1 Revision history for the Perl extension Time::HiRes.
4 - Correct stack discipline in stat(), which was screwing up list
5 operations in expressions containing calls to it [rt.cpan.org
7 - Add missing OUTPUT sections to the XS code [rt.cpan.org #70930].
8 - Skip itimer tests on GNU/Hurd, which has the API but lacks
9 the implementation [rt.cpan.org #72754].
10 - Fix a doubled word in the documentation [rt.cpan.org #72763].
13 - Correct XS parameter list, and therefore prototype, for
14 unimplemented-on-this-platform version of clock_nanosleep()
16 - Declare package variables with "our" rather than "use vars".
17 - Corresponding to "our" usage, check for minimum Perl version
19 - Declare module dependencies.
22 - Remove $ENV{PERL_CORE} logic from test suite, which is no
23 longer desired in the core.
24 - Convert test suite to use Test::More.
25 - Factor out watchdog code from test suite.
26 - In test suite, be consistent about using fully-qualified form
28 - Divide test suite into feature-specific scripts.
29 - Make ualarm timing test less vulnerable to delay-induced false
30 failure, from Dave Mitchell.
33 - Update for changes in build process in the core, patches
34 from BinGOs [rt.cpan.org #58858] and Craig Berry [rt.cpan.org
36 - Fix broken linkage on Windows with gcc 3.4 seen with ActivePerl,
37 report from Christian Walde [rt.cpan.org #61648], fix derived
39 - Jump through hoops to avoid compiler warnings.
42 - Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
43 additional fix from Gisle Aas
44 - Address [rt.cpan.org #55665] "Bad plan" on Windows,
45 report and fix from Jan Dubois
48 - Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
49 report and fix from Nicholas Clark
52 - As with QNX, Haiku has the API of interval timers but not
53 the implementation (bleadperl change #34630), hence skip
54 the tests, via David Mitchell.
57 - .xs code cleanup from Albert Dvornik
58 - in the #39 and #40 do not do us I did, mixing alarm() and
59 sleep(). Now instead spin until enough time has passed.
62 - Skip the tests added in 1.9716 (#39, #40) if there's no subsecond
63 alarm capability, like with the older subsecond alarm tests
66 - Change documentation to agree with reality: there are
67 no interval timers in Win32.
68 - Address [rt.cpan.org #35899] (problem in subsecond sleeps),
69 add two tests to guard against this problem
70 - Address [rt.cpan.org #36600] 'Division by zero' failure in test suite
71 - Address [rt.cpan.org #37340] [PATCH] Address timer process in test
72 - Address [rt.cpan.org#40311 ] bad implementation of hrt_usleep
73 with TIME_HIRES_NANOSLEEP
76 - Silly me: Makefile.PL does need to accept arguments other than mine.
77 Some testing frameworks obviously do this.
78 - Add retrying for tests 34..37, which are the most commonly
79 failing tests. If this helps, consider extending the retry
80 framework to all the tests. [Inspired by Slaven Rezic,
81 [rt.cpan.org #34711] Occasional failures of test 35 or 36 (FreeBSD)]
84 - Under Perl 5.6.* NVgf needs to be "g", reported by Zefram,
85 it seems that ppport.h 3.13 gets this wrong.
86 - remove the check in Makefile.PL for 5.7.2, shouldn't be
87 (a) necessary (b) relevant
88 - add logic to Makefile.PL to skip configure/write Makefile
89 step if the "xdefine" file already exists, indicating that
90 the configure step has already been done, one can still
91 force (re)configure by "perl Makefile.PL configure",
92 or of course by "make clean && perl Makefile.PL".
95 - for alarm() and ualarm() [Perl] prefer setitimer() [C]
96 instead of ualarm() [C] since ualarm() [C] cannot portably
97 (and standards-compliantly) be used for more than 999_999
98 microseconds (rt.cpan.org #34655)
99 - it seems that HP-UX has started (at least in 11.31 ia64)
100 #defining the CLOCK_REALTIME et alia (instead of having
102 - document all the diagnostics
105 - move the sub tick in the test file back to where it used to be
106 - in the "consider upgrading" message recommend at least Perl 5.8.8
107 and make the message to appear only for 5.8.0 since 5.8.1 and
108 later have the problem fixed
109 - VOS tweak for Makefile (core perl change #33259)
110 - since the test #17 seems to fail often, relax its limits a bit
113 - lost VMS test skippage from Craig Berry
114 - reformat the test code a little
117 - I got the sense of the QNX test the wrong way in an attempt
118 to generalize it for future
121 - casting fixes from Robin Barker for g++ and 64bitint
122 - in QNX skip the itimer tests because though the API
123 is there, the implementation isn't, from Matt Kraai
124 - raise the dead man timer to 180 seconds for really
126 - elaborate the UTF-8 locale warning from Makefile.PL
129 - [rt.cpan.org #29747]: Build failure with perl 5.005_05
130 Fixed by regenerating the ppport.h using Devel::PPPort 3.13.
133 - t/HiRes.t failed in Perl 5.6.2,
134 "action is not of type POSIX::SigAction at t/HiRes.t line 318",
135 reported and fixed by Anton Berezin, the reason was faulty
136 use of sigaction() when restoring the old action.
139 - with bleadperl in VMS the HiRes.t overrun the maximum number
140 of deferred signals because the libc SIGALRM was not strong
141 enough to interrupt select(), and select() got restarted every
142 time, solution is to use POSIX::SigAction if available.
143 A fix from Craig Berry (not 100% there, but helps).
144 - allow for more measuring noise for ualarm() tests 35..37
147 - nanosleep() and clock_nanosleep() detection and use were
148 quite broken; in Linux -lrt needed; fixes from Zefram
149 - [internal] slightly cleaner building of $DEFINE in Makefile.PL,
150 should avoid double/conflicting -D flags
153 - allow 10% of slop in test #14 (testing difference between
154 CORE::time() and Time::HiRes::time()), there seem to be often
155 transient failures from Perl smoke builds on this test
159 - use int main(int argc, char **argv) consistently in Makefile.PL,
161 [rt.cpan.org #23868] nanosleep not detected under Mac OS 10.3.9 starting with Time::HiRes 1.96
162 - if someone still has the locale-broken Perl 5.8.0,
163 suggest that they upgrade their Perl
166 - restore the -DATLEASTFIVEOHOHFIVE, Win32 needed it still
169 - upgrade to ppport.h 3.10_02
170 - remove the -DATLEASTFIVEOHOHFIVE
171 - use the ppport.h PL_ppaddr, PL_statcache, PL_laststatval
172 - use the ppport.h aTHXR for calling Perl stat()
173 - switch into four-digit version since 2.0 is coming up
174 awfully fast but not feeling like a major rewrite
177 - 1.95 broke building in Win32 (since pp_stat is not exported),
178 figured out how to call an op directly in 5.005 (use Perl_ppaddr
179 instead of PL_ppaddr)
180 - backport to Perl 5.004_05 (requires using statcache
181 and laststatval instead of PL_statcache and PL_laststatval)
182 (also checked to work in 5.005_04, 5.6.1, and 5.8.8 with threads)
185 - 1.95 broke builds for threaded Perls, rt.cpan.org tickets:
186 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
187 [rt.cpan.org #23712] Time-HiRes 1.95 Fails make on AIX 5.2 with Perl 5.8.8
188 [rt.cpan.org #23730] Time::HiRes 1.95 fails make on MacOS X 10.3.9/perl 5.8.8
189 - use main() prototype consistently in Makefile.PL
192 - integrate core change #29180: Silence VC++ compiler warnings
194 - do not use PL_ppaddr in stat() because that is not available
196 - regenerate fallback/*.inc for older Perls without
197 ExtUtils::Constant because of d_hires_stat, resolves
198 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
199 - Make Makefile.PL more defensive against false PERL_CORE
202 - file timestamps oddities seen: the atime and mtime
203 can be out of sync (modify first and read second can leave
204 atime < mtime) and mtime can be subsecond while atime is not.
205 So make the test more forgiving.
208 - the ualarm() tests (34-37) assumed that ualarm(N)
209 could never alarm in less than N seconds, widened
210 the acceptable relative range to 0.9..1.5. Addresses
211 [rt.cpan.org #22090] and [rt.cpan.org #22091].
213 - skip the stat() tests in cygwin and win32, because
214 if run on FAT the timestamp granularity is only 2 seconds.
215 Any good way to detect (cygwin or win32) whether we are
216 being run on NTFS or anywhere with better timestamps?
217 Addresses [rt.cpan.org #22089] and [rt.cpan.org #22098].
220 - scan for subsecond resolution timestamps in struct stat,
221 some known possibilities:
223 (1) struct timespec st_atimespec;
224 st_atimespec.tv_nsec;
229 (4) timestruc_t st_atim;
234 If something like this is found, one can do
237 my @stat = Time::HiRes::stat();
239 or even override the standard stat():
241 use Time::HiRes qw(stat);
243 to get the stat() timestamps
245 my ($atime, $mtime, $ctime) = @stat[8, 9, 10];
247 with subsecond resolution (assuming both the operating
248 system and the filesystem support that kind of thing).
250 Contributions for more systems (especially non-UNIX,
251 e.g. but not limited to: Win32, VMS, OS/2) gladly accepted.
252 (also more UNIX variants welcome: HP-UX? IRIX?)
254 Thanks to H.Merijn Brand, John Peacock, and Craig
255 Berry for brave beta testing.
258 - ualarm() in SuSE 10.1 was overflowing after ~4.2 seconds,
259 possibly due to a glibc bug/feature (suspected overflow at
260 2**32 microseconds?), workaround by using the setitimer()
261 implementation of ualarm() if either useconds or
262 interval > 999_999 (this case seems to vary between systems:
263 are useconds more than 999_999 for ualarm() defined or not)
264 Added more ualarm() tests to catch various overflow points,
265 hopefully no problems in various platforms.
266 (The problem report by Mark Seger and Jon Paul Sullivan of HP.)
269 - tweak still needed for Const64(), from Jerry Hedden
270 - get a freshly generated ppport.h
271 - update Copyright years
274 - Const64() already appends an 'LL' (or i64), so provide LL and i64
275 forms for the IV_1E[679] (effects Win32 and Cygwin), reported by
277 - the Changes entry for 1.88 talked about [IN]V_1[679],
281 - clean up the g++ warnings in HiRes.xs, all of them
282 about mixing integer and floating point, introduce
283 constants IV_1E[679] and NV_1E[679]
286 - [rt.cpan.org #17442] 'make test' frequently fails under
287 Cygwin Perl v5.8.8, reported and patched by J. R. Hedden
288 (two race condition bugs in the END block in the case the
289 main process dies before the timer process, unearthed
290 by a bug in Cygwin ualarm)
293 - HiRes.t:s/ok 32/ok 33/, from Dominic Dunlop
294 - tighten up the clock() test marginally by requiring non-negative
295 - clock_nanosleep() and clock() doc tweaks
298 - the interface to clock_nanosleep() is more natural
299 when it is like (hires) time() (instead of like nanosleep),
300 and the .xs implementation of clock_nanosleep() in 1.84
302 - the semantics of clock() are not quite so silly as I thought,
303 but still somewhat odd, documented as such
304 - additional enhancements to the clock() documentation
305 - add test for clock_nanosleep() (I cannot test this
306 since none of my systems have the function)
307 - add test for clock()
310 - add clock() which returns the processor time in
311 (floating point) seconds since an arbitrary era
312 - add clock_nanosleep() which suspends the current
313 thread until either absolute time or for relative time
314 - [rt.cpan.org #16486] printf missing value in HiRes.t
315 - add constants CLOCKS_PER_SEC, CLOCK_SOFTTIME, TIMER_ABSTIME
319 - has_symbol() was wrong since e.g. ITIMER_VIRTUAL is exported
320 via @EXPORT_OK even when it is not available. This is heinous.
321 @EXPORT_OK should be determined at Makefile.PL time.
322 - be more lenient is testing clock_gettime(): allow more slop,
323 and retry up to three times, sleeping a random nap between
325 - human months are one-based (noticed by Anton Berezin)
328 - CLOCK_REALTIME is an enum value (of the clockid_t enum)
329 in HP-UX (and might be so elsewhere, too), debugged by
331 - include const-c.inc as late as possible (from Randy Kobes,
332 [rt.cpan.org #15552] to avoid undefined usleep() on Win32
335 - try to be more robust and consistent in the detection of
336 CLOCK_REALTIME and ITIMER_VIRTUAL in HiRes.t: the proper
341 eval 'import Time::HiRes qw($symbol)';
342 return 0 unless $@ eq '';
343 return exists ${"Time::HiRes::$symbol"};
350 in the test. All these moves are needed because
352 1) one cannot directly do eval 'Time::HiRes::FOO_BAR'
353 because FOO_BAR might have a true value of zero
354 (or in the general case an empty string or even undef)
356 2) In case FOO_BAR is not available in this platform,
357 &FOO_BAR avoids the bareword warning
359 - wait more (1.5 seconds instead of 0.1) for the CLOCK_REALTIME test
360 but expect the 'customary' slop of 0.20 instead of 0.25
361 - fixed inside a comment HAS_POLL -> TIME_HIRES_NANOSLEEP
362 - at the end of HiRest.t tell how close we were to termination
365 - Gisle noticed a mistake (using HAS_NANOSLEEP) in 1.79
368 - try nanosleep for emulating usleep -- may help in some weird
369 embedded realtime places which have nanosleep but neither usleep
370 nor select nor poll (doesn't have to be weird embedded realtime
371 place, though -- in many places usleep is nanosleep anyway)
372 - try poll for emulating usleep -- this may help some obscure/old
373 SVR4 places that have neither usleep nor select
374 - a redundant test guard in HiRes.t
377 - ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
378 in the general case fail already at 'use' phase is suboptimal)
379 - fixes to the documentation of clock_gettime() and clock_getres()
382 - add support for the POSIX clock_gettime() and clock_getres(),
383 if available, either as library calls or as syscalls
384 - be more defensive about missing functionality: break out
385 early (during 'use') if no e.g. clock_getres() is available,
386 and protect our back by trapping those cases also in HiRes.xs
387 - the test added in 1.76 could cause an endless loop e.g. in Solaris,
388 due to mixing of sleep() and alarm() (bad programmer, no cookie!)
391 - testing for nanosleep had wrong logic which caused nanosleep
392 to become undefined for e.g. Mac OS X
393 - added a test for a core dump that was introduced by Perl 5.8.0
394 safe signals and was fixed for the time of 5.8.1 (one report of
395 the core dump was [perl #20920]), the test skipped pre-5.8.1.
396 - *cough* s/unanosleep/nanosleep/g; *cough*
399 - installation patch from Gisle Aas: in Perls 5.8.x and later
400 use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
403 - [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
404 (the workaround is not to use rel2abs, should not be necessary)
405 - [cpan #14642] U2time wrongly exported on the C API
406 (patch supplied by the reporter, SALVA@cpan.org)
407 - add release dates to Changes
410 - Time::HiRes::nanosleep support for Solaris [PATCH]
411 (POSIX::uname() not available if building with core perl,
412 from Gisle Aas, via perl5-porters, perl change #25295)
415 - going back to the 1.68 loader setup (using DynaLoader)
416 since too many weird things starting breaking
417 - fix a typo in José Auguste-Etienne's name
420 - a thinko in the nanosleep() detection
421 - move more changes stuff from the README to Changes
422 - add -w to the Makefile.PL
425 - oops in 1.69 about @ISA (not affecting anything but silly)
426 - add copyright 2005 to HiRes.pm
427 - add copyright and license to HiRes.xs
428 - add copyrights 2003, 2004, 2005 to README
431 - actually run a test for nanosleep
432 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
433 it seems that one can link in nanosleep() but then calling
434 it fails instantly and sets errno to ENOSYS (Not implemented).
435 This may be fixable in the AIX case by figuring out the right
436 (realtime POSIX?) libs and whatnot, but in the general case
437 running a real test case is better. (Of course, this change
438 will no doubt run into portability problems because of the
439 execution step...) Note that because of hysterical raisins
440 most Perls do NOT have $Config{d_nanosleep} (scanning for
441 it by Configure would in many platforms require linking in
442 things like -lrt, which would in many platforms be a bad idea
444 (from José Auguste-Etienne)
445 - support XSLoader also since it's much faster
446 (from Alexey Tourbin)
447 - add SEE ALSO (BSD::Resource and Time::TAI64)
450 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
451 error suspected), but miraculously it still worked since the
452 doubling took place below the __END__ token
453 - undef Pause() before defining it to avoid redefinition warnings
454 during compilation in case perl.h had already defined Pause()
455 (part of perl change #24271)
459 - (internal) don't ignore the return value of gettimeofday()
460 - (external) return undef or an empty if the C gettimeofday() fails
461 (affects Time::HiRes gettimeofday() and the hires time())
465 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
466 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
467 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
468 - close also [rt.cpan.org #5933] "Time::HiRes::time does not
469 pick up time adjustments like ntp" since ever reproducing it
470 (and therefore verifying a possible fix) in the same environment
471 has become rather unlikely
474 - one should not mix u?alarm and sleep (the tests modified
475 by 1.65, #12 and #13, hung in Solaris), now we just busy
476 loop executing an empty block
477 - in the documentation underline the unspecificity of mixing
479 - small spelling fixes
482 - regenerate ppport.h with Devel::PPPort 3.03,
483 now the MY_CXT_CLONE is defined in ppport.h,
484 we no more need to do that.
486 - the test #12 would often hang in sigsuspend() (at least that's
487 where Mac OS X' ktrace shows it hanging). With the sleep()s
488 changed to sleep(1)s, the tests still pass but no hang after
489 a few hundred repeats.
492 - Win32 and any ithread build: ppport.h didn't define
493 MY_CXT_CLONE, which seems to be a Time-HiRes-ism.
496 - Skip testing if under PERL_CORE and Time::HiRes has not
497 been Configured (from Marcus Holland-Moritz, core change
499 - Use ppport.h generated by Devel::PPPort 3.01,
500 allowing cutting away our own portability code.
501 - Don't use $ENV{PERL_CORE} for < 5.6.0.
502 - Don't use "for my $i" for <= 5.003.
503 - Don't use Pause() for <= 5.003.
504 - Can't use newSVpvf for <= 5.003.
505 (most of the changes from Marcus)
508 - Win32: reset reading from the performance counters every
509 five minutes to better track wall clock time (thanks to
510 PC timers being often quite bad), should help long-running
514 - Win32: Patch from Steve Hay
515 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
516 to [perl #30755] reported by Nigel Sandever
518 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
519 <w32api/windows.h> APIs are available. Cygwin testing by
520 Yitzchak Scott-Thoennes.
522 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
523 after that keep using -lrt, patch from Alan Burlison,
524 bug reported in [cpan #7165]
527 - Change the Win32 recalibration limit to 0.5 seconds and tweak
528 the documentation to blather less about the gory details of the
529 Win32 implementation and more about the complications in general
530 of meddling with the system clock.
533 - Document the 1.57 change better.
536 - Win32/Cygwin/MinGW: if the performance counter drifts by more
537 than two seconds from the system clock (due to ntp adjustments,
538 for example), recalibrate our internal counter: from Jan Dubois,
539 based on [cpan #5933] by Jerry D. Hedden.
542 - Give a clearer message if the tests timeout (perl change #22253)
543 - Don't use /tmp or its moral equivalents (perl bug #15036,
547 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
548 instead of VC-specific i64 suffix)
551 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
554 - Windows: higher resolution time() by using the Windows
555 performance counter API, from Jan Dubois and Anton Shcherbinin.
556 The exact new higher resolution depends on the hardware,
557 but it should be quite a bit better than using the basic
561 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
562 the subtest 18. No known analysis nor fix, but added
563 an alarm (that requires fork() and alarm()) to the test.
566 - doc tweaks from mjd (perl change #20456)
567 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
570 - add a message (for non-core builds) to Makefile.PL about
571 the LC_ALL=C workaround
572 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
573 - the nanosleep() probe was broken (perl change #20061)
574 - use existence instead of definedness for feature probes
576 - MPE/iX tweak (perl change #20042)
577 - do not use HAS_NANOSLEEP (perl change #19898)
580 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
581 - OS/2 can always mix subsecond sleeps with signals
582 (part of perl change #19789)
585 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
589 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
590 - unnecessary (nanosleep is in libc anyway)
591 - harmful (-lrt slows down execution)
592 - incompatible (with many distributions' pthreads)
595 - do not create files in blib directories under core
596 (perl change #19160, from rgs)
597 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
600 - guarantee that $xdefine in HiRes.t is always defined
601 (perl change #19109, from IlyaZ)
602 - a cleaner way to detect PERL_CORE (perl change #19111,
606 - add hints/irix.pl to turn off overly POSIX flags that
607 cause hide struct timespec to be hidden (and compilation
608 to fail) (bleadperl change #19085)
609 - documentation tweaks
612 - add c:/temp to the list of temp directories to probe
613 so that cygwin (and win*?) builds are happy. This was
614 needed at least in my cygwin 1.3.20/w2k setup.
617 - modernize the constants code (from Nicholas Clark)
620 - At some point the ability to figure our the correct incdir
621 for EXTERN.h (either a core perl build, or an installed perl)
622 had broken (which lead into all test compiles failing with
623 a core perl build, but thanks to the robustness of Makefile.PL
624 nothing of this was visible). The brokenness seemed to be
625 caused by $ENV{PERL_CORE} not being on for core builds?
626 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
627 right, and both styles of build should work again.
630 - Nicholas Clark noticed that the my_catdir() emulation function
631 was broken (which means that we didn't really work for Perls
633 - inspired by fixing the above made the whole Makefile.PL -w
635 - tightened up the Makefile.PL output, less whitespace
638 - fix from Craig Berry for better building in VMS with PERL_CORE
641 - no functional changes
642 - move lib/Time/HiRes.pm as Hires.pm
643 - libraries scanning was slightly broken (always scanned
644 for a library even when $Config{libs} already had it)
647 - Ray Zimmerman ran into a race condition in Mac OS X.
648 A 0.01-second alarm fired before the test expected.
649 The test first slept indefinitely (blocking for signals)
650 and only after that tested for the signal having been sent.
651 Since the signal had already been sent, the test #12 never
652 completed. The solution: test first, then block.
653 - default to being silent on all probing attempts, set the
654 environment variable VERBOSE to a true value to see the
655 details (the probing command and the possible errors)
658 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
659 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
660 cannot be upgraded from CPAN to override the 5.8.0 version
662 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
663 because the include directories did not adjust themselves
664 if $ENV{PERL_CORE} (Hugo van der Sanden)
665 - add documentation about the restart of select() under alarm()
668 - small documentation tweaks
672 - better VMS operation (Craig Berry)
675 - our time machine is accelerating: now works with Perl 5.004_01
676 (tried with 5.003_07 and 5.002 but I get segmentation faults
677 from running the Makefile.PL with those in Tru64 4.0D)
680 - backward compatibility (pre-5.6.0) tweaks:
681 - no XSLoader in 5.00503, use DynaLoader instead
682 - no SvPV_nolen, either
683 - no PerlProc_pause(), either
684 - now tested with 5.00404 and 5.00503
685 - Makefile.PL requires 5.00404 (no more 5.002)
686 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
687 (this means that one can mix subsecond sleeps with alarms)
688 - because of nanosleep we probe for -lrt and -lposix4
689 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
690 is available by exportable constants Time::HiRes::d_func
691 (since older Perl do not have them in %Config, and even
692 5.8.0 does not probe for nanosleep)
695 - backward compatibility (pre-5.6.1) tweaks:
696 - define NV if no NVTYPE
697 - define IVdf if needed (note: the Devel::PPPort
698 in 5.8.0 does not try hard hard enough since
699 the IVSIZE might not be defined)
700 - define NVgf if needed
701 - grab the typemap from 5.8.0 for the NV stuff
703 1.31 and 1.32 add more backward compatibility (now all the way
704 back to Perl 5.00404), and using nanosleep() (if available) for
709 - release 1.29_02 as 1.30
711 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
712 development cycle. Most notably portability across platforms has been
713 enhanced, and the interval timers (setitimer, getitimer) have been
714 added. Note that the version of Time::HiRes that is included in Perl
715 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
716 version. Note also that in 1.30 Wegscheid turns over the maintenance
717 to Jarkko Hietaniemi.
721 - fix a silly unclosed comment typo in HiRes.xs
722 - document and export REALTIME_REALPROF (Solaris)
726 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
727 (need to patch this also in Perl 5.[89])
728 - remove CVS revision log from HiRes.xs
732 The following numbered patches refer to the Perl 5.7 changes,
733 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
735 - 17558: Add #!./perl to the .t
736 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
737 - 16198: political correctness, from Simon Cozens
738 - 15857: doc tweaks, from Jarkko Hietaniemi
739 - 15593: optimization in .xs, from Paul Green
740 - 14892: pod fixes, from Robin Barker
741 - 14100: VOS fixes, from Paul Green
742 - 13422: XS segfault, from Marc Lehmann
743 - 13378: whether select() gets restarted on signals, depends
744 - 13354: timing constraints, again, from Andy Dougherty
745 - 13278: can't do subsecond alarms with ualarm;
746 break out early if alarms do not seem to be working
747 - 13266: test relaxation (cygwin gets lower hires
748 times than lores ones)
749 - 12846: protect against high load, from Jarkko Hietaniemi
750 - 12837: HiRes.t VMS tweak, from Craig A. Berry
751 - 12797: HiRes.t VMS tweak, from Charles Lane
752 - 12769: HiRes.t VMS tweak, from Craig A. Berry
753 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
754 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
755 - 12692: alarm() ain't gonna work if ualarm() ain't,
756 from Gurusamy Sarathy
757 - 12680: minor VMS tweak, from Charles Lane
758 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
759 - 12609: croak on negative time, from Jarkko Hietaniemi
760 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
761 - 12594: MacOS Classic timeofday, from Chris Nandor
762 - 12473: allow for more than one second for sleep() and usleep()
763 - 12458: test tuning, relax timing constraints,
764 from Jarkko Hietaniemi
765 - 12449: make sleep() and usleep() to return the number
766 of seconds and microseconds actually slept (analogously
767 with the builtin sleep()), also make usleep() croak if
768 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
769 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
770 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
771 - 12196: use ftime() on Win32, from Artur Bergman
772 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
773 - 12105: use GetSystemTime() on Win32, from Artur Bergman
774 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
775 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
776 - 11797: problem in HiRes.t, from John P. Linderman
777 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
778 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
779 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
780 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
781 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
782 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
784 - 10942: MPE/IX test tweaks, from Mark Bixby
785 - 10784: unnecessary pod2man calls, from Andy Dougherty
786 - 10354: ext/ + -Wall, from Doug MacEachern
787 - 10320: fix the BOOT section to call myU2time correctly
788 - 10317: correct casting for AIX< from H. Merijn Brand
789 - 10119: document that the core time() may be rounding, not truncating
790 - 10118: test fix, from John Peacock
791 - 9988: long =item, from Robin Barker
792 - 9714: correct test output
793 - 9708: test also the scalar aspect of getitimer()
794 - 9705: Add interval timers (setitimer, getitimer)
795 - 9692: do not require at least 5.005 using XS
797 The following changes were made on top of the changes
798 made for Time::HiRes during the Perl 5.7 development
799 cycle that culminated in the release of Perl 5.8.0.
801 - add "require 5.005" to the Makefile.PL
802 - remove the REVISION section (CVS log) from HiRes.pm
803 - add jhi's copyright alongside Douglas'
804 - move HiRes.pm to lib/Time/
806 - modify HiRes.t to use $ENV{PERL_CORE}
807 - modify the original Time::HiRes version 1.20 Makefile.PL
808 to work both with Perl 5.8.0 and the new code with pre-5.8.0
809 Perls (tried with 5.6.1)
810 - tiny tweaks and updates in README and TODO
811 - bump the VERSION to 1.29
813 1.20 Wed Feb 24 21:30 1999
814 - make our usleep and ualarm substitutes into hrt_usleep
815 and hrt_ualarm. This helps static links of Perl with other
816 packages that also have usleep, etc. From
817 Ilya Zakharevich <ilya@math.ohio-state.edu>
818 - add C API stuff. From Joshua Pritikin
819 <joshua.pritikin@db.com>
820 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
821 - hopefully correct "-lc" fix for SCO.
824 1.20 adds a platform neutral set of C accessible routines if you are
825 running 5.005+. All other changes are packaging changes and build
826 fixes(?) for statically linked Perl, SCO, and VMS.
828 1.19 Tue Sep 29 22:30 1998
829 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
830 <seb@stian.demon.co.uk>
831 - change GIMME_V to GIMME to help people with older versions of
833 - fix Win32 version of gettimeofday(). It didn't affect anything,
834 but it confuses people reading the code when the return value
835 is backwards (0 is success).
836 - fix Makefile.PL (more) so that detection of gettimeofday is
839 1.19 has better VMS support.
841 1.18 Mon Jul 6 22:40 1998
842 - add usleep() for Win32.
843 - fix Makefile.PL to fix reported HP/UX feature where unresolved
844 externals still cause an executable to be generated (though no
845 x bit set). Thanks to David Kozinn for report and explanation.
846 Problems with the fix are mine :)
848 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
849 Probably buggy. I'm sure I'll hear.
851 1.17 Wed Jul 1 20:10 1998
852 - fix setitimer calls so microseconds is not more than 1000000.
853 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
854 - make Win32. We only get gettimeofday (the select hack doesn't
855 seem to work on my Win95 system).
856 - fix test 4 on 01test.t. add test to see if time() and
857 Time::HiRes::time() are close.
859 1.16 Wed Nov 12 21:05 1997
860 - add missing EXTEND in new gettimeofday scalar code.
862 1.16+ should be closer to building out of the box on Linux. Thanks
863 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
865 If your underlying operating system doesn't implement ualarm(), then
866 a fake using setitimer() will be made. If the OS is missing usleep(),
867 a fake one using select() will be made. If a fake can't be made for
868 either ualarm() or usleep(), then the corresponding Perl function will
869 not be available. If the OS is missing gettimeofday(), you will get
870 unresolved externals, either at link- or run-time.
872 This is an improvement; the package used to not even build if
873 you were missing any of these bits. Roderick Schertler
875 <roderick@argon.org> did all the conditional compilation stuff,
876 look at HiRes.pm and the test suites; it's good educational reading.
878 1.15 Mon Nov 10 21:30 1997
879 - HiRes.pm: update pod. Provided by Gisle Aas.
880 - HiRes.xs: if gettimeofday() called in scalar context, do
881 something more useful than before. Provided by Gisle Aas.
882 - README: tell of xsubpp '-nolinenumber' woes. thanks to
883 Edward Henigin <ed@texas.net> for pointing out the problem.
885 1.14 Wed Nov 5 9:40 1997
886 - Makefile.PL: look for setitimer
887 - HiRes.xs: if missing ualarm, but we have setitimer, make up
888 our own setitimer. These were provided by Gisle Aas.
890 1.13 Tue Nov 4 23:30 1997
891 - Makefile.PL: fix autodetect mechanism to do try linking in addition
892 to just compiling; should fix Linux build problem. Fix was provided
895 1.12 Sun Oct 12 12:00:00 1997
896 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
897 you may need to comment this back out if you have an older xsubpp.
898 - HiRes.xs: set PROTOTYPES: DISABLE
900 1.11 Fri Sep 05 16:00:00 1997
902 Had some line commented out that shouldn't have been (testing
905 Previous version was corrupted.
907 1.10 Thu May 22 20:20:00 1997
908 - HiRes.xs, HiRes.pm, t/*:
909 - only compile what we have OS support for (or can
911 - only test what we compiled
912 - gross improvement to the test suite
914 This work was all done by Roderick Schertler
915 <roderick@argon.org>. If you run Linux or
916 one of the other ualarm-less platforms, and you like this
917 module, let Roderick know; without him, it still wouldn't
918 be working on those boxes...
919 - Makefile.PL: figure out what routines the OS has and
920 only build what we need. These bits were written by Jarkko
921 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
923 1.02 Mon Dec 30 08:00:00 1996
924 - HiRes.pm: update documentation to say what to do when missing
925 ualarm() and friends.
926 - README: update to warn that ualarm() and friends need to exist
928 1.01 Fri Oct 17 08:00:00 1996
929 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
930 - HiRes.pm: put blank line between __END__ and =head1 so that
933 1.00 Tue Sep 03 13:00:00 1996
934 - original version; created by h2xs 1.16