2,
177 "action is not of type POSIX::SigAction at t/HiRes.t line 318",
178 reported and fixed by Anton Berezin, the reason was faulty
179 use of sigaction() when restoring the old action.
180
1811.9706 [2007-02-25]
182 - with bleadperl in VMS the HiRes.t overrun the maximum number
183 of deferred signals because the libc SIGALRM was not strong
184 enough to interrupt select(), and select() got restarted every
185 time, solution is to use POSIX::SigAction if available.
186 A fix from Craig Berry (not 100% there, but helps).
187 - allow for more measuring noise for ualarm() tests 35..37
188
1891.9705 [2007-02-06]
190 - nanosleep() and clock_nanosleep() detection and use were
191 quite broken; in Linux -lrt needed; fixes from Zefram
192 - [internal] slightly cleaner building of $DEFINE in Makefile.PL,
193 should avoid double/conflicting -D flags
194
1951.9704 [2007-01-01]
196 - allow 10% of slop in test #14 (testing difference between
197 CORE::time() and Time::HiRes::time()), there seem to be often
198 transient failures from Perl smoke builds on this test
199 - small pod tweaks
200
2011.9703 [2006-12-08]
202 - use int main(int argc, char **argv) consistently in Makefile.PL,
203 should help with
204 [rt.cpan.org #23868] nanosleep not detected under Mac OS 10.3.9 starting with Time::HiRes 1.96
205 - if someone still has the locale-broken Perl 5.8.0,
206 suggest that they upgrade their Perl
207
2081.9702 [2006-12-06]
209 - restore the -DATLEASTFIVEOHOHFIVE, Win32 needed it still
210
2111.9701 [2006-12-04]
212 - upgrade to ppport.h 3.10_02
213 - remove the -DATLEASTFIVEOHOHFIVE
214 - use the ppport.h PL_ppaddr, PL_statcache, PL_laststatval
215 - use the ppport.h aTHXR for calling Perl stat()
216 - switch into four-digit version since 2.0 is coming up
217 awfully fast but not feeling like a major rewrite
218
2191.97 [2006-11-30]
220 - 1.95 broke building in Win32 (since pp_stat is not exported),
221 figured out how to call an op directly in 5.005 (use Perl_ppaddr
222 instead of PL_ppaddr)
223 - backport to Perl 5.004_05 (requires using statcache
224 and laststatval instead of PL_statcache and PL_laststatval)
225 (also checked to work in 5.005_04, 5.6.1, and 5.8.8 with threads)
226
2271.96 [2006-11-30]
228 - 1.95 broke builds for threaded Perls, rt.cpan.org tickets:
229 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
230 [rt.cpan.org #23712] Time-HiRes 1.95 Fails make on AIX 5.2 with Perl 5.8.8
231 [rt.cpan.org #23730] Time::HiRes 1.95 fails make on MacOS X 10.3.9/perl 5.8.8
232 - use main() prototype consistently in Makefile.PL
233
2341.95 [2006-11-29]
235 - integrate core change #29180: Silence VC++ compiler warnings
236 from Steve Hay
237 - do not use PL_ppaddr in stat() because that is not available
238 in Perl 5.005_04
239 - regenerate fallback/*.inc for older Perls without
240 ExtUtils::Constant because of d_hires_stat, resolves
241 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
242 - Make Makefile.PL more defensive against false PERL_CORE
243
2441.94 [2006-10-16]
245 - file timestamps oddities seen: the atime and mtime
246 can be out of sync (modify first and read second can leave
247 atime < mtime) and mtime can be subsecond while atime is not.
248 So make the test more forgiving.
249
2501.93 [2006-10-15]
251 - the ualarm() tests (34-37) assumed that ualarm(N)
252 could never alarm in less than N seconds, widened
253 the acceptable relative range to 0.9..1.5. Addresses
254 [rt.cpan.org #22090] and [rt.cpan.org #22091].
255
256 - skip the stat() tests in cygwin and win32, because
257 if run on FAT the timestamp granularity is only 2 seconds.
258 Any good way to detect (cygwin or win32) whether we are
259 being run on NTFS or anywhere with better timestamps?
260 Addresses [rt.cpan.org #22089] and [rt.cpan.org #22098].
261
2621.92 [2006-10-13]
263 - scan for subsecond resolution timestamps in struct stat,
264 some known possibilities:
265
266 (1) struct timespec st_atimespec;
267 st_atimespec.tv_nsec;
268 (2) time_t st_atime;
269 long st_atimensec;
270 (3) time_t st_atime;
271 int st_atime_n;
272 (4) timestruc_t st_atim;
273 st_atim.tv_nsec
274 (5) time_t st_atime;
275 int st_uatime;
276
277 If something like this is found, one can do
278
279 use Time::HiRes;
280 my @stat = Time::HiRes::stat();
281
282 or even override the standard stat():
b88265bf 283
df003a36
JH
284 use Time::HiRes qw(stat);
285
286 to get the stat() timestamps
287
288 my ($atime, $mtime, $ctime) = @stat[8, 9, 10];
289
290 with subsecond resolution (assuming both the operating
291 system and the filesystem support that kind of thing).
292
293 Contributions for more systems (especially non-UNIX,
294 e.g. but not limited to: Win32, VMS, OS/2) gladly accepted.
295 (also more UNIX variants welcome: HP-UX? IRIX?)
296
297 Thanks to H.Merijn Brand, John Peacock, and Craig
298 Berry for brave beta testing.
299
3001.91 [2006-09-29]
301 - ualarm() in SuSE 10.1 was overflowing after ~4.2 seconds,
302 possibly due to a glibc bug/feature (suspected overflow at
303 2**32 microseconds?), workaround by using the setitimer()
304 implementation of ualarm() if either useconds or
305 interval > 999_999 (this case seems to vary between systems:
306 are useconds more than 999_999 for ualarm() defined or not)
307 Added more ualarm() tests to catch various overflow points,
308 hopefully no problems in various platforms.
309 (The problem report by Mark Seger and Jon Paul Sullivan of HP.)
310
3111.90 [2006-08-22]
312 - tweak still needed for Const64(), from Jerry Hedden
313 - get a freshly generated ppport.h
314 - update Copyright years
315
3161.89 [2006-08-22]
317 - Const64() already appends an 'LL' (or i64), so provide LL and i64
318 forms for the IV_1E[679] (effects Win32 and Cygwin), reported by
319 Jerry Hedden.
320 - the Changes entry for 1.88 talked about [IN]V_1[679],
321 missing the 'E'.
322
3231.88 [2006-08-21]
324 - clean up the g++ warnings in HiRes.xs, all of them
325 about mixing integer and floating point, introduce
326 constants IV_1E[679] and NV_1E[679]
327
3281.87 [2006-02-13]
329 - [rt.cpan.org #17442] 'make test' frequently fails under
330 Cygwin Perl v5.8.8, reported and patched by J. R. Hedden
331 (two race condition bugs in the END block in the case the
332 main process dies before the timer process, unearthed
333 by a bug in Cygwin ualarm)
334
3351.86 [2005-12-17]
336 - HiRes.t:s/ok 32/ok 33/, from Dominic Dunlop
337 - tighten up the clock() test marginally by requiring non-negative
338 - clock_nanosleep() and clock() doc tweaks
339
3401.85 [2005-12-16]
341 - the interface to clock_nanosleep() is more natural
342 when it is like (hires) time() (instead of like nanosleep),
343 and the .xs implementation of clock_nanosleep() in 1.84
344 was broken anyway
345 - the semantics of clock() are not quite so silly as I thought,
346 but still somewhat odd, documented as such
347 - additional enhancements to the clock() documentation
348 - add test for clock_nanosleep() (I cannot test this
349 since none of my systems have the function)
350 - add test for clock()
351
3521.84 [2005-12-16]
353 - add clock() which returns the processor time in
354 (floating point) seconds since an arbitrary era
355 - add clock_nanosleep() which suspends the current
356 thread until either absolute time or for relative time
357 - [rt.cpan.org #16486] printf missing value in HiRes.t
358 - add constants CLOCKS_PER_SEC, CLOCK_SOFTTIME, TIMER_ABSTIME
359 - tiny typo fixes
360
3611.83 [2005-11-19]
362 - has_symbol() was wrong since e.g. ITIMER_VIRTUAL is exported
363 via @EXPORT_OK even when it is not available. This is heinous.
364 @EXPORT_OK should be determined at Makefile.PL time.
365 - be more lenient is testing clock_gettime(): allow more slop,
366 and retry up to three times, sleeping a random nap between
367 the retries
368 - human months are one-based (noticed by Anton Berezin)
369
3701.82 [2005-10-06]
371 - CLOCK_REALTIME is an enum value (of the clockid_t enum)
372 in HP-UX (and might be so elsewhere, too), debugged by
373 H. Merijn Brand
374 - include const-c.inc as late as possible (from Randy Kobes,
375 [rt.cpan.org #15552] to avoid undefined usleep() on Win32
376
3771.81 [2005-11-05]
378 - try to be more robust and consistent in the detection of
379 CLOCK_REALTIME and ITIMER_VIRTUAL in HiRes.t: the proper
380 way is
381
382 sub has_symbol {
383 my $symbol = shift;
384 eval 'import Time::HiRes qw($symbol)';
385 return 0 unless $@ eq '';
386 return exists ${"Time::HiRes::$symbol"};
387 }
388
389 and then use
390
391 &FOO_BAR
392
393 in the test. All these moves are needed because
394
395 1) one cannot directly do eval 'Time::HiRes::FOO_BAR'
396 because FOO_BAR might have a true value of zero
397 (or in the general case an empty string or even undef)
398
399 2) In case FOO_BAR is not available in this platform,
400 &FOO_BAR avoids the bareword warning
401
402 - wait more (1.5 seconds instead of 0.1) for the CLOCK_REALTIME test
403 but expect the 'customary' slop of 0.20 instead of 0.25
404 - fixed inside a comment HAS_POLL -> TIME_HIRES_NANOSLEEP
405 - at the end of HiRest.t tell how close we were to termination
406
4071.80 [2005-11-04]
408 - Gisle noticed a mistake (using HAS_NANOSLEEP) in 1.79
409
4101.79 [2005-11-03]
411 - try nanosleep for emulating usleep -- may help in some weird
412 embedded realtime places which have nanosleep but neither usleep
413 nor select nor poll (doesn't have to be weird embedded realtime
414 place, though -- in many places usleep is nanosleep anyway)
415 - try poll for emulating usleep -- this may help some obscure/old
416 SVR4 places that have neither usleep nor select
417 - a redundant test guard in HiRes.t
418
4191.78 [2005-11-03]
420 - ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
421 in the general case fail already at 'use' phase is suboptimal)
422 - fixes to the documentation of clock_gettime() and clock_getres()
423
4241.77 [2005-11-03]
425 - add support for the POSIX clock_gettime() and clock_getres(),
426 if available, either as library calls or as syscalls
427 - be more defensive about missing functionality: break out
428 early (during 'use') if no e.g. clock_getres() is available,
429 and protect our back by trapping those cases also in HiRes.xs
430 - the test added in 1.76 could cause an endless loop e.g. in Solaris,
431 due to mixing of sleep() and alarm() (bad programmer, no cookie!)
432
4331.76 [2005-10-22]
434 - testing for nanosleep had wrong logic which caused nanosleep
435 to become undefined for e.g. Mac OS X
436 - added a test for a core dump that was introduced by Perl 5.8.0
437 safe signals and was fixed for the time of 5.8.1 (one report of
438 the core dump was [perl #20920]), the test skipped pre-5.8.1.
439 - *cough* s/unanosleep/nanosleep/g; *cough*
440
4411.75 [2005-10-18]
442 - installation patch from Gisle Aas: in Perls 5.8.x and later
443 use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
444
4451.74 [2005-09-19]
446 - [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
447 (the workaround is not to use rel2abs, should not be necessary)
448 - [cpan #14642] U2time wrongly exported on the C API
449 (patch supplied by the reporter, SALVA@cpan.org)
450 - add release dates to Changes
451
4521.73 [2005-08-16]
453 - Time::HiRes::nanosleep support for Solaris [PATCH]
454 (POSIX::uname() not available if building with core perl,
455 from Gisle Aas, via perl5-porters, perl change #25295)
456
4571.72 [2005-07-01]
458 - going back to the 1.68 loader setup (using DynaLoader)
459 since too many weird things starting breaking
460 - fix a typo in José Auguste-Etienne's name
461
4621.71 [2005-06-28]
463 - a thinko in the nanosleep() detection
464 - move more changes stuff from the README to Changes
465 - add -w to the Makefile.PL
466
4671.70 [2005-06-26]
468 - oops in 1.69 about @ISA (not affecting anything but silly)
469 - add copyright 2005 to HiRes.pm
470 - add copyright and license to HiRes.xs
471 - add copyrights 2003, 2004, 2005 to README
472
4731.69 [2005-06-25]
474 - actually run a test for nanosleep
475 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
476 it seems that one can link in nanosleep() but then calling
477 it fails instantly and sets errno to ENOSYS (Not implemented).
478 This may be fixable in the AIX case by figuring out the right
479 (realtime POSIX?) libs and whatnot, but in the general case
480 running a real test case is better. (Of course, this change
481 will no doubt run into portability problems because of the
482 execution step...) Note that because of hysterical raisins
483 most Perls do NOT have $Config{d_nanosleep} (scanning for
484 it by Configure would in many platforms require linking in
485 things like -lrt, which would in many platforms be a bad idea
486 for Perl itself).
487 (from José Auguste-Etienne)
488 - support XSLoader also since it's much faster
489 (from Alexey Tourbin)
490 - add SEE ALSO (BSD::Resource and Time::TAI64)
491
4921.68 [2005-05-14]
493 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
494 error suspected), but miraculously it still worked since the
495 doubling took place below the __END__ token
496 - undef Pause() before defining it to avoid redefinition warnings
497 during compilation in case perl.h had already defined Pause()
498 (part of perl change #24271)
499 - minor doc tweaks
500
5011.67 [2005-05-04]
502 - (internal) don't ignore the return value of gettimeofday()
503 - (external) return undef or an empty if the C gettimeofday() fails
504 (affects Time::HiRes gettimeofday() and the hires time())
505
5061.66 [2004-12-19]
507 - add nanosleep()
508 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
509 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
510 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
511 - close also [rt.cpan.org #5933] "Time::HiRes::time does not
512 pick up time adjustments like ntp" since ever reproducing it
b88265bf 513 (and therefore verifying a possible fix) in the same environment
df003a36
JH
514 has become rather unlikely
515
5161.65 [2004-09-18]
517 - one should not mix u?alarm and sleep (the tests modified
518 by 1.65, #12 and #13, hung in Solaris), now we just busy
519 loop executing an empty block
520 - in the documentation underline the unspecificity of mixing
521 sleeps and alarms
522 - small spelling fixes
523
5241.64 [2004-09-16]
525 - regenerate ppport.h with Devel::PPPort 3.03,
526 now the MY_CXT_CLONE is defined in ppport.h,
527 we no more need to do that.
528
529 - the test #12 would often hang in sigsuspend() (at least that's
530 where Mac OS X' ktrace shows it hanging). With the sleep()s
531 changed to sleep(1)s, the tests still pass but no hang after
532 a few hundred repeats.
533
5341.63 [2004-09-01]
535 - Win32 and any ithread build: ppport.h didn't define
536 MY_CXT_CLONE, which seems to be a Time-HiRes-ism.
537
5381.62 [2004-08-31]
539 - Skip testing if under PERL_CORE and Time::HiRes has not
540 been Configured (from Marcus Holland-Moritz, core change
541 #23246)
542 - Use ppport.h generated by Devel::PPPort 3.01,
543 allowing cutting away our own portability code.
544 - Don't use $ENV{PERL_CORE} for < 5.6.0.
545 - Don't use "for my $i" for <= 5.003.
546 - Don't use Pause() for <= 5.003.
547 - Can't use newSVpvf for <= 5.003.
548 (most of the changes from Marcus)
549
5501.61 [2004-08-21]
551 - Win32: reset reading from the performance counters every
552 five minutes to better track wall clock time (thanks to
553 PC timers being often quite bad), should help long-running
554 programs.
555
5561.60 [2004-08-15]
557 - Win32: Patch from Steve Hay
558 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
559 to [perl #30755] reported by Nigel Sandever
560
561 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
562 <w32api/windows.h> APIs are available. Cygwin testing by
563 Yitzchak Scott-Thoennes.
564
565 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
566 after that keep using -lrt, patch from Alan Burlison,
567 bug reported in [cpan #7165]
568
5691.59 [2004-04-08]
570 - Change the Win32 recalibration limit to 0.5 seconds and tweak
571 the documentation to blather less about the gory details of the
572 Win32 implementation and more about the complications in general
573 of meddling with the system clock.
574
5751.58 [2004-04-08]
576 - Document the 1.57 change better.
577
5781.57 [2004-07-04]
579 - Win32/Cygwin/MinGW: if the performance counter drifts by more
580 than two seconds from the system clock (due to ntp adjustments,
581 for example), recalibrate our internal counter: from Jan Dubois,
582 based on [cpan #5933] by Jerry D. Hedden.
583
5841.56 [2004-29-02]
585 - Give a clearer message if the tests timeout (perl change #22253)
586 - Don't use /tmp or its moral equivalents (perl bug #15036,
587 perl change #22258)
588
5891.55 [2004-01-14]
590 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
591 instead of VC-specific i64 suffix)
592
5931.54 [2003-12-31]
594 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
595
5961.53 [2003-12-30]
597 - Windows: higher resolution time() by using the Windows
598 performance counter API, from Jan Dubois and Anton Shcherbinin.
599 The exact new higher resolution depends on the hardware,
600 but it should be quite a bit better than using the basic
601 Windows timers.
602
6031.52 [2003-10-28]
604 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
605 the subtest 18. No known analysis nor fix, but added
606 an alarm (that requires fork() and alarm()) to the test.
607
6081.51 [2003-09-22]
609 - doc tweaks from mjd (perl change #20456)
610 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
611
6121.50 [2003-08-02]
613 - add a message (for non-core builds) to Makefile.PL about
614 the LC_ALL=C workaround
615 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
616 - the nanosleep() probe was broken (perl change #20061)
617 - use existence instead of definedness for feature probes
618 (perl change #20043)
619 - MPE/iX tweak (perl change #20042)
620 - do not use HAS_NANOSLEEP (perl change #19898)
621
6221.49 [2003-06-23]
623 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
624 - OS/2 can always mix subsecond sleeps with signals
625 (part of perl change #19789)
626
6271.48 [2003-06-04]
628 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
629 (perl change #19592)
630
6311.47 [2003-05-03]
632 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
633 - unnecessary (nanosleep is in libc anyway)
634 - harmful (-lrt slows down execution)
635 - incompatible (with many distributions' pthreads)
636
6371.46 [2003-04-25]
638 - do not create files in blib directories under core
639 (perl change #19160, from rgs)
640 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
641
6421.45 [2003-04-01]
643 - guarantee that $xdefine in HiRes.t is always defined
644 (perl change #19109, from IlyaZ)
645 - a cleaner way to detect PERL_CORE (perl change #19111,
646 from IlyaZ)
647
6481.44 [2003-03-30]
649 - add hints/irix.pl to turn off overly POSIX flags that
650 cause hide struct timespec to be hidden (and compilation
651 to fail) (bleadperl change #19085)
652 - documentation tweaks
653
6541.43 [2003-03-11]
655 - add c:/temp to the list of temp directories to probe
656 so that cygwin (and win*?) builds are happy. This was
657 needed at least in my cygwin 1.3.20/w2k setup.
658
6591.42 [2003-01-07]
660 - modernize the constants code (from Nicholas Clark)
661
6621.41 [2003-01-03]
663 - At some point the ability to figure our the correct incdir
664 for EXTERN.h (either a core perl build, or an installed perl)
665 had broken (which lead into all test compiles failing with
666 a core perl build, but thanks to the robustness of Makefile.PL
667 nothing of this was visible). The brokenness seemed to be
668 caused by $ENV{PERL_CORE} not being on for core builds?
669 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
670 right, and both styles of build should work again.
671
6721.40 [2003-01-03]
673 - Nicholas Clark noticed that the my_catdir() emulation function
674 was broken (which means that we didn't really work for Perls
675 5.002 and 5.003)
676 - inspired by fixing the above made the whole Makefile.PL -w
677 and strict clean
678 - tightened up the Makefile.PL output, less whitespace
679
6801.39 [2003-10-20]
681 - fix from Craig Berry for better building in VMS with PERL_CORE
682
6831.38 [2003-10-13]
684 - no functional changes
685 - move lib/Time/HiRes.pm as Hires.pm
686 - libraries scanning was slightly broken (always scanned
687 for a library even when $Config{libs} already had it)
688
6891.37 [2003-09-23]
690 - Ray Zimmerman ran into a race condition in Mac OS X.
691 A 0.01-second alarm fired before the test expected.
692 The test first slept indefinitely (blocking for signals)
693 and only after that tested for the signal having been sent.
694 Since the signal had already been sent, the test #12 never
695 completed. The solution: test first, then block.
696 - default to being silent on all probing attempts, set the
697 environment variable VERBOSE to a true value to see the
698 details (the probing command and the possible errors)
699
7001.36 [2003-09-12]
701 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
702 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
703 cannot be upgraded from CPAN to override the 5.8.0 version
704 (Guido A. Ostkamp)
705 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
706 because the include directories did not adjust themselves
707 if $ENV{PERL_CORE} (Hugo van der Sanden)
708 - add documentation about the restart of select() under alarm()
709
7101.35 [2003-08-24]
711 - small documentation tweaks
712
713
7141.34 [2003-08-22]
715 - better VMS operation (Craig Berry)
716
7171.33 [2003-08-20]
718 - our time machine is accelerating: now works with Perl 5.004_01
719 (tried with 5.003_07 and 5.002 but I get segmentation faults
720 from running the Makefile.PL with those in Tru64 4.0D)
721
7221.32 [2003-08-20]
723 - backward compatibility (pre-5.6.0) tweaks:
724 - no XSLoader in 5.00503, use DynaLoader instead
725 - no SvPV_nolen, either
726 - no PerlProc_pause(), either
727 - now tested with 5.00404 and 5.00503
728 - Makefile.PL requires 5.00404 (no more 5.002)
729 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
730 (this means that one can mix subsecond sleeps with alarms)
731 - because of nanosleep we probe for -lrt and -lposix4
732 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
733 is available by exportable constants Time::HiRes::d_func
734 (since older Perl do not have them in %Config, and even
735 5.8.0 does not probe for nanosleep)
736
7371.31 [2003-08-19]
738 - backward compatibility (pre-5.6.1) tweaks:
739 - define NV if no NVTYPE
740 - define IVdf if needed (note: the Devel::PPPort
741 in 5.8.0 does not try hard hard enough since
742 the IVSIZE might not be defined)
743 - define NVgf if needed
744 - grab the typemap from 5.8.0 for the NV stuff
745
746 1.31 and 1.32 add more backward compatibility (now all the way
747 back to Perl 5.00404), and using nanosleep() (if available) for
748 subsecond sleeps.
749
7501.30 [2003-08-16]
751
752 - release 1.29_02 as 1.30
753
754 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
755 development cycle. Most notably portability across platforms has been
756 enhanced, and the interval timers (setitimer, getitimer) have been
757 added. Note that the version of Time::HiRes that is included in Perl
758 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
759 version. Note also that in 1.30 Wegscheid turns over the maintenance
760 to Jarkko Hietaniemi.
761
7621.29_02 [2003-08-16]
763
764 - fix a silly unclosed comment typo in HiRes.xs
765 - document and export REALTIME_REALPROF (Solaris)
766
7671.29_01 [2003-08-16]
768
769 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
770 (need to patch this also in Perl 5.[89])
771 - remove CVS revision log from HiRes.xs
772
7731.29_00 [2003-08-14]
774
775 The following numbered patches refer to the Perl 5.7 changes,
776 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
777
778 - 17558: Add #!./perl to the .t
779 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
780 - 16198: political correctness, from Simon Cozens
781 - 15857: doc tweaks, from Jarkko Hietaniemi
782 - 15593: optimization in .xs, from Paul Green
783 - 14892: pod fixes, from Robin Barker
784 - 14100: VOS fixes, from Paul Green
785 - 13422: XS segfault, from Marc Lehmann
786 - 13378: whether select() gets restarted on signals, depends
787 - 13354: timing constraints, again, from Andy Dougherty
788 - 13278: can't do subsecond alarms with ualarm;
789 break out early if alarms do not seem to be working
790 - 13266: test relaxation (cygwin gets lower hires
791 times than lores ones)
792 - 12846: protect against high load, from Jarkko Hietaniemi
793 - 12837: HiRes.t VMS tweak, from Craig A. Berry
794 - 12797: HiRes.t VMS tweak, from Charles Lane
795 - 12769: HiRes.t VMS tweak, from Craig A. Berry
796 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
797 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
798 - 12692: alarm() ain't gonna work if ualarm() ain't,
799 from Gurusamy Sarathy
800 - 12680: minor VMS tweak, from Charles Lane
801 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
802 - 12609: croak on negative time, from Jarkko Hietaniemi
803 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
b88265bf 804 - 12594: MacOS Classic timeofday, from Chris Nandor
df003a36
JH
805 - 12473: allow for more than one second for sleep() and usleep()
806 - 12458: test tuning, relax timing constraints,
807 from Jarkko Hietaniemi
808 - 12449: make sleep() and usleep() to return the number
809 of seconds and microseconds actually slept (analogously
810 with the builtin sleep()), also make usleep() croak if
811 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
812 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
813 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
814 - 12196: use ftime() on Win32, from Artur Bergman
815 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
816 - 12105: use GetSystemTime() on Win32, from Artur Bergman
817 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
818 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
819 - 11797: problem in HiRes.t, from John P. Linderman
820 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
821 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
b88265bf 822 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
df003a36
JH
823 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
824 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
825 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
826 from Jonathan Stowe
827 - 10942: MPE/IX test tweaks, from Mark Bixby
b88265bf 828 - 10784: unnecessary pod2man calls, from Andy Dougherty
df003a36
JH
829 - 10354: ext/ + -Wall, from Doug MacEachern
830 - 10320: fix the BOOT section to call myU2time correctly
831 - 10317: correct casting for AIX< from H. Merijn Brand
832 - 10119: document that the core time() may be rounding, not truncating
833 - 10118: test fix, from John Peacock
834 - 9988: long =item, from Robin Barker
835 - 9714: correct test output
836 - 9708: test also the scalar aspect of getitimer()
837 - 9705: Add interval timers (setitimer, getitimer)
838 - 9692: do not require at least 5.005 using XS
b88265bf 839
df003a36
JH
840 The following changes were made on top of the changes
841 made for Time::HiRes during the Perl 5.7 development
b88265bf 842 cycle that culminated in the release of Perl 5.8.0.
df003a36
JH
843
844 - add "require 5.005" to the Makefile.PL
845 - remove the REVISION section (CVS log) from HiRes.pm
846 - add jhi's copyright alongside Douglas'
847 - move HiRes.pm to lib/Time/
848 - move HiRes.t to t/
849 - modify HiRes.t to use $ENV{PERL_CORE}
850 - modify the original Time::HiRes version 1.20 Makefile.PL
851 to work both with Perl 5.8.0 and the new code with pre-5.8.0
852 Perls (tried with 5.6.1)
853 - tiny tweaks and updates in README and TODO
854 - bump the VERSION to 1.29
855
8561.20 Wed Feb 24 21:30 1999
b88265bf 857 - make our usleep and ualarm substitutes into hrt_usleep
df003a36
JH
858 and hrt_ualarm. This helps static links of Perl with other
859 packages that also have usleep, etc. From
860 Ilya Zakharevich <ilya@math.ohio-state.edu>
861 - add C API stuff. From Joshua Pritikin
862 <joshua.pritikin@db.com>
863 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
864 - hopefully correct "-lc" fix for SCO.
865 - add PPD stuff
866
867 1.20 adds a platform neutral set of C accessible routines if you are
868 running 5.005+. All other changes are packaging changes and build
869 fixes(?) for statically linked Perl, SCO, and VMS.
870
8711.19 Tue Sep 29 22:30 1998
872 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
873 <seb@stian.demon.co.uk>
874 - change GIMME_V to GIMME to help people with older versions of
875 Perl.
876 - fix Win32 version of gettimeofday(). It didn't affect anything,
877 but it confuses people reading the code when the return value
878 is backwards (0 is success).
879 - fix Makefile.PL (more) so that detection of gettimeofday is
880 more correct.
881
882 1.19 has better VMS support.
883
8841.18 Mon Jul 6 22:40 1998
885 - add usleep() for Win32.
886 - fix Makefile.PL to fix reported HP/UX feature where unresolved
887 externals still cause an executable to be generated (though no
888 x bit set). Thanks to David Kozinn for report and explanation.
889 Problems with the fix are mine :)
890
891 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
892 Probably buggy. I'm sure I'll hear.
893
8941.17 Wed Jul 1 20:10 1998
895 - fix setitimer calls so microseconds is not more than 1000000.
896 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
897 - make Win32. We only get gettimeofday (the select hack doesn't
898 seem to work on my Win95 system).
b88265bf 899 - fix test 4 on 01test.t. add test to see if time() and
df003a36
JH
900 Time::HiRes::time() are close.
901
9021.16 Wed Nov 12 21:05 1997
903 - add missing EXTEND in new gettimeofday scalar code.
904
905 1.16+ should be closer to building out of the box on Linux. Thanks
906 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
907
908 If your underlying operating system doesn't implement ualarm(), then
909 a fake using setitimer() will be made. If the OS is missing usleep(),
910 a fake one using select() will be made. If a fake can't be made for
911 either ualarm() or usleep(), then the corresponding Perl function will
912 not be available. If the OS is missing gettimeofday(), you will get
913 unresolved externals, either at link- or run-time.
914
915 This is an improvement; the package used to not even build if
916 you were missing any of these bits. Roderick Schertler
917
918 <roderick@argon.org> did all the conditional compilation stuff,
919 look at HiRes.pm and the test suites; it's good educational reading.
920
9211.15 Mon Nov 10 21:30 1997
922 - HiRes.pm: update pod. Provided by Gisle Aas.
923 - HiRes.xs: if gettimeofday() called in scalar context, do
924 something more useful than before. Provided by Gisle Aas.
925 - README: tell of xsubpp '-nolinenumber' woes. thanks to
926 Edward Henigin <ed@texas.net> for pointing out the problem.
927
9281.14 Wed Nov 5 9:40 1997
929 - Makefile.PL: look for setitimer
930 - HiRes.xs: if missing ualarm, but we have setitimer, make up
931 our own setitimer. These were provided by Gisle Aas.
932
9331.13 Tue Nov 4 23:30 1997
934 - Makefile.PL: fix autodetect mechanism to do try linking in addition
935 to just compiling; should fix Linux build problem. Fix was provided
936 by Gisle Aas.
937
9381.12 Sun Oct 12 12:00:00 1997
939 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
940 you may need to comment this back out if you have an older xsubpp.
941 - HiRes.xs: set PROTOTYPES: DISABLE
942
9431.11 Fri Sep 05 16:00:00 1997
944 - Makefile.PL:
945 Had some line commented out that shouldn't have been (testing
946 remnants)
947 - README:
948 Previous version was corrupted.
949
9501.10 Thu May 22 20:20:00 1997
951 - HiRes.xs, HiRes.pm, t/*:
b88265bf 952 - only compile what we have OS support for (or can
df003a36 953 fake with select())
b88265bf 954 - only test what we compiled
df003a36 955 - gross improvement to the test suite
b88265bf 956 - fix EXPORT_FAIL.
df003a36
JH
957 This work was all done by Roderick Schertler
958 <roderick@argon.org>. If you run Linux or
b88265bf
JH
959 one of the other ualarm-less platforms, and you like this
960 module, let Roderick know; without him, it still wouldn't
df003a36
JH
961 be working on those boxes...
962 - Makefile.PL: figure out what routines the OS has and
b88265bf 963 only build what we need. These bits were written by Jarkko
df003a36
JH
964 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
965
9661.02 Mon Dec 30 08:00:00 1996
967 - HiRes.pm: update documentation to say what to do when missing
968 ualarm() and friends.
969 - README: update to warn that ualarm() and friends need to exist
970
9711.01 Fri Oct 17 08:00:00 1996
972 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
b88265bf 973 - HiRes.pm: put blank line between __END__ and =head1 so that
df003a36
JH
974 pod2man works.
975
9761.00 Tue Sep 03 13:00:00 1996
977 - original version; created by h2xs 1.16