1 Revision history for the Perl extension Time::HiRes.
4 - C90 declaration-after-statement error with darwin threads: blead de1003b4
8 - MUTEX_LOCK, not PERL_MUTEX_LOCK: blead e5b02b5d
9 - also hrt_ualarm_itimer() is unused: 1cb6cce3
10 - the mutex needs init: 2d639e20
11 - version bump to 1.9732: 730d7fdc
14 - mark unused variable as such: blead a914236c
15 - OS X emulation mutex accidentally unused: da7a6455
16 - remove hrt_ualarm: 6da77c36
17 - pod error fixes: 919ca095
18 - nanosleep and clock_nanosleep cleanups: c8ea02b8..e3ff671b
19 - static funcs in HiRes.xs: 4e922c26
20 - Remove unused variable: 52ffb1b5
23 - TIME_HIRES_DONT_RUN_PROBES=1 to build the probes but not run them
29 - clock_getres(), clock_gettime(), clock_nanosleep() emulation for OS X
33 - No changes since 1.9727_03
35 1.9727_03 [2015-11-06]
36 - use PERL_UNUSED_ARG() in the not-here-croak branches
37 - Makefile.PL/metadata improvements
39 1.9727_02 [2015-08-29]
40 - correct documentation to note that Time::NVtime is NV, not double
42 - avoid a warning about clock_t on OS X / clang 3.6 - 3.6
45 1.9727_01 [2015-08-17]
46 - Remove usage of 5.005 thread APIs in favor of 5.6 and up thread APIs
47 [rt.cpan.org #102799].
48 - Silence a C++ compiler warning [rt.cpan.org #101497].
49 - respect $Config{run} for cross-compiling [rt.cpan.org #92607]
52 - Correct s/us splitting of usleep(1000000) [rt.cpan.org #78266].
53 - Avoid integer overflow in itimer-based alarm() with large
54 argument [rt.cpan.org #87160].
55 - Define PERL_NO_INLINE_FUNCTIONS during probe compilations, to
56 avoid false negatives caused by not linking with the perl core.
57 - Be more careful about context in stat().
58 - Install into "site" directories by default on perl 5.11+
60 - Fix a couple of doc typos [rt.cpan.org #85365].
61 - Fix function name in a doc example [rt.cpan.org #86318].
62 - Provide lstat() that yields high-res timestamps, alongside
63 the existing high-res stat() [rt.cpan.org #78732].
66 - Correct stack discipline in stat(), which was screwing up list
67 operations in expressions containing calls to it [rt.cpan.org
69 - Add missing OUTPUT sections to the XS code [rt.cpan.org #70930].
70 - Skip itimer tests on GNU/Hurd, which has the API but lacks
71 the implementation [rt.cpan.org #72754].
72 - Fix a doubled word in the documentation [rt.cpan.org #72763].
75 - Correct XS parameter list, and therefore prototype, for
76 unimplemented-on-this-platform version of clock_nanosleep()
78 - Declare package variables with "our" rather than "use vars".
79 - Corresponding to "our" usage, check for minimum Perl version
81 - Declare module dependencies.
84 - Remove $ENV{PERL_CORE} logic from test suite, which is no
85 longer desired in the core.
86 - Convert test suite to use Test::More.
87 - Factor out watchdog code from test suite.
88 - In test suite, be consistent about using fully-qualified form
90 - Divide test suite into feature-specific scripts.
91 - Make ualarm timing test less vulnerable to delay-induced false
92 failure, from Dave Mitchell.
95 - Update for changes in build process in the core, patches
96 from BinGOs [rt.cpan.org #58858] and Craig Berry [rt.cpan.org
98 - Fix broken linkage on Windows with gcc 3.4 seen with ActivePerl,
99 report from Christian Walde [rt.cpan.org #61648], fix derived
101 - Jump through hoops to avoid compiler warnings.
104 - Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
105 additional fix from Gisle Aas
106 - Address [rt.cpan.org #55665] "Bad plan" on Windows,
107 report and fix from Jan Dubois
110 - Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
111 report and fix from Nicholas Clark
114 - As with QNX, Haiku has the API of interval timers but not
115 the implementation (bleadperl change #34630), hence skip
116 the tests, via David Mitchell.
119 - .xs code cleanup from Albert Dvornik
120 - in the #39 and #40 do not do us I did, mixing alarm() and
121 sleep(). Now instead spin until enough time has passed.
124 - Skip the tests added in 1.9716 (#39, #40) if there's no subsecond
125 alarm capability, like with the older subsecond alarm tests
128 - Change documentation to agree with reality: there are
129 no interval timers in Win32.
130 - Address [rt.cpan.org #35899] (problem in subsecond sleeps),
131 add two tests to guard against this problem
132 - Address [rt.cpan.org #36600] 'Division by zero' failure in test suite
133 - Address [rt.cpan.org #37340] [PATCH] Address timer process in test
134 - Address [rt.cpan.org#40311 ] bad implementation of hrt_usleep
135 with TIME_HIRES_NANOSLEEP
138 - Silly me: Makefile.PL does need to accept arguments other than mine.
139 Some testing frameworks obviously do this.
140 - Add retrying for tests 34..37, which are the most commonly
141 failing tests. If this helps, consider extending the retry
142 framework to all the tests. [Inspired by Slaven Rezic,
143 [rt.cpan.org #34711] Occasional failures of test 35 or 36 (FreeBSD)]
146 - Under Perl 5.6.* NVgf needs to be "g", reported by Zefram,
147 it seems that ppport.h 3.13 gets this wrong.
148 - remove the check in Makefile.PL for 5.7.2, shouldn't be
149 (a) necessary (b) relevant
150 - add logic to Makefile.PL to skip configure/write Makefile
151 step if the "xdefine" file already exists, indicating that
152 the configure step has already been done, one can still
153 force (re)configure by "perl Makefile.PL configure",
154 or of course by "make clean && perl Makefile.PL".
157 - for alarm() and ualarm() [Perl] prefer setitimer() [C]
158 instead of ualarm() [C] since ualarm() [C] cannot portably
159 (and standards-compliantly) be used for more than 999_999
160 microseconds (rt.cpan.org #34655)
161 - it seems that HP-UX has started (at least in 11.31 ia64)
162 #defining the CLOCK_REALTIME et alia (instead of having
164 - document all the diagnostics
167 - move the sub tick in the test file back to where it used to be
168 - in the "consider upgrading" message recommend at least Perl 5.8.8
169 and make the message to appear only for 5.8.0 since 5.8.1 and
170 later have the problem fixed
171 - VOS tweak for Makefile (core perl change #33259)
172 - since the test #17 seems to fail often, relax its limits a bit
175 - lost VMS test skippage from Craig Berry
176 - reformat the test code a little
179 - I got the sense of the QNX test the wrong way in an attempt
180 to generalize it for future
183 - casting fixes from Robin Barker for g++ and 64bitint
184 - in QNX skip the itimer tests because though the API
185 is there, the implementation isn't, from Matt Kraai
186 - raise the dead man timer to 180 seconds for really
188 - elaborate the UTF-8 locale warning from Makefile.PL
191 - [rt.cpan.org #29747]: Build failure with perl 5.005_05
192 Fixed by regenerating the ppport.h using Devel::PPPort 3.13.
195 - t/HiRes.t failed in Perl 5.6.2,
196 "action is not of type POSIX::SigAction at t/HiRes.t line 318",
197 reported and fixed by Anton Berezin, the reason was faulty
198 use of sigaction() when restoring the old action.
201 - with bleadperl in VMS the HiRes.t overrun the maximum number
202 of deferred signals because the libc SIGALRM was not strong
203 enough to interrupt select(), and select() got restarted every
204 time, solution is to use POSIX::SigAction if available.
205 A fix from Craig Berry (not 100% there, but helps).
206 - allow for more measuring noise for ualarm() tests 35..37
209 - nanosleep() and clock_nanosleep() detection and use were
210 quite broken; in Linux -lrt needed; fixes from Zefram
211 - [internal] slightly cleaner building of $DEFINE in Makefile.PL,
212 should avoid double/conflicting -D flags
215 - allow 10% of slop in test #14 (testing difference between
216 CORE::time() and Time::HiRes::time()), there seem to be often
217 transient failures from Perl smoke builds on this test
221 - use int main(int argc, char **argv) consistently in Makefile.PL,
223 [rt.cpan.org #23868] nanosleep not detected under Mac OS 10.3.9 starting with Time::HiRes 1.96
224 - if someone still has the locale-broken Perl 5.8.0,
225 suggest that they upgrade their Perl
228 - restore the -DATLEASTFIVEOHOHFIVE, Win32 needed it still
231 - upgrade to ppport.h 3.10_02
232 - remove the -DATLEASTFIVEOHOHFIVE
233 - use the ppport.h PL_ppaddr, PL_statcache, PL_laststatval
234 - use the ppport.h aTHXR for calling Perl stat()
235 - switch into four-digit version since 2.0 is coming up
236 awfully fast but not feeling like a major rewrite
239 - 1.95 broke building in Win32 (since pp_stat is not exported),
240 figured out how to call an op directly in 5.005 (use Perl_ppaddr
241 instead of PL_ppaddr)
242 - backport to Perl 5.004_05 (requires using statcache
243 and laststatval instead of PL_statcache and PL_laststatval)
244 (also checked to work in 5.005_04, 5.6.1, and 5.8.8 with threads)
247 - 1.95 broke builds for threaded Perls, rt.cpan.org tickets:
248 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
249 [rt.cpan.org #23712] Time-HiRes 1.95 Fails make on AIX 5.2 with Perl 5.8.8
250 [rt.cpan.org #23730] Time::HiRes 1.95 fails make on MacOS X 10.3.9/perl 5.8.8
251 - use main() prototype consistently in Makefile.PL
254 - integrate core change #29180: Silence VC++ compiler warnings
256 - do not use PL_ppaddr in stat() because that is not available
258 - regenerate fallback/*.inc for older Perls without
259 ExtUtils::Constant because of d_hires_stat, resolves
260 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
261 - Make Makefile.PL more defensive against false PERL_CORE
264 - file timestamps oddities seen: the atime and mtime
265 can be out of sync (modify first and read second can leave
266 atime < mtime) and mtime can be subsecond while atime is not.
267 So make the test more forgiving.
270 - the ualarm() tests (34-37) assumed that ualarm(N)
271 could never alarm in less than N seconds, widened
272 the acceptable relative range to 0.9..1.5. Addresses
273 [rt.cpan.org #22090] and [rt.cpan.org #22091].
275 - skip the stat() tests in cygwin and win32, because
276 if run on FAT the timestamp granularity is only 2 seconds.
277 Any good way to detect (cygwin or win32) whether we are
278 being run on NTFS or anywhere with better timestamps?
279 Addresses [rt.cpan.org #22089] and [rt.cpan.org #22098].
282 - scan for subsecond resolution timestamps in struct stat,
283 some known possibilities:
285 (1) struct timespec st_atimespec;
286 st_atimespec.tv_nsec;
291 (4) timestruc_t st_atim;
296 If something like this is found, one can do
299 my @stat = Time::HiRes::stat();
301 or even override the standard stat():
303 use Time::HiRes qw(stat);
305 to get the stat() timestamps
307 my ($atime, $mtime, $ctime) = @stat[8, 9, 10];
309 with subsecond resolution (assuming both the operating
310 system and the filesystem support that kind of thing).
312 Contributions for more systems (especially non-UNIX,
313 e.g. but not limited to: Win32, VMS, OS/2) gladly accepted.
314 (also more UNIX variants welcome: HP-UX? IRIX?)
316 Thanks to H.Merijn Brand, John Peacock, and Craig
317 Berry for brave beta testing.
320 - ualarm() in SuSE 10.1 was overflowing after ~4.2 seconds,
321 possibly due to a glibc bug/feature (suspected overflow at
322 2**32 microseconds?), workaround by using the setitimer()
323 implementation of ualarm() if either useconds or
324 interval > 999_999 (this case seems to vary between systems:
325 are useconds more than 999_999 for ualarm() defined or not)
326 Added more ualarm() tests to catch various overflow points,
327 hopefully no problems in various platforms.
328 (The problem report by Mark Seger and Jon Paul Sullivan of HP.)
331 - tweak still needed for Const64(), from Jerry Hedden
332 - get a freshly generated ppport.h
333 - update Copyright years
336 - Const64() already appends an 'LL' (or i64), so provide LL and i64
337 forms for the IV_1E[679] (effects Win32 and Cygwin), reported by
339 - the Changes entry for 1.88 talked about [IN]V_1[679],
343 - clean up the g++ warnings in HiRes.xs, all of them
344 about mixing integer and floating point, introduce
345 constants IV_1E[679] and NV_1E[679]
348 - [rt.cpan.org #17442] 'make test' frequently fails under
349 Cygwin Perl v5.8.8, reported and patched by J. R. Hedden
350 (two race condition bugs in the END block in the case the
351 main process dies before the timer process, unearthed
352 by a bug in Cygwin ualarm)
355 - HiRes.t:s/ok 32/ok 33/, from Dominic Dunlop
356 - tighten up the clock() test marginally by requiring non-negative
357 - clock_nanosleep() and clock() doc tweaks
360 - the interface to clock_nanosleep() is more natural
361 when it is like (hires) time() (instead of like nanosleep),
362 and the .xs implementation of clock_nanosleep() in 1.84
364 - the semantics of clock() are not quite so silly as I thought,
365 but still somewhat odd, documented as such
366 - additional enhancements to the clock() documentation
367 - add test for clock_nanosleep() (I cannot test this
368 since none of my systems have the function)
369 - add test for clock()
372 - add clock() which returns the processor time in
373 (floating point) seconds since an arbitrary era
374 - add clock_nanosleep() which suspends the current
375 thread until either absolute time or for relative time
376 - [rt.cpan.org #16486] printf missing value in HiRes.t
377 - add constants CLOCKS_PER_SEC, CLOCK_SOFTTIME, TIMER_ABSTIME
381 - has_symbol() was wrong since e.g. ITIMER_VIRTUAL is exported
382 via @EXPORT_OK even when it is not available. This is heinous.
383 @EXPORT_OK should be determined at Makefile.PL time.
384 - be more lenient is testing clock_gettime(): allow more slop,
385 and retry up to three times, sleeping a random nap between
387 - human months are one-based (noticed by Anton Berezin)
390 - CLOCK_REALTIME is an enum value (of the clockid_t enum)
391 in HP-UX (and might be so elsewhere, too), debugged by
393 - include const-c.inc as late as possible (from Randy Kobes,
394 [rt.cpan.org #15552] to avoid undefined usleep() on Win32
397 - try to be more robust and consistent in the detection of
398 CLOCK_REALTIME and ITIMER_VIRTUAL in HiRes.t: the proper
403 eval 'import Time::HiRes qw($symbol)';
404 return 0 unless $@ eq '';
405 return exists ${"Time::HiRes::$symbol"};
412 in the test. All these moves are needed because
414 1) one cannot directly do eval 'Time::HiRes::FOO_BAR'
415 because FOO_BAR might have a true value of zero
416 (or in the general case an empty string or even undef)
418 2) In case FOO_BAR is not available in this platform,
419 &FOO_BAR avoids the bareword warning
421 - wait more (1.5 seconds instead of 0.1) for the CLOCK_REALTIME test
422 but expect the 'customary' slop of 0.20 instead of 0.25
423 - fixed inside a comment HAS_POLL -> TIME_HIRES_NANOSLEEP
424 - at the end of HiRest.t tell how close we were to termination
427 - Gisle noticed a mistake (using HAS_NANOSLEEP) in 1.79
430 - try nanosleep for emulating usleep -- may help in some weird
431 embedded realtime places which have nanosleep but neither usleep
432 nor select nor poll (doesn't have to be weird embedded realtime
433 place, though -- in many places usleep is nanosleep anyway)
434 - try poll for emulating usleep -- this may help some obscure/old
435 SVR4 places that have neither usleep nor select
436 - a redundant test guard in HiRes.t
439 - ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
440 in the general case fail already at 'use' phase is suboptimal)
441 - fixes to the documentation of clock_gettime() and clock_getres()
444 - add support for the POSIX clock_gettime() and clock_getres(),
445 if available, either as library calls or as syscalls
446 - be more defensive about missing functionality: break out
447 early (during 'use') if no e.g. clock_getres() is available,
448 and protect our back by trapping those cases also in HiRes.xs
449 - the test added in 1.76 could cause an endless loop e.g. in Solaris,
450 due to mixing of sleep() and alarm() (bad programmer, no cookie!)
453 - testing for nanosleep had wrong logic which caused nanosleep
454 to become undefined for e.g. Mac OS X
455 - added a test for a core dump that was introduced by Perl 5.8.0
456 safe signals and was fixed for the time of 5.8.1 (one report of
457 the core dump was [perl #20920]), the test skipped pre-5.8.1.
458 - *cough* s/unanosleep/nanosleep/g; *cough*
461 - installation patch from Gisle Aas: in Perls 5.8.x and later
462 use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
465 - [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
466 (the workaround is not to use rel2abs, should not be necessary)
467 - [cpan #14642] U2time wrongly exported on the C API
468 (patch supplied by the reporter, SALVA@cpan.org)
469 - add release dates to Changes
472 - Time::HiRes::nanosleep support for Solaris [PATCH]
473 (POSIX::uname() not available if building with core perl,
474 from Gisle Aas, via perl5-porters, perl change #25295)
477 - going back to the 1.68 loader setup (using DynaLoader)
478 since too many weird things starting breaking
479 - fix a typo in José Auguste-Etienne's name
482 - a thinko in the nanosleep() detection
483 - move more changes stuff from the README to Changes
484 - add -w to the Makefile.PL
487 - oops in 1.69 about @ISA (not affecting anything but silly)
488 - add copyright 2005 to HiRes.pm
489 - add copyright and license to HiRes.xs
490 - add copyrights 2003, 2004, 2005 to README
493 - actually run a test for nanosleep
494 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
495 it seems that one can link in nanosleep() but then calling
496 it fails instantly and sets errno to ENOSYS (Not implemented).
497 This may be fixable in the AIX case by figuring out the right
498 (realtime POSIX?) libs and whatnot, but in the general case
499 running a real test case is better. (Of course, this change
500 will no doubt run into portability problems because of the
501 execution step...) Note that because of hysterical raisins
502 most Perls do NOT have $Config{d_nanosleep} (scanning for
503 it by Configure would in many platforms require linking in
504 things like -lrt, which would in many platforms be a bad idea
506 (from José Auguste-Etienne)
507 - support XSLoader also since it's much faster
508 (from Alexey Tourbin)
509 - add SEE ALSO (BSD::Resource and Time::TAI64)
512 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
513 error suspected), but miraculously it still worked since the
514 doubling took place below the __END__ token
515 - undef Pause() before defining it to avoid redefinition warnings
516 during compilation in case perl.h had already defined Pause()
517 (part of perl change #24271)
521 - (internal) don't ignore the return value of gettimeofday()
522 - (external) return undef or an empty if the C gettimeofday() fails
523 (affects Time::HiRes gettimeofday() and the hires time())
527 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
528 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
529 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
530 - close also [rt.cpan.org #5933] "Time::HiRes::time does not
531 pick up time adjustments like ntp" since ever reproducing it
532 (and therefore verifying a possible fix) in the same environment
533 has become rather unlikely
536 - one should not mix u?alarm and sleep (the tests modified
537 by 1.65, #12 and #13, hung in Solaris), now we just busy
538 loop executing an empty block
539 - in the documentation underline the unspecificity of mixing
541 - small spelling fixes
544 - regenerate ppport.h with Devel::PPPort 3.03,
545 now the MY_CXT_CLONE is defined in ppport.h,
546 we no more need to do that.
548 - the test #12 would often hang in sigsuspend() (at least that's
549 where Mac OS X' ktrace shows it hanging). With the sleep()s
550 changed to sleep(1)s, the tests still pass but no hang after
551 a few hundred repeats.
554 - Win32 and any ithread build: ppport.h didn't define
555 MY_CXT_CLONE, which seems to be a Time-HiRes-ism.
558 - Skip testing if under PERL_CORE and Time::HiRes has not
559 been Configured (from Marcus Holland-Moritz, core change
561 - Use ppport.h generated by Devel::PPPort 3.01,
562 allowing cutting away our own portability code.
563 - Don't use $ENV{PERL_CORE} for < 5.6.0.
564 - Don't use "for my $i" for <= 5.003.
565 - Don't use Pause() for <= 5.003.
566 - Can't use newSVpvf for <= 5.003.
567 (most of the changes from Marcus)
570 - Win32: reset reading from the performance counters every
571 five minutes to better track wall clock time (thanks to
572 PC timers being often quite bad), should help long-running
576 - Win32: Patch from Steve Hay
577 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
578 to [perl #30755] reported by Nigel Sandever
580 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
581 <w32api/windows.h> APIs are available. Cygwin testing by
582 Yitzchak Scott-Thoennes.
584 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
585 after that keep using -lrt, patch from Alan Burlison,
586 bug reported in [cpan #7165]
589 - Change the Win32 recalibration limit to 0.5 seconds and tweak
590 the documentation to blather less about the gory details of the
591 Win32 implementation and more about the complications in general
592 of meddling with the system clock.
595 - Document the 1.57 change better.
598 - Win32/Cygwin/MinGW: if the performance counter drifts by more
599 than two seconds from the system clock (due to ntp adjustments,
600 for example), recalibrate our internal counter: from Jan Dubois,
601 based on [cpan #5933] by Jerry D. Hedden.
604 - Give a clearer message if the tests timeout (perl change #22253)
605 - Don't use /tmp or its moral equivalents (perl bug #15036,
609 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
610 instead of VC-specific i64 suffix)
613 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
616 - Windows: higher resolution time() by using the Windows
617 performance counter API, from Jan Dubois and Anton Shcherbinin.
618 The exact new higher resolution depends on the hardware,
619 but it should be quite a bit better than using the basic
623 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
624 the subtest 18. No known analysis nor fix, but added
625 an alarm (that requires fork() and alarm()) to the test.
628 - doc tweaks from mjd (perl change #20456)
629 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
632 - add a message (for non-core builds) to Makefile.PL about
633 the LC_ALL=C workaround
634 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
635 - the nanosleep() probe was broken (perl change #20061)
636 - use existence instead of definedness for feature probes
638 - MPE/iX tweak (perl change #20042)
639 - do not use HAS_NANOSLEEP (perl change #19898)
642 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
643 - OS/2 can always mix subsecond sleeps with signals
644 (part of perl change #19789)
647 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
651 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
652 - unnecessary (nanosleep is in libc anyway)
653 - harmful (-lrt slows down execution)
654 - incompatible (with many distributions' pthreads)
657 - do not create files in blib directories under core
658 (perl change #19160, from rgs)
659 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
662 - guarantee that $xdefine in HiRes.t is always defined
663 (perl change #19109, from IlyaZ)
664 - a cleaner way to detect PERL_CORE (perl change #19111,
668 - add hints/irix.pl to turn off overly POSIX flags that
669 cause hide struct timespec to be hidden (and compilation
670 to fail) (bleadperl change #19085)
671 - documentation tweaks
674 - add c:/temp to the list of temp directories to probe
675 so that cygwin (and win*?) builds are happy. This was
676 needed at least in my cygwin 1.3.20/w2k setup.
679 - modernize the constants code (from Nicholas Clark)
682 - At some point the ability to figure our the correct incdir
683 for EXTERN.h (either a core perl build, or an installed perl)
684 had broken (which lead into all test compiles failing with
685 a core perl build, but thanks to the robustness of Makefile.PL
686 nothing of this was visible). The brokenness seemed to be
687 caused by $ENV{PERL_CORE} not being on for core builds?
688 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
689 right, and both styles of build should work again.
692 - Nicholas Clark noticed that the my_catdir() emulation function
693 was broken (which means that we didn't really work for Perls
695 - inspired by fixing the above made the whole Makefile.PL -w
697 - tightened up the Makefile.PL output, less whitespace
700 - fix from Craig Berry for better building in VMS with PERL_CORE
703 - no functional changes
704 - move lib/Time/HiRes.pm as Hires.pm
705 - libraries scanning was slightly broken (always scanned
706 for a library even when $Config{libs} already had it)
709 - Ray Zimmerman ran into a race condition in Mac OS X.
710 A 0.01-second alarm fired before the test expected.
711 The test first slept indefinitely (blocking for signals)
712 and only after that tested for the signal having been sent.
713 Since the signal had already been sent, the test #12 never
714 completed. The solution: test first, then block.
715 - default to being silent on all probing attempts, set the
716 environment variable VERBOSE to a true value to see the
717 details (the probing command and the possible errors)
720 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
721 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
722 cannot be upgraded from CPAN to override the 5.8.0 version
724 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
725 because the include directories did not adjust themselves
726 if $ENV{PERL_CORE} (Hugo van der Sanden)
727 - add documentation about the restart of select() under alarm()
730 - small documentation tweaks
734 - better VMS operation (Craig Berry)
737 - our time machine is accelerating: now works with Perl 5.004_01
738 (tried with 5.003_07 and 5.002 but I get segmentation faults
739 from running the Makefile.PL with those in Tru64 4.0D)
742 - backward compatibility (pre-5.6.0) tweaks:
743 - no XSLoader in 5.00503, use DynaLoader instead
744 - no SvPV_nolen, either
745 - no PerlProc_pause(), either
746 - now tested with 5.00404 and 5.00503
747 - Makefile.PL requires 5.00404 (no more 5.002)
748 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
749 (this means that one can mix subsecond sleeps with alarms)
750 - because of nanosleep we probe for -lrt and -lposix4
751 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
752 is available by exportable constants Time::HiRes::d_func
753 (since older Perl do not have them in %Config, and even
754 5.8.0 does not probe for nanosleep)
757 - backward compatibility (pre-5.6.1) tweaks:
758 - define NV if no NVTYPE
759 - define IVdf if needed (note: the Devel::PPPort
760 in 5.8.0 does not try hard hard enough since
761 the IVSIZE might not be defined)
762 - define NVgf if needed
763 - grab the typemap from 5.8.0 for the NV stuff
765 1.31 and 1.32 add more backward compatibility (now all the way
766 back to Perl 5.00404), and using nanosleep() (if available) for
771 - release 1.29_02 as 1.30
773 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
774 development cycle. Most notably portability across platforms has been
775 enhanced, and the interval timers (setitimer, getitimer) have been
776 added. Note that the version of Time::HiRes that is included in Perl
777 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
778 version. Note also that in 1.30 Wegscheid turns over the maintenance
779 to Jarkko Hietaniemi.
783 - fix a silly unclosed comment typo in HiRes.xs
784 - document and export REALTIME_REALPROF (Solaris)
788 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
789 (need to patch this also in Perl 5.[89])
790 - remove CVS revision log from HiRes.xs
794 The following numbered patches refer to the Perl 5.7 changes,
795 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
797 - 17558: Add #!./perl to the .t
798 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
799 - 16198: political correctness, from Simon Cozens
800 - 15857: doc tweaks, from Jarkko Hietaniemi
801 - 15593: optimization in .xs, from Paul Green
802 - 14892: pod fixes, from Robin Barker
803 - 14100: VOS fixes, from Paul Green
804 - 13422: XS segfault, from Marc Lehmann
805 - 13378: whether select() gets restarted on signals, depends
806 - 13354: timing constraints, again, from Andy Dougherty
807 - 13278: can't do subsecond alarms with ualarm;
808 break out early if alarms do not seem to be working
809 - 13266: test relaxation (cygwin gets lower hires
810 times than lores ones)
811 - 12846: protect against high load, from Jarkko Hietaniemi
812 - 12837: HiRes.t VMS tweak, from Craig A. Berry
813 - 12797: HiRes.t VMS tweak, from Charles Lane
814 - 12769: HiRes.t VMS tweak, from Craig A. Berry
815 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
816 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
817 - 12692: alarm() ain't gonna work if ualarm() ain't,
818 from Gurusamy Sarathy
819 - 12680: minor VMS tweak, from Charles Lane
820 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
821 - 12609: croak on negative time, from Jarkko Hietaniemi
822 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
823 - 12594: MacOS Classic timeofday, from Chris Nandor
824 - 12473: allow for more than one second for sleep() and usleep()
825 - 12458: test tuning, relax timing constraints,
826 from Jarkko Hietaniemi
827 - 12449: make sleep() and usleep() to return the number
828 of seconds and microseconds actually slept (analogously
829 with the builtin sleep()), also make usleep() croak if
830 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
831 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
832 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
833 - 12196: use ftime() on Win32, from Artur Bergman
834 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
835 - 12105: use GetSystemTime() on Win32, from Artur Bergman
836 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
837 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
838 - 11797: problem in HiRes.t, from John P. Linderman
839 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
840 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
841 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
842 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
843 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
844 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
846 - 10942: MPE/IX test tweaks, from Mark Bixby
847 - 10784: unnecessary pod2man calls, from Andy Dougherty
848 - 10354: ext/ + -Wall, from Doug MacEachern
849 - 10320: fix the BOOT section to call myU2time correctly
850 - 10317: correct casting for AIX< from H. Merijn Brand
851 - 10119: document that the core time() may be rounding, not truncating
852 - 10118: test fix, from John Peacock
853 - 9988: long =item, from Robin Barker
854 - 9714: correct test output
855 - 9708: test also the scalar aspect of getitimer()
856 - 9705: Add interval timers (setitimer, getitimer)
857 - 9692: do not require at least 5.005 using XS
859 The following changes were made on top of the changes
860 made for Time::HiRes during the Perl 5.7 development
861 cycle that culminated in the release of Perl 5.8.0.
863 - add "require 5.005" to the Makefile.PL
864 - remove the REVISION section (CVS log) from HiRes.pm
865 - add jhi's copyright alongside Douglas'
866 - move HiRes.pm to lib/Time/
868 - modify HiRes.t to use $ENV{PERL_CORE}
869 - modify the original Time::HiRes version 1.20 Makefile.PL
870 to work both with Perl 5.8.0 and the new code with pre-5.8.0
871 Perls (tried with 5.6.1)
872 - tiny tweaks and updates in README and TODO
873 - bump the VERSION to 1.29
875 1.20 Wed Feb 24 21:30 1999
876 - make our usleep and ualarm substitutes into hrt_usleep
877 and hrt_ualarm. This helps static links of Perl with other
878 packages that also have usleep, etc. From
879 Ilya Zakharevich <ilya@math.ohio-state.edu>
880 - add C API stuff. From Joshua Pritikin
881 <joshua.pritikin@db.com>
882 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
883 - hopefully correct "-lc" fix for SCO.
886 1.20 adds a platform neutral set of C accessible routines if you are
887 running 5.005+. All other changes are packaging changes and build
888 fixes(?) for statically linked Perl, SCO, and VMS.
890 1.19 Tue Sep 29 22:30 1998
891 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
892 <seb@stian.demon.co.uk>
893 - change GIMME_V to GIMME to help people with older versions of
895 - fix Win32 version of gettimeofday(). It didn't affect anything,
896 but it confuses people reading the code when the return value
897 is backwards (0 is success).
898 - fix Makefile.PL (more) so that detection of gettimeofday is
901 1.19 has better VMS support.
903 1.18 Mon Jul 6 22:40 1998
904 - add usleep() for Win32.
905 - fix Makefile.PL to fix reported HP/UX feature where unresolved
906 externals still cause an executable to be generated (though no
907 x bit set). Thanks to David Kozinn for report and explanation.
908 Problems with the fix are mine :)
910 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
911 Probably buggy. I'm sure I'll hear.
913 1.17 Wed Jul 1 20:10 1998
914 - fix setitimer calls so microseconds is not more than 1000000.
915 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
916 - make Win32. We only get gettimeofday (the select hack doesn't
917 seem to work on my Win95 system).
918 - fix test 4 on 01test.t. add test to see if time() and
919 Time::HiRes::time() are close.
921 1.16 Wed Nov 12 21:05 1997
922 - add missing EXTEND in new gettimeofday scalar code.
924 1.16+ should be closer to building out of the box on Linux. Thanks
925 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
927 If your underlying operating system doesn't implement ualarm(), then
928 a fake using setitimer() will be made. If the OS is missing usleep(),
929 a fake one using select() will be made. If a fake can't be made for
930 either ualarm() or usleep(), then the corresponding Perl function will
931 not be available. If the OS is missing gettimeofday(), you will get
932 unresolved externals, either at link- or run-time.
934 This is an improvement; the package used to not even build if
935 you were missing any of these bits. Roderick Schertler
937 <roderick@argon.org> did all the conditional compilation stuff,
938 look at HiRes.pm and the test suites; it's good educational reading.
940 1.15 Mon Nov 10 21:30 1997
941 - HiRes.pm: update pod. Provided by Gisle Aas.
942 - HiRes.xs: if gettimeofday() called in scalar context, do
943 something more useful than before. Provided by Gisle Aas.
944 - README: tell of xsubpp '-nolinenumber' woes. thanks to
945 Edward Henigin <ed@texas.net> for pointing out the problem.
947 1.14 Wed Nov 5 9:40 1997
948 - Makefile.PL: look for setitimer
949 - HiRes.xs: if missing ualarm, but we have setitimer, make up
950 our own setitimer. These were provided by Gisle Aas.
952 1.13 Tue Nov 4 23:30 1997
953 - Makefile.PL: fix autodetect mechanism to do try linking in addition
954 to just compiling; should fix Linux build problem. Fix was provided
957 1.12 Sun Oct 12 12:00:00 1997
958 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
959 you may need to comment this back out if you have an older xsubpp.
960 - HiRes.xs: set PROTOTYPES: DISABLE
962 1.11 Fri Sep 05 16:00:00 1997
964 Had some line commented out that shouldn't have been (testing
967 Previous version was corrupted.
969 1.10 Thu May 22 20:20:00 1997
970 - HiRes.xs, HiRes.pm, t/*:
971 - only compile what we have OS support for (or can
973 - only test what we compiled
974 - gross improvement to the test suite
976 This work was all done by Roderick Schertler
977 <roderick@argon.org>. If you run Linux or
978 one of the other ualarm-less platforms, and you like this
979 module, let Roderick know; without him, it still wouldn't
980 be working on those boxes...
981 - Makefile.PL: figure out what routines the OS has and
982 only build what we need. These bits were written by Jarkko
983 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
985 1.02 Mon Dec 30 08:00:00 1996
986 - HiRes.pm: update documentation to say what to do when missing
987 ualarm() and friends.
988 - README: update to warn that ualarm() and friends need to exist
990 1.01 Fri Oct 17 08:00:00 1996
991 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
992 - HiRes.pm: put blank line between __END__ and =head1 so that
995 1.00 Tue Sep 03 13:00:00 1996
996 - original version; created by h2xs 1.16