1 Revision history for the Perl extension Time::HiRes.
4 - As with QNX, Haiku has the API of interval timers but not
5 the implementation (bleadperl change #34630), hence skip
6 the tests, via David Mitchell.
9 - .xs code cleanup from Albert Dvornik
10 - in the #39 and #40 do not do us I did, mixing alarm() and
11 sleep(). Now instead spin until enough time has passed.
14 - Skip the tests added in 1.9716 (#39, #40) if there's no subsecond
15 alarm capability, like with the older subsecond alarm tests
18 - Change documentation to agree with reality: there are
19 no interval timers in Win32.
20 - Address [rt.cpan.org #35899] (problem in subsecond sleeps),
21 add two tests to guard against this problem
22 - Address [rt.cpan.org #36600] 'Division by zero' failure in test suite
23 - Address [rt.cpan.org #37340] [PATCH] Address timer process in test
24 - Address [rt.cpan.org#40311 ] bad implementation of hrt_usleep
25 with TIME_HIRES_NANOSLEEP
28 - Silly me: Makefile.PL does need to accept arguments other than mine.
29 Some testing frameworks obviously do this.
30 - Add retrying for tests 34..37, which are the most commonly
31 failing tests. If this helps, consider extending the retry
32 framework to all the tests. [Inspired by Slaven Rezic,
33 [rt.cpan.org #34711] Occasional failures of test 35 or 36 (FreeBSD)]
36 - Under Perl 5.6.* NVgf needs to be "g", reported by Zefram,
37 it seems that ppport.h 3.13 gets this wrong.
38 - remove the check in Makefile.PL for 5.7.2, shouldn't be
39 (a) necessary (b) relevant
40 - add logic to Makefile.PL to skip configure/write Makefile
41 step if the "xdefine" file already exists, indicating that
42 the configure step has already been done, one can still
43 force (re)configure by "perl Makefile.PL configure",
44 or of course by "make clean && perl Makefile.PL".
47 - for alarm() and ualarm() [Perl] prefer setitimer() [C]
48 instead of ualarm() [C] since ualarm() [C] cannot portably
49 (and standards-compliantly) be used for more than 999_999
50 microseconds (rt.cpan.org #34655)
51 - it seems that HP-UX has started (at least in 11.31 ia64)
52 #defining the CLOCK_REALTIME et alia (instead of having
54 - document all the diagnostics
57 - move the sub tick in the test file back to where it used to be
58 - in the "consider upgrading" message recommend at least Perl 5.8.8
59 and make the message to appear only for 5.8.0 since 5.8.1 and
60 later have the problem fixed
61 - VOS tweak for Makefile (core perl change #33259)
62 - since the test #17 seems to fail often, relax its limits a bit
65 - lost VMS test skippage from Craig Berry
66 - reformat the test code a little
69 - I got the sense of the QNX test the wrong way in an attempt
70 to generalize it for future
73 - casting fixes from Robin Barker for g++ and 64bitint
74 - in QNX skip the itimer tests because though the API
75 is there, the implementation isn't, from Matt Kraai
76 - raise the dead man timer to 180 seconds for really
78 - elaborate the UTF-8 locale warning from Makefile.PL
81 - [rt.cpan.org #29747]: Build failure with perl 5.005_05
82 Fixed by regenerating the ppport.h using Devel::PPPort 3.13.
85 - t/HiRes.t failed in Perl 5.6.2,
86 "action is not of type POSIX::SigAction at t/HiRes.t line 318",
87 reported and fixed by Anton Berezin, the reason was faulty
88 use of sigaction() when restoring the old action.
91 - with bleadperl in VMS the HiRes.t overrun the maximum number
92 of deferred signals because the libc SIGALRM was not strong
93 enough to interrupt select(), and select() got restarted every
94 time, solution is to use POSIX::SigAction if available.
95 A fix from Craig Berry (not 100% there, but helps).
96 - allow for more measuring noise for ualarm() tests 35..37
99 - nanosleep() and clock_nanosleep() detection and use were
100 quite broken; in Linux -lrt needed; fixes from Zefram
101 - [internal] slightly cleaner building of $DEFINE in Makefile.PL,
102 should avoid double/conflicting -D flags
105 - allow 10% of slop in test #14 (testing difference between
106 CORE::time() and Time::HiRes::time()), there seem to be often
107 transient failures from Perl smoke builds on this test
111 - use int main(int argc, char **argv) consistently in Makefile.PL,
113 [rt.cpan.org #23868] nanosleep not detected under Mac OS 10.3.9 starting with Time::HiRes 1.96
114 - if someone still has the locale-broken Perl 5.8.0,
115 suggest that they upgrade their Perl
118 - restore the -DATLEASTFIVEOHOHFIVE, Win32 needed it still
121 - upgrade to ppport.h 3.10_02
122 - remove the -DATLEASTFIVEOHOHFIVE
123 - use the ppport.h PL_ppaddr, PL_statcache, PL_laststatval
124 - use the ppport.h aTHXR for calling Perl stat()
125 - switch into four-digit version since 2.0 is coming up
126 awfully fast but not feeling like a major rewrite
129 - 1.95 broke building in Win32 (since pp_stat is not exported),
130 figured out how to call an op directly in 5.005 (use Perl_ppaddr
131 instead of PL_ppaddr)
132 - backport to Perl 5.004_05 (requires using statcache
133 and laststatval instead of PL_statcache and PL_laststatval)
134 (also checked to work in 5.005_04, 5.6.1, and 5.8.8 with threads)
137 - 1.95 broke builds for threaded Perls, rt.cpan.org tickets:
138 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
139 [rt.cpan.org #23712] Time-HiRes 1.95 Fails make on AIX 5.2 with Perl 5.8.8
140 [rt.cpan.org #23730] Time::HiRes 1.95 fails make on MacOS X 10.3.9/perl 5.8.8
141 - use main() prototype consistently in Makefile.PL
144 - integrate core change #29180: Silence VC++ compiler warnings
146 - do not use PL_ppaddr in stat() because that is not available
148 - regenerate fallback/*.inc for older Perls without
149 ExtUtils::Constant because of d_hires_stat, resolves
150 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
151 - Make Makefile.PL more defensive against false PERL_CORE
154 - file timestamps oddities seen: the atime and mtime
155 can be out of sync (modify first and read second can leave
156 atime < mtime) and mtime can be subsecond while atime is not.
157 So make the test more forgiving.
160 - the ualarm() tests (34-37) assumed that ualarm(N)
161 could never alarm in less than N seconds, widened
162 the acceptable relative range to 0.9..1.5. Addresses
163 [rt.cpan.org #22090] and [rt.cpan.org #22091].
165 - skip the stat() tests in cygwin and win32, because
166 if run on FAT the timestamp granularity is only 2 seconds.
167 Any good way to detect (cygwin or win32) whether we are
168 being run on NTFS or anywhere with better timestamps?
169 Addresses [rt.cpan.org #22089] and [rt.cpan.org #22098].
172 - scan for subsecond resolution timestamps in struct stat,
173 some known possibilities:
175 (1) struct timespec st_atimespec;
176 st_atimespec.tv_nsec;
181 (4) timestruc_t st_atim;
186 If something like this is found, one can do
189 my @stat = Time::HiRes::stat();
191 or even override the standard stat():
193 use Time::HiRes qw(stat);
195 to get the stat() timestamps
197 my ($atime, $mtime, $ctime) = @stat[8, 9, 10];
199 with subsecond resolution (assuming both the operating
200 system and the filesystem support that kind of thing).
202 Contributions for more systems (especially non-UNIX,
203 e.g. but not limited to: Win32, VMS, OS/2) gladly accepted.
204 (also more UNIX variants welcome: HP-UX? IRIX?)
206 Thanks to H.Merijn Brand, John Peacock, and Craig
207 Berry for brave beta testing.
210 - ualarm() in SuSE 10.1 was overflowing after ~4.2 seconds,
211 possibly due to a glibc bug/feature (suspected overflow at
212 2**32 microseconds?), workaround by using the setitimer()
213 implementation of ualarm() if either useconds or
214 interval > 999_999 (this case seems to vary between systems:
215 are useconds more than 999_999 for ualarm() defined or not)
216 Added more ualarm() tests to catch various overflow points,
217 hopefully no problems in various platforms.
218 (The problem report by Mark Seger and Jon Paul Sullivan of HP.)
221 - tweak still needed for Const64(), from Jerry Hedden
222 - get a freshly generated ppport.h
223 - update Copyright years
226 - Const64() already appends an 'LL' (or i64), so provide LL and i64
227 forms for the IV_1E[679] (effects Win32 and Cygwin), reported by
229 - the Changes entry for 1.88 talked about [IN]V_1[679],
233 - clean up the g++ warnings in HiRes.xs, all of them
234 about mixing integer and floating point, introduce
235 constants IV_1E[679] and NV_1E[679]
238 - [rt.cpan.org #17442] 'make test' frequently fails under
239 Cygwin Perl v5.8.8, reported and patched by J. R. Hedden
240 (two race condition bugs in the END block in the case the
241 main process dies before the timer process, unearthed
242 by a bug in Cygwin ualarm)
245 - HiRes.t:s/ok 32/ok 33/, from Dominic Dunlop
246 - tighten up the clock() test marginally by requiring non-negative
247 - clock_nanosleep() and clock() doc tweaks
250 - the interface to clock_nanosleep() is more natural
251 when it is like (hires) time() (instead of like nanosleep),
252 and the .xs implementation of clock_nanosleep() in 1.84
254 - the semantics of clock() are not quite so silly as I thought,
255 but still somewhat odd, documented as such
256 - additional enhancements to the clock() documentation
257 - add test for clock_nanosleep() (I cannot test this
258 since none of my systems have the function)
259 - add test for clock()
262 - add clock() which returns the processor time in
263 (floating point) seconds since an arbitrary era
264 - add clock_nanosleep() which suspends the current
265 thread until either absolute time or for relative time
266 - [rt.cpan.org #16486] printf missing value in HiRes.t
267 - add constants CLOCKS_PER_SEC, CLOCK_SOFTTIME, TIMER_ABSTIME
271 - has_symbol() was wrong since e.g. ITIMER_VIRTUAL is exported
272 via @EXPORT_OK even when it is not available. This is heinous.
273 @EXPORT_OK should be determined at Makefile.PL time.
274 - be more lenient is testing clock_gettime(): allow more slop,
275 and retry up to three times, sleeping a random nap between
277 - human months are one-based (noticed by Anton Berezin)
280 - CLOCK_REALTIME is an enum value (of the clockid_t enum)
281 in HP-UX (and might be so elsewhere, too), debugged by
283 - include const-c.inc as late as possible (from Randy Kobes,
284 [rt.cpan.org #15552] to avoid undefined usleep() on Win32
287 - try to be more robust and consistent in the detection of
288 CLOCK_REALTIME and ITIMER_VIRTUAL in HiRes.t: the proper
293 eval 'import Time::HiRes qw($symbol)';
294 return 0 unless $@ eq '';
295 return exists ${"Time::HiRes::$symbol"};
302 in the test. All these moves are needed because
304 1) one cannot directly do eval 'Time::HiRes::FOO_BAR'
305 because FOO_BAR might have a true value of zero
306 (or in the general case an empty string or even undef)
308 2) In case FOO_BAR is not available in this platform,
309 &FOO_BAR avoids the bareword warning
311 - wait more (1.5 seconds instead of 0.1) for the CLOCK_REALTIME test
312 but expect the 'customary' slop of 0.20 instead of 0.25
313 - fixed inside a comment HAS_POLL -> TIME_HIRES_NANOSLEEP
314 - at the end of HiRest.t tell how close we were to termination
317 - Gisle noticed a mistake (using HAS_NANOSLEEP) in 1.79
320 - try nanosleep for emulating usleep -- may help in some weird
321 embedded realtime places which have nanosleep but neither usleep
322 nor select nor poll (doesn't have to be weird embedded realtime
323 place, though -- in many places usleep is nanosleep anyway)
324 - try poll for emulating usleep -- this may help some obscure/old
325 SVR4 places that have neither usleep nor select
326 - a redundant test guard in HiRes.t
329 - ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
330 in the general case fail already at 'use' phase is suboptimal)
331 - fixes to the documentation of clock_gettime() and clock_getres()
334 - add support for the POSIX clock_gettime() and clock_getres(),
335 if available, either as library calls or as syscalls
336 - be more defensive about missing functionality: break out
337 early (during 'use') if no e.g. clock_getres() is available,
338 and protect our back by trapping those cases also in HiRes.xs
339 - the test added in 1.76 could cause an endless loop e.g. in Solaris,
340 due to mixing of sleep() and alarm() (bad programmer, no cookie!)
343 - testing for nanosleep had wrong logic which caused nanosleep
344 to become undefined for e.g. Mac OS X
345 - added a test for a core dump that was introduced by Perl 5.8.0
346 safe signals and was fixed for the time of 5.8.1 (one report of
347 the core dump was [perl #20920]), the test skipped pre-5.8.1.
348 - *cough* s/unanosleep/nanosleep/g; *cough*
351 - installation patch from Gisle Aas: in Perls 5.8.x and later
352 use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
355 - [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
356 (the workaround is not to use rel2abs, should not be necessary)
357 - [cpan #14642] U2time wrongly exported on the C API
358 (patch supplied by the reporter, SALVA@cpan.org)
359 - add release dates to Changes
362 - Time::HiRes::nanosleep support for Solaris [PATCH]
363 (POSIX::uname() not available if building with core perl,
364 from Gisle Aas, via perl5-porters, perl change #25295)
367 - going back to the 1.68 loader setup (using DynaLoader)
368 since too many weird things starting breaking
369 - fix a typo in José Auguste-Etienne's name
372 - a thinko in the nanosleep() detection
373 - move more changes stuff from the README to Changes
374 - add -w to the Makefile.PL
377 - oops in 1.69 about @ISA (not affecting anything but silly)
378 - add copyright 2005 to HiRes.pm
379 - add copyright and license to HiRes.xs
380 - add copyrights 2003, 2004, 2005 to README
383 - actually run a test for nanosleep
384 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
385 it seems that one can link in nanosleep() but then calling
386 it fails instantly and sets errno to ENOSYS (Not implemented).
387 This may be fixable in the AIX case by figuring out the right
388 (realtime POSIX?) libs and whatnot, but in the general case
389 running a real test case is better. (Of course, this change
390 will no doubt run into portability problems because of the
391 execution step...) Note that because of hysterical raisins
392 most Perls do NOT have $Config{d_nanosleep} (scanning for
393 it by Configure would in many platforms require linking in
394 things like -lrt, which would in many platforms be a bad idea
396 (from José Auguste-Etienne)
397 - support XSLoader also since it's much faster
398 (from Alexey Tourbin)
399 - add SEE ALSO (BSD::Resource and Time::TAI64)
402 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
403 error suspected), but miraculously it still worked since the
404 doubling took place below the __END__ token
405 - undef Pause() before defining it to avoid redefinition warnings
406 during compilation in case perl.h had already defined Pause()
407 (part of perl change #24271)
411 - (internal) don't ignore the return value of gettimeofday()
412 - (external) return undef or an empty if the C gettimeofday() fails
413 (affects Time::HiRes gettimeofday() and the hires time())
417 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
418 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
419 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
420 - close also [rt.cpan.org #5933] "Time::HiRes::time does not
421 pick up time adjustments like ntp" since ever reproducing it
422 (and therefore verifying a possible fix) in the same environment
423 has become rather unlikely
426 - one should not mix u?alarm and sleep (the tests modified
427 by 1.65, #12 and #13, hung in Solaris), now we just busy
428 loop executing an empty block
429 - in the documentation underline the unspecificity of mixing
431 - small spelling fixes
434 - regenerate ppport.h with Devel::PPPort 3.03,
435 now the MY_CXT_CLONE is defined in ppport.h,
436 we no more need to do that.
438 - the test #12 would often hang in sigsuspend() (at least that's
439 where Mac OS X' ktrace shows it hanging). With the sleep()s
440 changed to sleep(1)s, the tests still pass but no hang after
441 a few hundred repeats.
444 - Win32 and any ithread build: ppport.h didn't define
445 MY_CXT_CLONE, which seems to be a Time-HiRes-ism.
448 - Skip testing if under PERL_CORE and Time::HiRes has not
449 been Configured (from Marcus Holland-Moritz, core change
451 - Use ppport.h generated by Devel::PPPort 3.01,
452 allowing cutting away our own portability code.
453 - Don't use $ENV{PERL_CORE} for < 5.6.0.
454 - Don't use "for my $i" for <= 5.003.
455 - Don't use Pause() for <= 5.003.
456 - Can't use newSVpvf for <= 5.003.
457 (most of the changes from Marcus)
460 - Win32: reset reading from the performance counters every
461 five minutes to better track wall clock time (thanks to
462 PC timers being often quite bad), should help long-running
466 - Win32: Patch from Steve Hay
467 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
468 to [perl #30755] reported by Nigel Sandever
470 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
471 <w32api/windows.h> APIs are available. Cygwin testing by
472 Yitzchak Scott-Thoennes.
474 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
475 after that keep using -lrt, patch from Alan Burlison,
476 bug reported in [cpan #7165]
479 - Change the Win32 recalibration limit to 0.5 seconds and tweak
480 the documentation to blather less about the gory details of the
481 Win32 implementation and more about the complications in general
482 of meddling with the system clock.
485 - Document the 1.57 change better.
488 - Win32/Cygwin/MinGW: if the performance counter drifts by more
489 than two seconds from the system clock (due to ntp adjustments,
490 for example), recalibrate our internal counter: from Jan Dubois,
491 based on [cpan #5933] by Jerry D. Hedden.
494 - Give a clearer message if the tests timeout (perl change #22253)
495 - Don't use /tmp or its moral equivalents (perl bug #15036,
499 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
500 instead of VC-specific i64 suffix)
503 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
506 - Windows: higher resolution time() by using the Windows
507 performance counter API, from Jan Dubois and Anton Shcherbinin.
508 The exact new higher resolution depends on the hardware,
509 but it should be quite a bit better than using the basic
513 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
514 the subtest 18. No known analysis nor fix, but added
515 an alarm (that requires fork() and alarm()) to the test.
518 - doc tweaks from mjd (perl change #20456)
519 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
522 - add a message (for non-core builds) to Makefile.PL about
523 the LC_ALL=C workaround
524 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
525 - the nanosleep() probe was broken (perl change #20061)
526 - use existence instead of definedness for feature probes
528 - MPE/iX tweak (perl change #20042)
529 - do not use HAS_NANOSLEEP (perl change #19898)
532 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
533 - OS/2 can always mix subsecond sleeps with signals
534 (part of perl change #19789)
537 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
541 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
542 - unnecessary (nanosleep is in libc anyway)
543 - harmful (-lrt slows down execution)
544 - incompatible (with many distributions' pthreads)
547 - do not create files in blib directories under core
548 (perl change #19160, from rgs)
549 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
552 - guarantee that $xdefine in HiRes.t is always defined
553 (perl change #19109, from IlyaZ)
554 - a cleaner way to detect PERL_CORE (perl change #19111,
558 - add hints/irix.pl to turn off overly POSIX flags that
559 cause hide struct timespec to be hidden (and compilation
560 to fail) (bleadperl change #19085)
561 - documentation tweaks
564 - add c:/temp to the list of temp directories to probe
565 so that cygwin (and win*?) builds are happy. This was
566 needed at least in my cygwin 1.3.20/w2k setup.
569 - modernize the constants code (from Nicholas Clark)
572 - At some point the ability to figure our the correct incdir
573 for EXTERN.h (either a core perl build, or an installed perl)
574 had broken (which lead into all test compiles failing with
575 a core perl build, but thanks to the robustness of Makefile.PL
576 nothing of this was visible). The brokenness seemed to be
577 caused by $ENV{PERL_CORE} not being on for core builds?
578 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
579 right, and both styles of build should work again.
582 - Nicholas Clark noticed that the my_catdir() emulation function
583 was broken (which means that we didn't really work for Perls
585 - inspired by fixing the above made the whole Makefile.PL -w
587 - tightened up the Makefile.PL output, less whitespace
590 - fix from Craig Berry for better building in VMS with PERL_CORE
593 - no functional changes
594 - move lib/Time/HiRes.pm as Hires.pm
595 - libraries scanning was slightly broken (always scanned
596 for a library even when $Config{libs} already had it)
599 - Ray Zimmerman ran into a race condition in Mac OS X.
600 A 0.01-second alarm fired before the test expected.
601 The test first slept indefinitely (blocking for signals)
602 and only after that tested for the signal having been sent.
603 Since the signal had already been sent, the test #12 never
604 completed. The solution: test first, then block.
605 - default to being silent on all probing attempts, set the
606 environment variable VERBOSE to a true value to see the
607 details (the probing command and the possible errors)
610 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
611 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
612 cannot be upgraded from CPAN to override the 5.8.0 version
614 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
615 because the include directories did not adjust themselves
616 if $ENV{PERL_CORE} (Hugo van der Sanden)
617 - add documentation about the restart of select() under alarm()
620 - small documentation tweaks
624 - better VMS operation (Craig Berry)
627 - our time machine is accelerating: now works with Perl 5.004_01
628 (tried with 5.003_07 and 5.002 but I get segmentation faults
629 from running the Makefile.PL with those in Tru64 4.0D)
632 - backward compatibility (pre-5.6.0) tweaks:
633 - no XSLoader in 5.00503, use DynaLoader instead
634 - no SvPV_nolen, either
635 - no PerlProc_pause(), either
636 - now tested with 5.00404 and 5.00503
637 - Makefile.PL requires 5.00404 (no more 5.002)
638 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
639 (this means that one can mix subsecond sleeps with alarms)
640 - because of nanosleep we probe for -lrt and -lposix4
641 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
642 is available by exportable constants Time::HiRes::d_func
643 (since older Perl do not have them in %Config, and even
644 5.8.0 does not probe for nanosleep)
647 - backward compatibility (pre-5.6.1) tweaks:
648 - define NV if no NVTYPE
649 - define IVdf if needed (note: the Devel::PPPort
650 in 5.8.0 does not try hard hard enough since
651 the IVSIZE might not be defined)
652 - define NVgf if needed
653 - grab the typemap from 5.8.0 for the NV stuff
655 1.31 and 1.32 add more backward compatibility (now all the way
656 back to Perl 5.00404), and using nanosleep() (if available) for
661 - release 1.29_02 as 1.30
663 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
664 development cycle. Most notably portability across platforms has been
665 enhanced, and the interval timers (setitimer, getitimer) have been
666 added. Note that the version of Time::HiRes that is included in Perl
667 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
668 version. Note also that in 1.30 Wegscheid turns over the maintenance
669 to Jarkko Hietaniemi.
673 - fix a silly unclosed comment typo in HiRes.xs
674 - document and export REALTIME_REALPROF (Solaris)
678 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
679 (need to patch this also in Perl 5.[89])
680 - remove CVS revision log from HiRes.xs
684 The following numbered patches refer to the Perl 5.7 changes,
685 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
687 - 17558: Add #!./perl to the .t
688 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
689 - 16198: political correctness, from Simon Cozens
690 - 15857: doc tweaks, from Jarkko Hietaniemi
691 - 15593: optimization in .xs, from Paul Green
692 - 14892: pod fixes, from Robin Barker
693 - 14100: VOS fixes, from Paul Green
694 - 13422: XS segfault, from Marc Lehmann
695 - 13378: whether select() gets restarted on signals, depends
696 - 13354: timing constraints, again, from Andy Dougherty
697 - 13278: can't do subsecond alarms with ualarm;
698 break out early if alarms do not seem to be working
699 - 13266: test relaxation (cygwin gets lower hires
700 times than lores ones)
701 - 12846: protect against high load, from Jarkko Hietaniemi
702 - 12837: HiRes.t VMS tweak, from Craig A. Berry
703 - 12797: HiRes.t VMS tweak, from Charles Lane
704 - 12769: HiRes.t VMS tweak, from Craig A. Berry
705 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
706 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
707 - 12692: alarm() ain't gonna work if ualarm() ain't,
708 from Gurusamy Sarathy
709 - 12680: minor VMS tweak, from Charles Lane
710 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
711 - 12609: croak on negative time, from Jarkko Hietaniemi
712 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
713 - 12594: MacOS Classic timeofday, from Chris Nandor
714 - 12473: allow for more than one second for sleep() and usleep()
715 - 12458: test tuning, relax timing constraints,
716 from Jarkko Hietaniemi
717 - 12449: make sleep() and usleep() to return the number
718 of seconds and microseconds actually slept (analogously
719 with the builtin sleep()), also make usleep() croak if
720 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
721 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
722 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
723 - 12196: use ftime() on Win32, from Artur Bergman
724 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
725 - 12105: use GetSystemTime() on Win32, from Artur Bergman
726 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
727 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
728 - 11797: problem in HiRes.t, from John P. Linderman
729 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
730 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
731 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
732 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
733 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
734 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
736 - 10942: MPE/IX test tweaks, from Mark Bixby
737 - 10784: unnecessary pod2man calls, from Andy Dougherty
738 - 10354: ext/ + -Wall, from Doug MacEachern
739 - 10320: fix the BOOT section to call myU2time correctly
740 - 10317: correct casting for AIX< from H. Merijn Brand
741 - 10119: document that the core time() may be rounding, not truncating
742 - 10118: test fix, from John Peacock
743 - 9988: long =item, from Robin Barker
744 - 9714: correct test output
745 - 9708: test also the scalar aspect of getitimer()
746 - 9705: Add interval timers (setitimer, getitimer)
747 - 9692: do not require at least 5.005 using XS
749 The following changes were made on top of the changes
750 made for Time::HiRes during the Perl 5.7 development
751 cycle that culminated in the release of Perl 5.8.0.
753 - add "require 5.005" to the Makefile.PL
754 - remove the REVISION section (CVS log) from HiRes.pm
755 - add jhi's copyright alongside Douglas'
756 - move HiRes.pm to lib/Time/
758 - modify HiRes.t to use $ENV{PERL_CORE}
759 - modify the original Time::HiRes version 1.20 Makefile.PL
760 to work both with Perl 5.8.0 and the new code with pre-5.8.0
761 Perls (tried with 5.6.1)
762 - tiny tweaks and updates in README and TODO
763 - bump the VERSION to 1.29
765 1.20 Wed Feb 24 21:30 1999
766 - make our usleep and ualarm substitutes into hrt_usleep
767 and hrt_ualarm. This helps static links of Perl with other
768 packages that also have usleep, etc. From
769 Ilya Zakharevich <ilya@math.ohio-state.edu>
770 - add C API stuff. From Joshua Pritikin
771 <joshua.pritikin@db.com>
772 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
773 - hopefully correct "-lc" fix for SCO.
776 1.20 adds a platform neutral set of C accessible routines if you are
777 running 5.005+. All other changes are packaging changes and build
778 fixes(?) for statically linked Perl, SCO, and VMS.
780 1.19 Tue Sep 29 22:30 1998
781 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
782 <seb@stian.demon.co.uk>
783 - change GIMME_V to GIMME to help people with older versions of
785 - fix Win32 version of gettimeofday(). It didn't affect anything,
786 but it confuses people reading the code when the return value
787 is backwards (0 is success).
788 - fix Makefile.PL (more) so that detection of gettimeofday is
791 1.19 has better VMS support.
793 1.18 Mon Jul 6 22:40 1998
794 - add usleep() for Win32.
795 - fix Makefile.PL to fix reported HP/UX feature where unresolved
796 externals still cause an executable to be generated (though no
797 x bit set). Thanks to David Kozinn for report and explanation.
798 Problems with the fix are mine :)
800 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
801 Probably buggy. I'm sure I'll hear.
803 1.17 Wed Jul 1 20:10 1998
804 - fix setitimer calls so microseconds is not more than 1000000.
805 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
806 - make Win32. We only get gettimeofday (the select hack doesn't
807 seem to work on my Win95 system).
808 - fix test 4 on 01test.t. add test to see if time() and
809 Time::HiRes::time() are close.
811 1.16 Wed Nov 12 21:05 1997
812 - add missing EXTEND in new gettimeofday scalar code.
814 1.16+ should be closer to building out of the box on Linux. Thanks
815 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
817 If your underlying operating system doesn't implement ualarm(), then
818 a fake using setitimer() will be made. If the OS is missing usleep(),
819 a fake one using select() will be made. If a fake can't be made for
820 either ualarm() or usleep(), then the corresponding Perl function will
821 not be available. If the OS is missing gettimeofday(), you will get
822 unresolved externals, either at link- or run-time.
824 This is an improvement; the package used to not even build if
825 you were missing any of these bits. Roderick Schertler
827 <roderick@argon.org> did all the conditional compilation stuff,
828 look at HiRes.pm and the test suites; it's good educational reading.
830 1.15 Mon Nov 10 21:30 1997
831 - HiRes.pm: update pod. Provided by Gisle Aas.
832 - HiRes.xs: if gettimeofday() called in scalar context, do
833 something more useful than before. Provided by Gisle Aas.
834 - README: tell of xsubpp '-nolinenumber' woes. thanks to
835 Edward Henigin <ed@texas.net> for pointing out the problem.
837 1.14 Wed Nov 5 9:40 1997
838 - Makefile.PL: look for setitimer
839 - HiRes.xs: if missing ualarm, but we have setitimer, make up
840 our own setitimer. These were provided by Gisle Aas.
842 1.13 Tue Nov 4 23:30 1997
843 - Makefile.PL: fix autodetect mechanism to do try linking in addition
844 to just compiling; should fix Linux build problem. Fix was provided
847 1.12 Sun Oct 12 12:00:00 1997
848 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
849 you may need to comment this back out if you have an older xsubpp.
850 - HiRes.xs: set PROTOTYPES: DISABLE
852 1.11 Fri Sep 05 16:00:00 1997
854 Had some line commented out that shouldn't have been (testing
857 Previous version was corrupted.
859 1.10 Thu May 22 20:20:00 1997
860 - HiRes.xs, HiRes.pm, t/*:
861 - only compile what we have OS support for (or can
863 - only test what we compiled
864 - gross improvement to the test suite
866 This work was all done by Roderick Schertler
867 <roderick@argon.org>. If you run Linux or
868 one of the other ualarm-less platforms, and you like this
869 module, let Roderick know; without him, it still wouldn't
870 be working on those boxes...
871 - Makefile.PL: figure out what routines the OS has and
872 only build what we need. These bits were written by Jarkko
873 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
875 1.02 Mon Dec 30 08:00:00 1996
876 - HiRes.pm: update documentation to say what to do when missing
877 ualarm() and friends.
878 - README: update to warn that ualarm() and friends need to exist
880 1.01 Fri Oct 17 08:00:00 1996
881 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
882 - HiRes.pm: put blank line between __END__ and =head1 so that
885 1.00 Tue Sep 03 13:00:00 1996
886 - original version; created by h2xs 1.16