1 Revision history for Perl extension Time::HiRes.
4 - try to be more robust and consistent in the detection of
5 CLOCK_REALTIME and ITIMER_VIRTUAL in HiRes.t: the proper
10 eval 'import Time::HiRes qw($symbol)';
11 return 0 unless $@ eq '';
12 return exists ${"Time::HiRes::$symbol"};
19 in the test. All these moves are needed because
21 1) one cannot directly do eval 'Time::HiRes::FOO_BAR'
22 because FOO_BAR might have a true value of zero
23 (or in the general case an empty string or even undef)
25 2) In case FOO_BAR is not available in this platform,
26 &FOO_BAR avoids the bareword warning
28 - wait more (1.5 seconds instead of 0.1) for the CLOCK_REALTIME test
29 but expect the 'customary' slop of 0.20 instead of 0.25
30 - fixed inside a comment HAS_POLL -> TIME_HIRES_NANOSLEEP
31 - at the end of HiRest.t tell how close we were to termination
34 - Gisle noticed a mistake (using HAS_NANOSLEEP) in 1.79
37 - try nanosleep for emulating usleep -- may help in some weird
38 embedded realtime places which have nanosleep but neither usleep
39 nor select nor poll (doesn't have to be weird embedded realtime
40 place, though -- in many places usleep is nanosleep anyway)
41 - try poll for emulating usleep -- this may help some obscure/old
42 SVR4 places that have neither usleep nor select
43 - a redundant test guard in HiRes.t
46 - ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
47 in the general case fail already at 'use' phase is suboptimal)
48 - fixes to the documentation of clock_gettime() and clock_getres()
51 - add support for the POSIX clock_gettime() and clock_getres(),
52 if available, either as library calls or as syscalls
53 - be more defensive about missing functionality: break out
54 early (during 'use') if no e.g. clock_getres() is available,
55 and protect our back by trapping those cases also in HiRes.xs
56 - the test added in 1.76 could cause an endless loop e.g. in Solaris,
57 due to mixing of sleep() and alarm() (bad programmer, no cookie!)
60 - testing for nanosleep had wrong logic which caused nanosleep
61 to become undefined for e.g. Mac OS X
62 - added a test for a core dump that was introduced by Perl 5.8.0
63 safe signals and was fixed for the time of 5.8.1 (one report of
64 the core dump was [perl #20920]), the test skipped pre-5.8.1.
65 - *cough* s/unanosleep/nanosleep/g; *cough*
68 - installation patch from Gisle Aas: in Perls 5.8.x and later
69 use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
72 - [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
73 (the workaround is not to use rel2abs, should not be necessary)
74 - [cpan #14642] U2time wrongly exported on the C API
75 (patch supplied by the reporter, SALVA@cpan.org)
76 - add release dates to Changes
79 - Time::HiRes::nanosleep support for Solaris [PATCH]
80 (POSIX::uname() not available if building with core perl,
81 from Gisle Aas, via perl5-porters, perl change #25295)
84 - going back to the 1.68 loader setup (using DynaLoader)
85 since too many weird things starting breaking
86 - fix a typo in José Auguste-Etienne's name
89 - a thinko in the nanosleep() detection
90 - move more changes stuff from the README to Changes
91 - add -w to the Makefile.PL
94 - oops in 1.69 about @ISA (not affecting anything but silly)
95 - add copyright 2005 to HiRes.pm
96 - add copyright and license to HiRes.xs
97 - add copyrights 2003, 2004, 2005 to README
100 - actually run a test for nanosleep
101 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
102 it seems that one can link in nanosleep() but then calling
103 it fails instantly and sets errno to ENOSYS (Not implemented).
104 This may be fixable in the AIX case by figuring out the right
105 (realtime POSIX?) libs and whatnot, but in the general case
106 running a real test case is better. (Of course, this change
107 will no doubt run into portability problems because of the
108 execution step...) Note that because of hysterical raisins
109 most Perls do NOT have $Config{d_nanosleep} (scanning for
110 it by Configure would in many platforms require linking in
111 things like -lrt, which would in many platforms be a bad idea
113 (from José Auguste-Etienne)
114 - support XSLoader also since it's much faster
115 (from Alexey Tourbin)
116 - add SEE ALSO (BSD::Resource and Time::TAI64)
119 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
120 error suspected), but miraculously it still worked since the
121 doubling took place below the __END__ token
122 - undef Pause() before defining it to avoid redefinition warnings
123 during compilation in case perl.h had already defined Pause()
124 (part of perl change #24271)
128 - (internal) don't ignore the return value of gettimeofday()
129 - (external) return undef or an empty if the C gettimeofday() fails
130 (affects Time::HiRes gettimeofday() and the hires time())
134 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
135 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
136 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
137 - close also [rt.cpan.org #5933] "Time::HiRes::time does not pick up time adjustments like ntp" since ever reproducing it in the same environment
138 has become rather unlikely
141 - one should not mix u?alarm and sleep (the tests modified
142 by 1.65, #12 and #13, hung in Solaris), now we just busy
143 loop executing an empty block
144 - in the documentation underline the unspecificity of mixing
146 - small spelling fixes
149 - regenerate ppport.h with Devel::PPPort 3.03,
150 now the MY_CXT_CLONE is defined in ppport.h,
151 we no more need to do that.
153 - the test #12 would often hang in sigsuspend() (at least that's
154 where Mac OS X' ktrace shows it hanging). With the sleep()s
155 changed to sleep(1)s, the tests still pass but no hang after
156 a few hundred repeats.
159 - Win32 and any ithread build: ppport.h didn't define
160 MY_CXT_CLONE, which seems to be a Time-HiResism.
163 - Skip testing if under PERL_CORE and Time::HiRes has not
164 been Configured (from Marcus Holland-Moritz, core change
166 - Use ppport.h generated by Devel::PPPort 3.01,
167 allowing cutting away our own portability code.
168 - Don't use $ENV{PERL_CORE} for < 5.6.0.
169 - Don't use "for my $i" for <= 5.003.
170 - Don't use Pause() for <= 5.003.
171 - Can't use newSVpvf for <= 5.003.
172 (most of the changes from Marcus)
175 - Win32: reset reading from the performance counters every
176 five minutes to better track wall clock time (thanks to
177 PC timers being often quite bad), should help long-running
181 - Win32: Patch from Steve Hay
182 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
183 to [perl #30755] reported by Nigel Sandever
185 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
186 <w32api/windows.h> APIs are available. Cygwin testing by
187 Yitzchak Scott-Thoennes.
189 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
190 after that keep using -lrt, patch from Alan Burlison,
191 bug reported in [cpan #7165]
194 - Change the Win32 recalibration limit to 0.5 seconds and tweak
195 the documentation to blather less about the gory details of the
196 Win32 implementation and more about the complications in general
197 of meddling with the system clock.
200 - Document the 1.57 change better.
203 - Win32/Cygwin/MinGW: if the performance counter drifts by more
204 than two seconds from the system clock (due to ntp adjustments,
205 for example), recalibrate our internal counter: from Jan Dubois,
206 based on [cpan #5933] by Jerry D. Hedden.
209 - Give a clearer message if the tests timeout (perl change #22253)
210 - Don't use /tmp or its moral equivalents (perl bug #15036,
214 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
215 instead of VC-specific i64 suffix)
218 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
221 - Windows: higher resolution time() by using the Windows
222 performance counter API, from Jan Dubois and Anton Shcherbinin.
223 The exact new higher resolution depends on the hardware,
224 but it should be quite a bit better than using the basic
228 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
229 the subtest 18. No known analysis nor fix, but added
230 an alarm (that requires fork() and alarm()) to the test.
233 - doc tweaks from mjd (perl change #20456)
234 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
237 - add a message (for non-core builds) to Makefile.PL about
238 the LC_ALL=C workaround
239 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
240 - the nanosleep() probe was broken (perl change #20061)
241 - use existence instead of definedness for feature probes
243 - MPE/iX tweak (perl change #20042)
244 - do not use HAS_NANOSLEEP (perl change #19898)
247 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
248 - OS/2 can always mix subsecond sleeps with signals
249 (part of perl change #19789)
252 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
256 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
257 - unnecessary (nanosleep is in libc anyway)
258 - harmful (-lrt slows down execution)
259 - incompatible (with many distributions' pthreads)
262 - do not create files in blib directories under core
263 (perl change #19160, from rgs)
264 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
267 - guarantee that $xdefine in HiRes.t is always defined
268 (perl change #19109, from IlyaZ)
269 - a cleaner way to detect PERL_CORE (perl change #19111,
273 - add hints/irix.pl to turn off overly POSIX flags that
274 cause hide struct timespec to be hidden (and compilation
275 to fail) (bleadperl change #19085)
276 - documentation tweaks
279 - add c:/temp to the list of temp directories to probe
280 so that cygwin (and win*?) builds are happy. This was
281 needed at least in my cygwin 1.3.20/w2k setup.
284 - modernize the constants code (from Nicholas Clark)
287 - At some point the ability to figure our the correct incdir
288 for EXTERN.h (either a core perl build, or an installed perl)
289 had broken (which lead into all test compiles failing with
290 a core perl build, but thanks to the robustness of Makefile.PL
291 nothing of this was visible). The brokenness seemed to be
292 caused by $ENV{PERL_CORE} not being on for core builds?
293 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
294 right, and both styles of build should work again.
297 - Nicholas Clark noticed that the my_catdir() emulation function
298 was broken (which means that we didn't really work for Perls
300 - inspired by fixing the above made the whole Makefile.PL -w
302 - tightened up the Makefile.PL output, less whitespace
305 - fix from Craig Berry for better building in VMS with PERL_CORE
308 - no functional changes
309 - move lib/Time/HiRes.pm as Hires.pm
310 - libraries scanning was slightly broken (always scanned
311 for a library even when $Config{libs} already had it)
314 - Ray Zimmerman ran into a race condition in Mac OS X.
315 A 0.01-second alarm fired before the test expected.
316 The test first slept indefinitely (blocking for signals)
317 and only after that tested for the signal having been sent.
318 Since the signal had already been sent, the test #12 never
319 completed. The solution: test first, then block.
320 - default to being silent on all probing attempts, set the
321 environment variable VERBOSE to a true value to see the
322 details (the probing command and the possible errors)
325 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
326 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
327 cannot be upgraded from CPAN to override the 5.8.0 version
329 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
330 because the include directories did not adjust themselves
331 if $ENV{PERL_CORE} (Hugo van der Sanden)
332 - add documentation about the restart of select() under alarm()
335 - small documentation tweaks
339 - better VMS operation (Craig Berry)
342 - our time machine is accelerating: now works with Perl 5.004_01
343 (tried with 5.003_07 and 5.002 but I get segmentation faults
344 from running the Makefile.PL with those in Tru64 4.0D)
347 - backward compatibility (pre-5.6.0) tweaks:
348 - no XSLoader in 5.00503, use DynaLoader instead
349 - no SvPV_nolen, either
350 - no PerlProc_pause(), either
351 - now tested with 5.00404 and 5.00503
352 - Makefile.PL requires 5.00404 (no more 5.002)
353 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
354 (this means that one can mix subsecond sleeps with alarms)
355 - because of nanosleep we probe for -lrt and -lposix4
356 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
357 is available by exportable constants Time::HiRes::d_func
358 (since older Perl do not have them in %Config, and even
359 5.8.0 does not probe for nanosleep)
362 - backward compatibility (pre-5.6.1) tweaks:
363 - define NV if no NVTYPE
364 - define IVdf if needed (note: the Devel::PPPort
365 in 5.8.0 does not try hard hard enough since
366 the IVSIZE might not be defined)
367 - define NVgf if needed
368 - grab the typemap from 5.8.0 for the NV stuff
370 1.31 and 1.32 add more backward compatibility (now all the way
371 back to Perl 5.00404), and using nanosleep() (if available) for
376 - release 1.29_02 as 1.30
378 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
379 development cycle. Most notably portability across platforms has been
380 enhanced, and the interval timers (setitimer, getitimer) have been
381 added. Note that the version of Time::HiRes that is included in Perl
382 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
383 version. Note also that in 1.30 Wegscheid turns over the maintenance
384 to Jarkko Hietaniemi.
388 - fix a silly unclosed comment typo in HiRes.xs
389 - document and export REALTIME_REALPROF (Solaris)
393 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
394 (need to patch this also in Perl 5.[89])
395 - remove CVS revision log from HiRes.xs
399 The following numbered patches refer to the Perl 5.7 changes,
400 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
402 - 17558: Add #!./perl to the .t
403 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
404 - 16198: political correctness, from Simon Cozens
405 - 15857: doc tweaks, from Jarkko Hietaniemi
406 - 15593: optimization in .xs, from Paul Green
407 - 14892: pod fixes, from Robin Barker
408 - 14100: VOS fixes, from Paul Green
409 - 13422: XS segfault, from Marc Lehmann
410 - 13378: whether select() gets restarted on signals, depends
411 - 13354: timing constraints, again, from Andy Dougherty
412 - 13278: can't do subecond alarms with ualarm;
413 break out early if alarms do not seem to be working
414 - 13266: test relaxation (cygwin gets lower hires
415 times than lores ones)
416 - 12846: protect against high load, from Jarkko Hietaniemi
417 - 12837: HiRes.t VMS tweak, from Craig A. Berry
418 - 12797: HiRes.t VMS tweak, from Charles Lane
419 - 12769: HiRes.t VMS tweak, from Craig A. Berry
420 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
421 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
422 - 12692: alarm() ain't gonna work if ualarm() ain't,
423 from Gurusamy Sarathy
424 - 12680: minor VMS tweak, from Charles Lane
425 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
426 - 12609: croak on negative time, from Jarkko Hietaniemi
427 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
428 - 12594: MacOS Classic timeofday, from Chris Nandor
429 - 12473: allow for more than one second for sleep() and usleep()
430 - 12458: test tuning, relax timing constraints,
431 from Jarkko Hietaniemi
432 - 12449: make sleep() and usleep() to return the number
433 of seconds and microseconds actually slept (analogously
434 with the builtin sleep()), also make usleep() croak if
435 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
436 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
437 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
438 - 12196: use ftime() on Win32, from Artur Bergman
439 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
440 - 12105: use GetSystemTime() on Win32, from Artur Bergman
441 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
442 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
443 - 11797: problem in HiRes.t, from John P. Linderman
444 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
445 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
446 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
447 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
448 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
449 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
451 - 10942: MPE/IX test tweaks, from Mark Bixby
452 - 10784: unnecessary pod2man calls, from Andy Dougherty
453 - 10354: ext/ + -Wall, from Doug MacEachern
454 - 10320: fix the BOOT section to call myU2time correctly
455 - 10317: correct casting for AIX< from H. Merijn Brand
456 - 10119: document that the core time() may be rounding, not truncating
457 - 10118: test fix, from John Peacock
458 - 9988: long =item, from Robin Barker
459 - 9714: correct test output
460 - 9708: test also the scalar aspect of getitimer()
461 - 9705: Add interval timers (setitimer, getitimer)
462 - 9692: do not require at least 5.005 using XS
464 The following changes were made on top of the changes
465 made for Time::HiRes during the Perl 5.7 development
466 cycle that culminated in the release of Perl 5.8.0.
468 - add "require 5.005" to the Makefile.PL
469 - remove the REVISION section (CVS log) from HiRes.pm
470 - add jhi's copyright alongside Douglas'
471 - move HiRes.pm to lib/Time/
473 - modify HiRes.t to use $ENV{PERL_CORE}
474 - modify the original Time::HiRes version 1.20 Makefile.PL
475 to work both with Perl 5.8.0 and the new code with pre-5.8.0
476 Perls (tried with 5.6.1)
477 - tiny tweaks and updates in README and TODO
478 - bump the VERSION to 1.29
480 1.20 Wed Feb 24 21:30 1999
481 - make our usleep and ualarm substitutes into hrt_usleep
482 and hrt_ualarm. This helps static links of Perl with other
483 packages that also have usleep, etc. From
484 Ilya Zakharevich <ilya@math.ohio-state.edu>
485 - add C API stuff. From Joshua Pritikin
486 <joshua.pritikin@db.com>
487 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
488 - hopefully correct "-lc" fix for SCO.
491 1.20 adds a platform neutral set of C accessible routines if you are
492 running 5.005+. All other changes are packaging changes and build
493 fixes(?) for statically linked Perl, SCO, and VMS.
495 1.19 Tue Sep 29 22:30 1998
496 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
497 <seb@stian.demon.co.uk>
498 - change GIMME_V to GIMME to help people with older versions of
500 - fix Win32 version of gettimeofday(). It didn't affect anything,
501 but it confuses people reading the code when the return value
502 is backwards (0 is success).
503 - fix Makefile.PL (more) so that detection of gettimeofday is
506 1.19 has better VMS support.
508 1.18 Mon Jul 6 22:40 1998
509 - add usleep() for Win32.
510 - fix Makefile.PL to fix reported HP/UX feature where unresolved
511 externals still cause an executable to be generated (though no
512 x bit set). Thanks to David Kozinn for report and explanation.
513 Problems with the fix are mine :)
515 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
516 Probably buggy. I'm sure I'll hear.
518 1.17 Wed Jul 1 20:10 1998
519 - fix setitimer calls so microseconds is not more than 1000000.
520 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
521 - make Win32. We only get gettimeofday (the select hack doesn't
522 seem to work on my Win95 system).
523 - fix test 4 on 01test.t. add test to see if time() and
524 Time::HiRes::time() are close.
526 1.16 Wed Nov 12 21:05 1997
527 - add missing EXTEND in new gettimeofday scalar code.
529 1.16+ should be closer to building out of the box on Linux. Thanks
530 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
532 If your underlying operating system doesn't implement ualarm(), then
533 a fake using setitimer() will be made. If the OS is missing usleep(),
534 a fake one using select() will be made. If a fake can't be made for
535 either ualarm() or usleep(), then the corresponding Perl function will
536 not be available. If the OS is missing gettimeofday(), you will get
537 unresolved externals, either at link- or run-time.
539 This is an improvement; the package used to not even build if
540 you were missing any of these bits. Roderick Schertler
542 <roderick@argon.org> did all the conditional compilation stuff,
543 look at HiRes.pm and the test suites; it's good educational reading.
545 1.15 Mon Nov 10 21:30 1997
546 - HiRes.pm: update pod. Provided by Gisle Aas.
547 - HiRes.xs: if gettimeofday() called in scalar context, do
548 something more useful than before. Provided by Gisle Aas.
549 - README: tell of xsubpp '-nolinenumber' woes. thanks to
550 Edward Henigin <ed@texas.net> for pointing out the problem.
552 1.14 Wed Nov 5 9:40 1997
553 - Makefile.PL: look for setitimer
554 - HiRes.xs: if missing ualarm, but we have setitimer, make up
555 our own setitimer. These were provided by Gisle Aas.
557 1.13 Tue Nov 4 23:30 1997
558 - Makefile.PL: fix autodetect mechanism to do try linking in addition
559 to just compiling; should fix Linux build problem. Fix was provided
562 1.12 Sun Oct 12 12:00:00 1997
563 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
564 you may need to comment this back out if you have an older xsubpp.
565 - HiRes.xs: set PROTOTYPES: DISABLE
567 1.11 Fri Sep 05 16:00:00 1997
569 Had some line commented out that shouldn't have been (testing
572 Previous version was corrupted.
574 1.10 Thu May 22 20:20:00 1997
575 - HiRes.xs, HiRes.pm, t/*:
576 - only compile what we have OS support for (or can
578 - only test what we compiled
579 - gross improvement to the test suite
581 This work was all done by Roderick Schertler
582 <roderick@argon.org>. If you run Linux or
583 one of the other ualarm-less platoforms, and you like this
584 module, let Roderick know; without him, it still wouldn't
585 be working on those boxes...
586 - Makefile.PL: figure out what routines the OS has and
587 only build what we need. These bits were written by Jarkko
588 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
590 1.02 Mon Dec 30 08:00:00 1996
591 - HiRes.pm: update documentation to say what to do when missing
592 ualarm() and friends.
593 - README: update to warn that ualarm() and friends need to exist
595 1.01 Fri Oct 17 08:00:00 1996
596 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
597 - HiRes.pm: put blank line between __END__ and =head1 so that
600 1.00 Tue Sep 03 13:00:00 1996
601 - original version; created by h2xs 1.16