This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Time-HiRes: record latest changes
[perl5.git] / dist / Time-HiRes / Changes
CommitLineData
df003a36
JH
1Revision history for the Perl extension Time::HiRes.
2
919ca095
JH
31.9731 [2016-03-??]
4 - nanosleep and clock_nanosleep cleanups: blead c8ea02b8..e3ff671b
5 - static funcs in HiRes.xs: 4e922c26
6 - Remove unused variable: 52ffb1b5
7
1625f1d3
JH
81.9730 [2016-02-17]
9 - TIME_HIRES_DONT_RUN_PROBES=1 to build the probes but not run them
10 [rt.cpan.org #111391]
11
bcd2c0ef
JH
121.9729 [2016-02-13]
13 - new constants
14 [rt.cpan.org #111170]
15 - clock_getres(), clock_gettime(), clock_nanosleep() emulation for OS X
16 [rt.cpan.org #78566]
17
df003a36
JH
181.9728 [2015-11-13]
19 - No changes since 1.9727_03
20
211.9727_03 [2015-11-06]
22 - use PERL_UNUSED_ARG() in the not-here-croak branches
23 - Makefile.PL/metadata improvements
24
251.9727_02 [2015-08-29]
26 - correct documentation to note that Time::NVtime is NV, not double
27 [perl #123879]
28 - avoid a warning about clock_t on OS X / clang 3.6 - 3.6
29 [rt.cpan.org #102718]
30
311.9727_01 [2015-08-17]
32 - Remove usage of 5.005 thread APIs in favor of 5.6 and up thread APIs
33 [rt.cpan.org #102799].
34 - Silence a C++ compiler warning [rt.cpan.org #101497].
35 - respect $Config{run} for cross-compiling [rt.cpan.org #92607]
36
371.9726 [2013-08-17]
38 - Correct s/us splitting of usleep(1000000) [rt.cpan.org #78266].
39 - Avoid integer overflow in itimer-based alarm() with large
40 argument [rt.cpan.org #87160].
41 - Define PERL_NO_INLINE_FUNCTIONS during probe compilations, to
42 avoid false negatives caused by not linking with the perl core.
43 - Be more careful about context in stat().
44 - Install into "site" directories by default on perl 5.11+
45 [rt.cpan.org #79797].
46 - Fix a couple of doc typos [rt.cpan.org #85365].
47 - Fix function name in a doc example [rt.cpan.org #86318].
48 - Provide lstat() that yields high-res timestamps, alongside
49 the existing high-res stat() [rt.cpan.org #78732].
50
511.9725 [2012-02-01]
52 - Correct stack discipline in stat(), which was screwing up list
53 operations in expressions containing calls to it [rt.cpan.org
54 #72926].
55 - Add missing OUTPUT sections to the XS code [rt.cpan.org #70930].
56 - Skip itimer tests on GNU/Hurd, which has the API but lacks
57 the implementation [rt.cpan.org #72754].
58 - Fix a doubled word in the documentation [rt.cpan.org #72763].
59
601.9724 [2011-06-09]
61 - Correct XS parameter list, and therefore prototype, for
62 unimplemented-on-this-platform version of clock_nanosleep()
63 [rt.cpan.org #68700].
64 - Declare package variables with "our" rather than "use vars".
65 - Corresponding to "our" usage, check for minimum Perl version
66 5.006.
67 - Declare module dependencies.
68
691.9723 [2011-06-07]
70 - Remove $ENV{PERL_CORE} logic from test suite, which is no
71 longer desired in the core.
72 - Convert test suite to use Test::More.
73 - Factor out watchdog code from test suite.
74 - In test suite, be consistent about using fully-qualified form
75 of function names.
76 - Divide test suite into feature-specific scripts.
77 - Make ualarm timing test less vulnerable to delay-induced false
78 failure, from Dave Mitchell.
79
801.9722 [2011-05-18]
81 - Update for changes in build process in the core, patches
82 from BinGOs [rt.cpan.org #58858] and Craig Berry [rt.cpan.org
83 #63363].
84 - Fix broken linkage on Windows with gcc 3.4 seen with ActivePerl,
85 report from Christian Walde [rt.cpan.org #61648], fix derived
86 from Vincent Pit.
87 - Jump through hoops to avoid compiler warnings.
88
891.9721 [2010-03-17]
90 - Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
91 additional fix from Gisle Aas
92 - Address [rt.cpan.org #55665] "Bad plan" on Windows,
93 report and fix from Jan Dubois
94
951.9720 [2010-02-14]
96 - Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
97 report and fix from Nicholas Clark
98
991.9719 [2009-01-04]
100 - As with QNX, Haiku has the API of interval timers but not
101 the implementation (bleadperl change #34630), hence skip
102 the tests, via David Mitchell.
103
1041.9718 [2008-12-31]
105 - .xs code cleanup from Albert Dvornik
106 - in the #39 and #40 do not do us I did, mixing alarm() and
107 sleep(). Now instead spin until enough time has passed.
108
1091.9717 [2008-12-30]
110 - Skip the tests added in 1.9716 (#39, #40) if there's no subsecond
111 alarm capability, like with the older subsecond alarm tests
112
1131.9716 [2008-12-26]
114 - Change documentation to agree with reality: there are
115 no interval timers in Win32.
116 - Address [rt.cpan.org #35899] (problem in subsecond sleeps),
117 add two tests to guard against this problem
118 - Address [rt.cpan.org #36600] 'Division by zero' failure in test suite
119 - Address [rt.cpan.org #37340] [PATCH] Address timer process in test
120 - Address [rt.cpan.org#40311 ] bad implementation of hrt_usleep
121 with TIME_HIRES_NANOSLEEP
122
1231.9715 [2008-04-08]
124 - Silly me: Makefile.PL does need to accept arguments other than mine.
125 Some testing frameworks obviously do this.
126 - Add retrying for tests 34..37, which are the most commonly
127 failing tests. If this helps, consider extending the retry
128 framework to all the tests. [Inspired by Slaven Rezic,
129 [rt.cpan.org #34711] Occasional failures of test 35 or 36 (FreeBSD)]
130
1311.9714 [2008-04-07]
132 - Under Perl 5.6.* NVgf needs to be "g", reported by Zefram,
133 it seems that ppport.h 3.13 gets this wrong.
134 - remove the check in Makefile.PL for 5.7.2, shouldn't be
135 (a) necessary (b) relevant
136 - add logic to Makefile.PL to skip configure/write Makefile
137 step if the "xdefine" file already exists, indicating that
138 the configure step has already been done, one can still
139 force (re)configure by "perl Makefile.PL configure",
140 or of course by "make clean && perl Makefile.PL".
141
1421.9713 [2008-04-04]
143 - for alarm() and ualarm() [Perl] prefer setitimer() [C]
144 instead of ualarm() [C] since ualarm() [C] cannot portably
145 (and standards-compliantly) be used for more than 999_999
146 microseconds (rt.cpan.org #34655)
147 - it seems that HP-UX has started (at least in 11.31 ia64)
148 #defining the CLOCK_REALTIME et alia (instead of having
149 them just as enums)
b88265bf 150 - document all the diagnostics
df003a36
JH
151
1521.9712 [2008-02-09]
153 - move the sub tick in the test file back to where it used to be
154 - in the "consider upgrading" message recommend at least Perl 5.8.8
155 and make the message to appear only for 5.8.0 since 5.8.1 and
156 later have the problem fixed
157 - VOS tweak for Makefile (core perl change #33259)
158 - since the test #17 seems to fail often, relax its limits a bit
159
1601.9711 [2007-11-29]
161 - lost VMS test skippage from Craig Berry
162 - reformat the test code a little
163
1641.9710 [2007-11-29]
165 - I got the sense of the QNX test the wrong way in an attempt
166 to generalize it for future
167
1681.9709 [2007-11-28]
169 - casting fixes from Robin Barker for g++ and 64bitint
170 - in QNX skip the itimer tests because though the API
171 is there, the implementation isn't, from Matt Kraai
172 - raise the dead man timer to 180 seconds for really
173 slow/busy systems
174 - elaborate the UTF-8 locale warning from Makefile.PL
175
1761.9708 [2007-10-05]
177 - [rt.cpan.org #29747]: Build failure with perl 5.005_05
178 Fixed by regenerating the ppport.h using Devel::PPPort 3.13.
179
1801.9707 [2007-02-27]
181 - t/HiRes.t failed in Perl 5.6.2,
182 "action is not of type POSIX::SigAction at t/HiRes.t line 318",
183 reported and fixed by Anton Berezin, the reason was faulty
184 use of sigaction() when restoring the old action.
185
1861.9706 [2007-02-25]
187 - with bleadperl in VMS the HiRes.t overrun the maximum number
188 of deferred signals because the libc SIGALRM was not strong
189 enough to interrupt select(), and select() got restarted every
190 time, solution is to use POSIX::SigAction if available.
191 A fix from Craig Berry (not 100% there, but helps).
192 - allow for more measuring noise for ualarm() tests 35..37
193
1941.9705 [2007-02-06]
195 - nanosleep() and clock_nanosleep() detection and use were
196 quite broken; in Linux -lrt needed; fixes from Zefram
197 - [internal] slightly cleaner building of $DEFINE in Makefile.PL,
198 should avoid double/conflicting -D flags
199
2001.9704 [2007-01-01]
201 - allow 10% of slop in test #14 (testing difference between
202 CORE::time() and Time::HiRes::time()), there seem to be often
203 transient failures from Perl smoke builds on this test
204 - small pod tweaks
205
2061.9703 [2006-12-08]
207 - use int main(int argc, char **argv) consistently in Makefile.PL,
208 should help with
209 [rt.cpan.org #23868] nanosleep not detected under Mac OS 10.3.9 starting with Time::HiRes 1.96
210 - if someone still has the locale-broken Perl 5.8.0,
211 suggest that they upgrade their Perl
212
2131.9702 [2006-12-06]
214 - restore the -DATLEASTFIVEOHOHFIVE, Win32 needed it still
215
2161.9701 [2006-12-04]
217 - upgrade to ppport.h 3.10_02
218 - remove the -DATLEASTFIVEOHOHFIVE
219 - use the ppport.h PL_ppaddr, PL_statcache, PL_laststatval
220 - use the ppport.h aTHXR for calling Perl stat()
221 - switch into four-digit version since 2.0 is coming up
222 awfully fast but not feeling like a major rewrite
223
2241.97 [2006-11-30]
225 - 1.95 broke building in Win32 (since pp_stat is not exported),
226 figured out how to call an op directly in 5.005 (use Perl_ppaddr
227 instead of PL_ppaddr)
228 - backport to Perl 5.004_05 (requires using statcache
229 and laststatval instead of PL_statcache and PL_laststatval)
230 (also checked to work in 5.005_04, 5.6.1, and 5.8.8 with threads)
231
2321.96 [2006-11-30]
233 - 1.95 broke builds for threaded Perls, rt.cpan.org tickets:
234 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
235 [rt.cpan.org #23712] Time-HiRes 1.95 Fails make on AIX 5.2 with Perl 5.8.8
236 [rt.cpan.org #23730] Time::HiRes 1.95 fails make on MacOS X 10.3.9/perl 5.8.8
237 - use main() prototype consistently in Makefile.PL
238
2391.95 [2006-11-29]
240 - integrate core change #29180: Silence VC++ compiler warnings
241 from Steve Hay
242 - do not use PL_ppaddr in stat() because that is not available
243 in Perl 5.005_04
244 - regenerate fallback/*.inc for older Perls without
245 ExtUtils::Constant because of d_hires_stat, resolves
246 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
247 - Make Makefile.PL more defensive against false PERL_CORE
248
2491.94 [2006-10-16]
250 - file timestamps oddities seen: the atime and mtime
251 can be out of sync (modify first and read second can leave
252 atime < mtime) and mtime can be subsecond while atime is not.
253 So make the test more forgiving.
254
2551.93 [2006-10-15]
256 - the ualarm() tests (34-37) assumed that ualarm(N)
257 could never alarm in less than N seconds, widened
258 the acceptable relative range to 0.9..1.5. Addresses
259 [rt.cpan.org #22090] and [rt.cpan.org #22091].
260
261 - skip the stat() tests in cygwin and win32, because
262 if run on FAT the timestamp granularity is only 2 seconds.
263 Any good way to detect (cygwin or win32) whether we are
264 being run on NTFS or anywhere with better timestamps?
265 Addresses [rt.cpan.org #22089] and [rt.cpan.org #22098].
266
2671.92 [2006-10-13]
268 - scan for subsecond resolution timestamps in struct stat,
269 some known possibilities:
270
271 (1) struct timespec st_atimespec;
272 st_atimespec.tv_nsec;
273 (2) time_t st_atime;
274 long st_atimensec;
275 (3) time_t st_atime;
276 int st_atime_n;
277 (4) timestruc_t st_atim;
278 st_atim.tv_nsec
279 (5) time_t st_atime;
280 int st_uatime;
281
282 If something like this is found, one can do
283
284 use Time::HiRes;
285 my @stat = Time::HiRes::stat();
286
287 or even override the standard stat():
b88265bf 288
df003a36
JH
289 use Time::HiRes qw(stat);
290
291 to get the stat() timestamps
292
293 my ($atime, $mtime, $ctime) = @stat[8, 9, 10];
294
295 with subsecond resolution (assuming both the operating
296 system and the filesystem support that kind of thing).
297
298 Contributions for more systems (especially non-UNIX,
299 e.g. but not limited to: Win32, VMS, OS/2) gladly accepted.
300 (also more UNIX variants welcome: HP-UX? IRIX?)
301
302 Thanks to H.Merijn Brand, John Peacock, and Craig
303 Berry for brave beta testing.
304
3051.91 [2006-09-29]
306 - ualarm() in SuSE 10.1 was overflowing after ~4.2 seconds,
307 possibly due to a glibc bug/feature (suspected overflow at
308 2**32 microseconds?), workaround by using the setitimer()
309 implementation of ualarm() if either useconds or
310 interval > 999_999 (this case seems to vary between systems:
311 are useconds more than 999_999 for ualarm() defined or not)
312 Added more ualarm() tests to catch various overflow points,
313 hopefully no problems in various platforms.
314 (The problem report by Mark Seger and Jon Paul Sullivan of HP.)
315
3161.90 [2006-08-22]
317 - tweak still needed for Const64(), from Jerry Hedden
318 - get a freshly generated ppport.h
319 - update Copyright years
320
3211.89 [2006-08-22]
322 - Const64() already appends an 'LL' (or i64), so provide LL and i64
323 forms for the IV_1E[679] (effects Win32 and Cygwin), reported by
324 Jerry Hedden.
325 - the Changes entry for 1.88 talked about [IN]V_1[679],
326 missing the 'E'.
327
3281.88 [2006-08-21]
329 - clean up the g++ warnings in HiRes.xs, all of them
330 about mixing integer and floating point, introduce
331 constants IV_1E[679] and NV_1E[679]
332
3331.87 [2006-02-13]
334 - [rt.cpan.org #17442] 'make test' frequently fails under
335 Cygwin Perl v5.8.8, reported and patched by J. R. Hedden
336 (two race condition bugs in the END block in the case the
337 main process dies before the timer process, unearthed
338 by a bug in Cygwin ualarm)
339
3401.86 [2005-12-17]
341 - HiRes.t:s/ok 32/ok 33/, from Dominic Dunlop
342 - tighten up the clock() test marginally by requiring non-negative
343 - clock_nanosleep() and clock() doc tweaks
344
3451.85 [2005-12-16]
346 - the interface to clock_nanosleep() is more natural
347 when it is like (hires) time() (instead of like nanosleep),
348 and the .xs implementation of clock_nanosleep() in 1.84
349 was broken anyway
350 - the semantics of clock() are not quite so silly as I thought,
351 but still somewhat odd, documented as such
352 - additional enhancements to the clock() documentation
353 - add test for clock_nanosleep() (I cannot test this
354 since none of my systems have the function)
355 - add test for clock()
356
3571.84 [2005-12-16]
358 - add clock() which returns the processor time in
359 (floating point) seconds since an arbitrary era
360 - add clock_nanosleep() which suspends the current
361 thread until either absolute time or for relative time
362 - [rt.cpan.org #16486] printf missing value in HiRes.t
363 - add constants CLOCKS_PER_SEC, CLOCK_SOFTTIME, TIMER_ABSTIME
364 - tiny typo fixes
365
3661.83 [2005-11-19]
367 - has_symbol() was wrong since e.g. ITIMER_VIRTUAL is exported
368 via @EXPORT_OK even when it is not available. This is heinous.
369 @EXPORT_OK should be determined at Makefile.PL time.
370 - be more lenient is testing clock_gettime(): allow more slop,
371 and retry up to three times, sleeping a random nap between
372 the retries
373 - human months are one-based (noticed by Anton Berezin)
374
3751.82 [2005-10-06]
376 - CLOCK_REALTIME is an enum value (of the clockid_t enum)
377 in HP-UX (and might be so elsewhere, too), debugged by
378 H. Merijn Brand
379 - include const-c.inc as late as possible (from Randy Kobes,
380 [rt.cpan.org #15552] to avoid undefined usleep() on Win32
381
3821.81 [2005-11-05]
383 - try to be more robust and consistent in the detection of
384 CLOCK_REALTIME and ITIMER_VIRTUAL in HiRes.t: the proper
385 way is
386
387 sub has_symbol {
388 my $symbol = shift;
389 eval 'import Time::HiRes qw($symbol)';
390 return 0 unless $@ eq '';
391 return exists ${"Time::HiRes::$symbol"};
392 }
393
394 and then use
395
396 &FOO_BAR
397
398 in the test. All these moves are needed because
399
400 1) one cannot directly do eval 'Time::HiRes::FOO_BAR'
401 because FOO_BAR might have a true value of zero
402 (or in the general case an empty string or even undef)
403
404 2) In case FOO_BAR is not available in this platform,
405 &FOO_BAR avoids the bareword warning
406
407 - wait more (1.5 seconds instead of 0.1) for the CLOCK_REALTIME test
408 but expect the 'customary' slop of 0.20 instead of 0.25
409 - fixed inside a comment HAS_POLL -> TIME_HIRES_NANOSLEEP
410 - at the end of HiRest.t tell how close we were to termination
411
4121.80 [2005-11-04]
413 - Gisle noticed a mistake (using HAS_NANOSLEEP) in 1.79
414
4151.79 [2005-11-03]
416 - try nanosleep for emulating usleep -- may help in some weird
417 embedded realtime places which have nanosleep but neither usleep
418 nor select nor poll (doesn't have to be weird embedded realtime
419 place, though -- in many places usleep is nanosleep anyway)
420 - try poll for emulating usleep -- this may help some obscure/old
421 SVR4 places that have neither usleep nor select
422 - a redundant test guard in HiRes.t
423
4241.78 [2005-11-03]
425 - ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
426 in the general case fail already at 'use' phase is suboptimal)
427 - fixes to the documentation of clock_gettime() and clock_getres()
428
4291.77 [2005-11-03]
430 - add support for the POSIX clock_gettime() and clock_getres(),
431 if available, either as library calls or as syscalls
432 - be more defensive about missing functionality: break out
433 early (during 'use') if no e.g. clock_getres() is available,
434 and protect our back by trapping those cases also in HiRes.xs
435 - the test added in 1.76 could cause an endless loop e.g. in Solaris,
436 due to mixing of sleep() and alarm() (bad programmer, no cookie!)
437
4381.76 [2005-10-22]
439 - testing for nanosleep had wrong logic which caused nanosleep
440 to become undefined for e.g. Mac OS X
441 - added a test for a core dump that was introduced by Perl 5.8.0
442 safe signals and was fixed for the time of 5.8.1 (one report of
443 the core dump was [perl #20920]), the test skipped pre-5.8.1.
444 - *cough* s/unanosleep/nanosleep/g; *cough*
445
4461.75 [2005-10-18]
447 - installation patch from Gisle Aas: in Perls 5.8.x and later
448 use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
449
4501.74 [2005-09-19]
451 - [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
452 (the workaround is not to use rel2abs, should not be necessary)
453 - [cpan #14642] U2time wrongly exported on the C API
454 (patch supplied by the reporter, SALVA@cpan.org)
455 - add release dates to Changes
456
4571.73 [2005-08-16]
458 - Time::HiRes::nanosleep support for Solaris [PATCH]
459 (POSIX::uname() not available if building with core perl,
460 from Gisle Aas, via perl5-porters, perl change #25295)
461
4621.72 [2005-07-01]
463 - going back to the 1.68 loader setup (using DynaLoader)
464 since too many weird things starting breaking
465 - fix a typo in José Auguste-Etienne's name
466
4671.71 [2005-06-28]
468 - a thinko in the nanosleep() detection
469 - move more changes stuff from the README to Changes
470 - add -w to the Makefile.PL
471
4721.70 [2005-06-26]
473 - oops in 1.69 about @ISA (not affecting anything but silly)
474 - add copyright 2005 to HiRes.pm
475 - add copyright and license to HiRes.xs
476 - add copyrights 2003, 2004, 2005 to README
477
4781.69 [2005-06-25]
479 - actually run a test for nanosleep
480 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
481 it seems that one can link in nanosleep() but then calling
482 it fails instantly and sets errno to ENOSYS (Not implemented).
483 This may be fixable in the AIX case by figuring out the right
484 (realtime POSIX?) libs and whatnot, but in the general case
485 running a real test case is better. (Of course, this change
486 will no doubt run into portability problems because of the
487 execution step...) Note that because of hysterical raisins
488 most Perls do NOT have $Config{d_nanosleep} (scanning for
489 it by Configure would in many platforms require linking in
490 things like -lrt, which would in many platforms be a bad idea
491 for Perl itself).
492 (from José Auguste-Etienne)
493 - support XSLoader also since it's much faster
494 (from Alexey Tourbin)
495 - add SEE ALSO (BSD::Resource and Time::TAI64)
496
4971.68 [2005-05-14]
498 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
499 error suspected), but miraculously it still worked since the
500 doubling took place below the __END__ token
501 - undef Pause() before defining it to avoid redefinition warnings
502 during compilation in case perl.h had already defined Pause()
503 (part of perl change #24271)
504 - minor doc tweaks
505
5061.67 [2005-05-04]
507 - (internal) don't ignore the return value of gettimeofday()
508 - (external) return undef or an empty if the C gettimeofday() fails
509 (affects Time::HiRes gettimeofday() and the hires time())
510
5111.66 [2004-12-19]
512 - add nanosleep()
513 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
514 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
515 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
516 - close also [rt.cpan.org #5933] "Time::HiRes::time does not
517 pick up time adjustments like ntp" since ever reproducing it
b88265bf 518 (and therefore verifying a possible fix) in the same environment
df003a36
JH
519 has become rather unlikely
520
5211.65 [2004-09-18]
522 - one should not mix u?alarm and sleep (the tests modified
523 by 1.65, #12 and #13, hung in Solaris), now we just busy
524 loop executing an empty block
525 - in the documentation underline the unspecificity of mixing
526 sleeps and alarms
527 - small spelling fixes
528
5291.64 [2004-09-16]
530 - regenerate ppport.h with Devel::PPPort 3.03,
531 now the MY_CXT_CLONE is defined in ppport.h,
532 we no more need to do that.
533
534 - the test #12 would often hang in sigsuspend() (at least that's
535 where Mac OS X' ktrace shows it hanging). With the sleep()s
536 changed to sleep(1)s, the tests still pass but no hang after
537 a few hundred repeats.
538
5391.63 [2004-09-01]
540 - Win32 and any ithread build: ppport.h didn't define
541 MY_CXT_CLONE, which seems to be a Time-HiRes-ism.
542
5431.62 [2004-08-31]
544 - Skip testing if under PERL_CORE and Time::HiRes has not
545 been Configured (from Marcus Holland-Moritz, core change
546 #23246)
547 - Use ppport.h generated by Devel::PPPort 3.01,
548 allowing cutting away our own portability code.
549 - Don't use $ENV{PERL_CORE} for < 5.6.0.
550 - Don't use "for my $i" for <= 5.003.
551 - Don't use Pause() for <= 5.003.
552 - Can't use newSVpvf for <= 5.003.
553 (most of the changes from Marcus)
554
5551.61 [2004-08-21]
556 - Win32: reset reading from the performance counters every
557 five minutes to better track wall clock time (thanks to
558 PC timers being often quite bad), should help long-running
559 programs.
560
5611.60 [2004-08-15]
562 - Win32: Patch from Steve Hay
563 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
564 to [perl #30755] reported by Nigel Sandever
565
566 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
567 <w32api/windows.h> APIs are available. Cygwin testing by
568 Yitzchak Scott-Thoennes.
569
570 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
571 after that keep using -lrt, patch from Alan Burlison,
572 bug reported in [cpan #7165]
573
5741.59 [2004-04-08]
575 - Change the Win32 recalibration limit to 0.5 seconds and tweak
576 the documentation to blather less about the gory details of the
577 Win32 implementation and more about the complications in general
578 of meddling with the system clock.
579
5801.58 [2004-04-08]
581 - Document the 1.57 change better.
582
5831.57 [2004-07-04]
584 - Win32/Cygwin/MinGW: if the performance counter drifts by more
585 than two seconds from the system clock (due to ntp adjustments,
586 for example), recalibrate our internal counter: from Jan Dubois,
587 based on [cpan #5933] by Jerry D. Hedden.
588
5891.56 [2004-29-02]
590 - Give a clearer message if the tests timeout (perl change #22253)
591 - Don't use /tmp or its moral equivalents (perl bug #15036,
592 perl change #22258)
593
5941.55 [2004-01-14]
595 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
596 instead of VC-specific i64 suffix)
597
5981.54 [2003-12-31]
599 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
600
6011.53 [2003-12-30]
602 - Windows: higher resolution time() by using the Windows
603 performance counter API, from Jan Dubois and Anton Shcherbinin.
604 The exact new higher resolution depends on the hardware,
605 but it should be quite a bit better than using the basic
606 Windows timers.
607
6081.52 [2003-10-28]
609 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
610 the subtest 18. No known analysis nor fix, but added
611 an alarm (that requires fork() and alarm()) to the test.
612
6131.51 [2003-09-22]
614 - doc tweaks from mjd (perl change #20456)
615 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
616
6171.50 [2003-08-02]
618 - add a message (for non-core builds) to Makefile.PL about
619 the LC_ALL=C workaround
620 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
621 - the nanosleep() probe was broken (perl change #20061)
622 - use existence instead of definedness for feature probes
623 (perl change #20043)
624 - MPE/iX tweak (perl change #20042)
625 - do not use HAS_NANOSLEEP (perl change #19898)
626
6271.49 [2003-06-23]
628 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
629 - OS/2 can always mix subsecond sleeps with signals
630 (part of perl change #19789)
631
6321.48 [2003-06-04]
633 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
634 (perl change #19592)
635
6361.47 [2003-05-03]
637 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
638 - unnecessary (nanosleep is in libc anyway)
639 - harmful (-lrt slows down execution)
640 - incompatible (with many distributions' pthreads)
641
6421.46 [2003-04-25]
643 - do not create files in blib directories under core
644 (perl change #19160, from rgs)
645 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
646
6471.45 [2003-04-01]
648 - guarantee that $xdefine in HiRes.t is always defined
649 (perl change #19109, from IlyaZ)
650 - a cleaner way to detect PERL_CORE (perl change #19111,
651 from IlyaZ)
652
6531.44 [2003-03-30]
654 - add hints/irix.pl to turn off overly POSIX flags that
655 cause hide struct timespec to be hidden (and compilation
656 to fail) (bleadperl change #19085)
657 - documentation tweaks
658
6591.43 [2003-03-11]
660 - add c:/temp to the list of temp directories to probe
661 so that cygwin (and win*?) builds are happy. This was
662 needed at least in my cygwin 1.3.20/w2k setup.
663
6641.42 [2003-01-07]
665 - modernize the constants code (from Nicholas Clark)
666
6671.41 [2003-01-03]
668 - At some point the ability to figure our the correct incdir
669 for EXTERN.h (either a core perl build, or an installed perl)
670 had broken (which lead into all test compiles failing with
671 a core perl build, but thanks to the robustness of Makefile.PL
672 nothing of this was visible). The brokenness seemed to be
673 caused by $ENV{PERL_CORE} not being on for core builds?
674 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
675 right, and both styles of build should work again.
676
6771.40 [2003-01-03]
678 - Nicholas Clark noticed that the my_catdir() emulation function
679 was broken (which means that we didn't really work for Perls
680 5.002 and 5.003)
681 - inspired by fixing the above made the whole Makefile.PL -w
682 and strict clean
683 - tightened up the Makefile.PL output, less whitespace
684
6851.39 [2003-10-20]
686 - fix from Craig Berry for better building in VMS with PERL_CORE
687
6881.38 [2003-10-13]
689 - no functional changes
690 - move lib/Time/HiRes.pm as Hires.pm
691 - libraries scanning was slightly broken (always scanned
692 for a library even when $Config{libs} already had it)
693
6941.37 [2003-09-23]
695 - Ray Zimmerman ran into a race condition in Mac OS X.
696 A 0.01-second alarm fired before the test expected.
697 The test first slept indefinitely (blocking for signals)
698 and only after that tested for the signal having been sent.
699 Since the signal had already been sent, the test #12 never
700 completed. The solution: test first, then block.
701 - default to being silent on all probing attempts, set the
702 environment variable VERBOSE to a true value to see the
703 details (the probing command and the possible errors)
704
7051.36 [2003-09-12]
706 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
707 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
708 cannot be upgraded from CPAN to override the 5.8.0 version
709 (Guido A. Ostkamp)
710 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
711 because the include directories did not adjust themselves
712 if $ENV{PERL_CORE} (Hugo van der Sanden)
713 - add documentation about the restart of select() under alarm()
714
7151.35 [2003-08-24]
716 - small documentation tweaks
717
718
7191.34 [2003-08-22]
720 - better VMS operation (Craig Berry)
721
7221.33 [2003-08-20]
723 - our time machine is accelerating: now works with Perl 5.004_01
724 (tried with 5.003_07 and 5.002 but I get segmentation faults
725 from running the Makefile.PL with those in Tru64 4.0D)
726
7271.32 [2003-08-20]
728 - backward compatibility (pre-5.6.0) tweaks:
729 - no XSLoader in 5.00503, use DynaLoader instead
730 - no SvPV_nolen, either
731 - no PerlProc_pause(), either
732 - now tested with 5.00404 and 5.00503
733 - Makefile.PL requires 5.00404 (no more 5.002)
734 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
735 (this means that one can mix subsecond sleeps with alarms)
736 - because of nanosleep we probe for -lrt and -lposix4
737 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
738 is available by exportable constants Time::HiRes::d_func
739 (since older Perl do not have them in %Config, and even
740 5.8.0 does not probe for nanosleep)
741
7421.31 [2003-08-19]
743 - backward compatibility (pre-5.6.1) tweaks:
744 - define NV if no NVTYPE
745 - define IVdf if needed (note: the Devel::PPPort
746 in 5.8.0 does not try hard hard enough since
747 the IVSIZE might not be defined)
748 - define NVgf if needed
749 - grab the typemap from 5.8.0 for the NV stuff
750
751 1.31 and 1.32 add more backward compatibility (now all the way
752 back to Perl 5.00404), and using nanosleep() (if available) for
753 subsecond sleeps.
754
7551.30 [2003-08-16]
756
757 - release 1.29_02 as 1.30
758
759 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
760 development cycle. Most notably portability across platforms has been
761 enhanced, and the interval timers (setitimer, getitimer) have been
762 added. Note that the version of Time::HiRes that is included in Perl
763 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
764 version. Note also that in 1.30 Wegscheid turns over the maintenance
765 to Jarkko Hietaniemi.
766
7671.29_02 [2003-08-16]
768
769 - fix a silly unclosed comment typo in HiRes.xs
770 - document and export REALTIME_REALPROF (Solaris)
771
7721.29_01 [2003-08-16]
773
774 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
775 (need to patch this also in Perl 5.[89])
776 - remove CVS revision log from HiRes.xs
777
7781.29_00 [2003-08-14]
779
780 The following numbered patches refer to the Perl 5.7 changes,
781 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
782
783 - 17558: Add #!./perl to the .t
784 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
785 - 16198: political correctness, from Simon Cozens
786 - 15857: doc tweaks, from Jarkko Hietaniemi
787 - 15593: optimization in .xs, from Paul Green
788 - 14892: pod fixes, from Robin Barker
789 - 14100: VOS fixes, from Paul Green
790 - 13422: XS segfault, from Marc Lehmann
791 - 13378: whether select() gets restarted on signals, depends
792 - 13354: timing constraints, again, from Andy Dougherty
793 - 13278: can't do subsecond alarms with ualarm;
794 break out early if alarms do not seem to be working
795 - 13266: test relaxation (cygwin gets lower hires
796 times than lores ones)
797 - 12846: protect against high load, from Jarkko Hietaniemi
798 - 12837: HiRes.t VMS tweak, from Craig A. Berry
799 - 12797: HiRes.t VMS tweak, from Charles Lane
800 - 12769: HiRes.t VMS tweak, from Craig A. Berry
801 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
802 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
803 - 12692: alarm() ain't gonna work if ualarm() ain't,
804 from Gurusamy Sarathy
805 - 12680: minor VMS tweak, from Charles Lane
806 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
807 - 12609: croak on negative time, from Jarkko Hietaniemi
808 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
b88265bf 809 - 12594: MacOS Classic timeofday, from Chris Nandor
df003a36
JH
810 - 12473: allow for more than one second for sleep() and usleep()
811 - 12458: test tuning, relax timing constraints,
812 from Jarkko Hietaniemi
813 - 12449: make sleep() and usleep() to return the number
814 of seconds and microseconds actually slept (analogously
815 with the builtin sleep()), also make usleep() croak if
816 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
817 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
818 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
819 - 12196: use ftime() on Win32, from Artur Bergman
820 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
821 - 12105: use GetSystemTime() on Win32, from Artur Bergman
822 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
823 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
824 - 11797: problem in HiRes.t, from John P. Linderman
825 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
826 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
b88265bf 827 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
df003a36
JH
828 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
829 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
830 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
831 from Jonathan Stowe
832 - 10942: MPE/IX test tweaks, from Mark Bixby
b88265bf 833 - 10784: unnecessary pod2man calls, from Andy Dougherty
df003a36
JH
834 - 10354: ext/ + -Wall, from Doug MacEachern
835 - 10320: fix the BOOT section to call myU2time correctly
836 - 10317: correct casting for AIX< from H. Merijn Brand
837 - 10119: document that the core time() may be rounding, not truncating
838 - 10118: test fix, from John Peacock
839 - 9988: long =item, from Robin Barker
840 - 9714: correct test output
841 - 9708: test also the scalar aspect of getitimer()
842 - 9705: Add interval timers (setitimer, getitimer)
843 - 9692: do not require at least 5.005 using XS
b88265bf 844
df003a36
JH
845 The following changes were made on top of the changes
846 made for Time::HiRes during the Perl 5.7 development
b88265bf 847 cycle that culminated in the release of Perl 5.8.0.
df003a36
JH
848
849 - add "require 5.005" to the Makefile.PL
850 - remove the REVISION section (CVS log) from HiRes.pm
851 - add jhi's copyright alongside Douglas'
852 - move HiRes.pm to lib/Time/
853 - move HiRes.t to t/
854 - modify HiRes.t to use $ENV{PERL_CORE}
855 - modify the original Time::HiRes version 1.20 Makefile.PL
856 to work both with Perl 5.8.0 and the new code with pre-5.8.0
857 Perls (tried with 5.6.1)
858 - tiny tweaks and updates in README and TODO
859 - bump the VERSION to 1.29
860
8611.20 Wed Feb 24 21:30 1999
b88265bf 862 - make our usleep and ualarm substitutes into hrt_usleep
df003a36
JH
863 and hrt_ualarm. This helps static links of Perl with other
864 packages that also have usleep, etc. From
865 Ilya Zakharevich <ilya@math.ohio-state.edu>
866 - add C API stuff. From Joshua Pritikin
867 <joshua.pritikin@db.com>
868 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
869 - hopefully correct "-lc" fix for SCO.
870 - add PPD stuff
871
872 1.20 adds a platform neutral set of C accessible routines if you are
873 running 5.005+. All other changes are packaging changes and build
874 fixes(?) for statically linked Perl, SCO, and VMS.
875
8761.19 Tue Sep 29 22:30 1998
877 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
878 <seb@stian.demon.co.uk>
879 - change GIMME_V to GIMME to help people with older versions of
880 Perl.
881 - fix Win32 version of gettimeofday(). It didn't affect anything,
882 but it confuses people reading the code when the return value
883 is backwards (0 is success).
884 - fix Makefile.PL (more) so that detection of gettimeofday is
885 more correct.
886
887 1.19 has better VMS support.
888
8891.18 Mon Jul 6 22:40 1998
890 - add usleep() for Win32.
891 - fix Makefile.PL to fix reported HP/UX feature where unresolved
892 externals still cause an executable to be generated (though no
893 x bit set). Thanks to David Kozinn for report and explanation.
894 Problems with the fix are mine :)
895
896 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
897 Probably buggy. I'm sure I'll hear.
898
8991.17 Wed Jul 1 20:10 1998
900 - fix setitimer calls so microseconds is not more than 1000000.
901 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
902 - make Win32. We only get gettimeofday (the select hack doesn't
903 seem to work on my Win95 system).
b88265bf 904 - fix test 4 on 01test.t. add test to see if time() and
df003a36
JH
905 Time::HiRes::time() are close.
906
9071.16 Wed Nov 12 21:05 1997
908 - add missing EXTEND in new gettimeofday scalar code.
909
910 1.16+ should be closer to building out of the box on Linux. Thanks
911 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
912
913 If your underlying operating system doesn't implement ualarm(), then
914 a fake using setitimer() will be made. If the OS is missing usleep(),
915 a fake one using select() will be made. If a fake can't be made for
916 either ualarm() or usleep(), then the corresponding Perl function will
917 not be available. If the OS is missing gettimeofday(), you will get
918 unresolved externals, either at link- or run-time.
919
920 This is an improvement; the package used to not even build if
921 you were missing any of these bits. Roderick Schertler
922
923 <roderick@argon.org> did all the conditional compilation stuff,
924 look at HiRes.pm and the test suites; it's good educational reading.
925
9261.15 Mon Nov 10 21:30 1997
927 - HiRes.pm: update pod. Provided by Gisle Aas.
928 - HiRes.xs: if gettimeofday() called in scalar context, do
929 something more useful than before. Provided by Gisle Aas.
930 - README: tell of xsubpp '-nolinenumber' woes. thanks to
931 Edward Henigin <ed@texas.net> for pointing out the problem.
932
9331.14 Wed Nov 5 9:40 1997
934 - Makefile.PL: look for setitimer
935 - HiRes.xs: if missing ualarm, but we have setitimer, make up
936 our own setitimer. These were provided by Gisle Aas.
937
9381.13 Tue Nov 4 23:30 1997
939 - Makefile.PL: fix autodetect mechanism to do try linking in addition
940 to just compiling; should fix Linux build problem. Fix was provided
941 by Gisle Aas.
942
9431.12 Sun Oct 12 12:00:00 1997
944 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
945 you may need to comment this back out if you have an older xsubpp.
946 - HiRes.xs: set PROTOTYPES: DISABLE
947
9481.11 Fri Sep 05 16:00:00 1997
949 - Makefile.PL:
950 Had some line commented out that shouldn't have been (testing
951 remnants)
952 - README:
953 Previous version was corrupted.
954
9551.10 Thu May 22 20:20:00 1997
956 - HiRes.xs, HiRes.pm, t/*:
b88265bf 957 - only compile what we have OS support for (or can
df003a36 958 fake with select())
b88265bf 959 - only test what we compiled
df003a36 960 - gross improvement to the test suite
b88265bf 961 - fix EXPORT_FAIL.
df003a36
JH
962 This work was all done by Roderick Schertler
963 <roderick@argon.org>. If you run Linux or
b88265bf
JH
964 one of the other ualarm-less platforms, and you like this
965 module, let Roderick know; without him, it still wouldn't
df003a36
JH
966 be working on those boxes...
967 - Makefile.PL: figure out what routines the OS has and
b88265bf 968 only build what we need. These bits were written by Jarkko
df003a36
JH
969 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
970
9711.02 Mon Dec 30 08:00:00 1996
972 - HiRes.pm: update documentation to say what to do when missing
973 ualarm() and friends.
974 - README: update to warn that ualarm() and friends need to exist
975
9761.01 Fri Oct 17 08:00:00 1996
977 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
b88265bf 978 - HiRes.pm: put blank line between __END__ and =head1 so that
df003a36
JH
979 pod2man works.
980
9811.00 Tue Sep 03 13:00:00 1996
982 - original version; created by h2xs 1.16