1 Revision history for Perl extension Time::HiRes.
4 - try nanosleep for emulating usleep -- may help in some weird
5 embedded realtime places which have nanosleep but neither usleep
6 nor select nor poll (doesn't have to be weird embedded realtime
7 place, though -- in many places usleep is nanosleep anyway)
8 - try poll for emulating usleep -- this may help some obscure/old
9 SVR4 places that have neither usleep nor select
10 - a redundant test guard in HiRes.t
13 - ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
14 in the general case fail already at 'use' phase is suboptimal)
15 - fixes to the documentation of clock_gettime() and clock_getres()
18 - add support for the POSIX clock_gettime() and clock_getres(),
19 if available, either as library calls or as syscalls
20 - be more defensive about missing functionality: break out
21 early (during 'use') if no e.g. clock_getres() is available,
22 and protect our back by trapping those cases also in HiRes.xs
23 - the test added in 1.76 could cause an endless loop e.g. in Solaris,
24 due to mixing of sleep() and alarm() (bad programmer, no cookie!)
27 - testing for nanosleep had wrong logic which caused nanosleep
28 to become undefined for e.g. Mac OS X
29 - added a test for a core dump that was introduced by Perl 5.8.0
30 safe signals and was fixed for the time of 5.8.1 (one report of
31 the core dump was [perl #20920]), the test skipped pre-5.8.1.
32 - *cough* s/unanosleep/nanosleep/g; *cough*
35 - installation patch from Gisle Aas: in Perls 5.8.x and later
36 use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
39 - [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
40 (the workaround is not to use rel2abs, should not be necessary)
41 - [cpan #14642] U2time wrongly exported on the C API
42 (patch supplied by the reporter, SALVA@cpan.org)
43 - add release dates to Changes
46 - Time::HiRes::nanosleep support for Solaris [PATCH]
47 (POSIX::uname() not available if building with core perl,
48 from Gisle Aas, via perl5-porters, perl change #25295)
51 - going back to the 1.68 loader setup (using DynaLoader)
52 since too many weird things starting breaking
53 - fix a typo in José Auguste-Etienne's name
56 - a thinko in the nanosleep() detection
57 - move more changes stuff from the README to Changes
58 - add -w to the Makefile.PL
61 - oops in 1.69 about @ISA (not affecting anything but silly)
62 - add copyright 2005 to HiRes.pm
63 - add copyright and license to HiRes.xs
64 - add copyrights 2003, 2004, 2005 to README
67 - actually run a test for nanosleep
68 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
69 it seems that one can link in nanosleep() but then calling
70 it fails instantly and sets errno to ENOSYS (Not implemented).
71 This may be fixable in the AIX case by figuring out the right
72 (realtime POSIX?) libs and whatnot, but in the general case
73 running a real test case is better. (Of course, this change
74 will no doubt run into portability problems because of the
75 execution step...) Note that because of hysterical raisins
76 most Perls do NOT have $Config{d_nanosleep} (scanning for
77 it by Configure would in many platforms require linking in
78 things like -lrt, which would in many platforms be a bad idea
80 (from José Auguste-Etienne)
81 - support XSLoader also since it's much faster
83 - add SEE ALSO (BSD::Resource and Time::TAI64)
86 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
87 error suspected), but miraculously it still worked since the
88 doubling took place below the __END__ token
89 - undef Pause() before defining it to avoid redefinition warnings
90 during compilation in case perl.h had already defined Pause()
91 (part of perl change #24271)
95 - (internal) don't ignore the return value of gettimeofday()
96 - (external) return undef or an empty if the C gettimeofday() fails
97 (affects Time::HiRes gettimeofday() and the hires time())
101 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
102 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
103 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
104 - 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
105 has become rather unlikely
108 - one should not mix u?alarm and sleep (the tests modified
109 by 1.65, #12 and #13, hung in Solaris), now we just busy
110 loop executing an empty block
111 - in the documentation underline the unspecificity of mixing
113 - small spelling fixes
116 - regenerate ppport.h with Devel::PPPort 3.03,
117 now the MY_CXT_CLONE is defined in ppport.h,
118 we no more need to do that.
120 - the test #12 would often hang in sigsuspend() (at least that's
121 where Mac OS X' ktrace shows it hanging). With the sleep()s
122 changed to sleep(1)s, the tests still pass but no hang after
123 a few hundred repeats.
126 - Win32 and any ithread build: ppport.h didn't define
127 MY_CXT_CLONE, which seems to be a Time-HiResism.
130 - Skip testing if under PERL_CORE and Time::HiRes has not
131 been Configured (from Marcus Holland-Moritz, core change
133 - Use ppport.h generated by Devel::PPPort 3.01,
134 allowing cutting away our own portability code.
135 - Don't use $ENV{PERL_CORE} for < 5.6.0.
136 - Don't use "for my $i" for <= 5.003.
137 - Don't use Pause() for <= 5.003.
138 - Can't use newSVpvf for <= 5.003.
139 (most of the changes from Marcus)
142 - Win32: reset reading from the performance counters every
143 five minutes to better track wall clock time (thanks to
144 PC timers being often quite bad), should help long-running
148 - Win32: Patch from Steve Hay
149 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
150 to [perl #30755] reported by Nigel Sandever
152 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
153 <w32api/windows.h> APIs are available. Cygwin testing by
154 Yitzchak Scott-Thoennes.
156 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
157 after that keep using -lrt, patch from Alan Burlison,
158 bug reported in [cpan #7165]
161 - Change the Win32 recalibration limit to 0.5 seconds and tweak
162 the documentation to blather less about the gory details of the
163 Win32 implementation and more about the complications in general
164 of meddling with the system clock.
167 - Document the 1.57 change better.
170 - Win32/Cygwin/MinGW: if the performance counter drifts by more
171 than two seconds from the system clock (due to ntp adjustments,
172 for example), recalibrate our internal counter: from Jan Dubois,
173 based on [cpan #5933] by Jerry D. Hedden.
176 - Give a clearer message if the tests timeout (perl change #22253)
177 - Don't use /tmp or its moral equivalents (perl bug #15036,
181 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
182 instead of VC-specific i64 suffix)
185 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
188 - Windows: higher resolution time() by using the Windows
189 performance counter API, from Jan Dubois and Anton Shcherbinin.
190 The exact new higher resolution depends on the hardware,
191 but it should be quite a bit better than using the basic
195 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
196 the subtest 18. No known analysis nor fix, but added
197 an alarm (that requires fork() and alarm()) to the test.
200 - doc tweaks from mjd (perl change #20456)
201 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
204 - add a message (for non-core builds) to Makefile.PL about
205 the LC_ALL=C workaround
206 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
207 - the nanosleep() probe was broken (perl change #20061)
208 - use existence instead of definedness for feature probes
210 - MPE/iX tweak (perl change #20042)
211 - do not use HAS_NANOSLEEP (perl change #19898)
214 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
215 - OS/2 can always mix subsecond sleeps with signals
216 (part of perl change #19789)
219 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
223 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
224 - unnecessary (nanosleep is in libc anyway)
225 - harmful (-lrt slows down execution)
226 - incompatible (with many distributions' pthreads)
229 - do not create files in blib directories under core
230 (perl change #19160, from rgs)
231 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
234 - guarantee that $xdefine in HiRes.t is always defined
235 (perl change #19109, from IlyaZ)
236 - a cleaner way to detect PERL_CORE (perl change #19111,
240 - add hints/irix.pl to turn off overly POSIX flags that
241 cause hide struct timespec to be hidden (and compilation
242 to fail) (bleadperl change #19085)
243 - documentation tweaks
246 - add c:/temp to the list of temp directories to probe
247 so that cygwin (and win*?) builds are happy. This was
248 needed at least in my cygwin 1.3.20/w2k setup.
251 - modernize the constants code (from Nicholas Clark)
254 - At some point the ability to figure our the correct incdir
255 for EXTERN.h (either a core perl build, or an installed perl)
256 had broken (which lead into all test compiles failing with
257 a core perl build, but thanks to the robustness of Makefile.PL
258 nothing of this was visible). The brokenness seemed to be
259 caused by $ENV{PERL_CORE} not being on for core builds?
260 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
261 right, and both styles of build should work again.
264 - Nicholas Clark noticed that the my_catdir() emulation function
265 was broken (which means that we didn't really work for Perls
267 - inspired by fixing the above made the whole Makefile.PL -w
269 - tightened up the Makefile.PL output, less whitespace
272 - fix from Craig Berry for better building in VMS with PERL_CORE
275 - no functional changes
276 - move lib/Time/HiRes.pm as Hires.pm
277 - libraries scanning was slightly broken (always scanned
278 for a library even when $Config{libs} already had it)
281 - Ray Zimmerman ran into a race condition in Mac OS X.
282 A 0.01-second alarm fired before the test expected.
283 The test first slept indefinitely (blocking for signals)
284 and only after that tested for the signal having been sent.
285 Since the signal had already been sent, the test #12 never
286 completed. The solution: test first, then block.
287 - default to being silent on all probing attempts, set the
288 environment variable VERBOSE to a true value to see the
289 details (the probing command and the possible errors)
292 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
293 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
294 cannot be upgraded from CPAN to override the 5.8.0 version
296 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
297 because the include directories did not adjust themselves
298 if $ENV{PERL_CORE} (Hugo van der Sanden)
299 - add documentation about the restart of select() under alarm()
302 - small documentation tweaks
306 - better VMS operation (Craig Berry)
309 - our time machine is accelerating: now works with Perl 5.004_01
310 (tried with 5.003_07 and 5.002 but I get segmentation faults
311 from running the Makefile.PL with those in Tru64 4.0D)
314 - backward compatibility (pre-5.6.0) tweaks:
315 - no XSLoader in 5.00503, use DynaLoader instead
316 - no SvPV_nolen, either
317 - no PerlProc_pause(), either
318 - now tested with 5.00404 and 5.00503
319 - Makefile.PL requires 5.00404 (no more 5.002)
320 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
321 (this means that one can mix subsecond sleeps with alarms)
322 - because of nanosleep we probe for -lrt and -lposix4
323 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
324 is available by exportable constants Time::HiRes::d_func
325 (since older Perl do not have them in %Config, and even
326 5.8.0 does not probe for nanosleep)
329 - backward compatibility (pre-5.6.1) tweaks:
330 - define NV if no NVTYPE
331 - define IVdf if needed (note: the Devel::PPPort
332 in 5.8.0 does not try hard hard enough since
333 the IVSIZE might not be defined)
334 - define NVgf if needed
335 - grab the typemap from 5.8.0 for the NV stuff
337 1.31 and 1.32 add more backward compatibility (now all the way
338 back to Perl 5.00404), and using nanosleep() (if available) for
343 - release 1.29_02 as 1.30
345 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
346 development cycle. Most notably portability across platforms has been
347 enhanced, and the interval timers (setitimer, getitimer) have been
348 added. Note that the version of Time::HiRes that is included in Perl
349 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
350 version. Note also that in 1.30 Wegscheid turns over the maintenance
351 to Jarkko Hietaniemi.
355 - fix a silly unclosed comment typo in HiRes.xs
356 - document and export REALTIME_REALPROF (Solaris)
360 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
361 (need to patch this also in Perl 5.[89])
362 - remove CVS revision log from HiRes.xs
366 The following numbered patches refer to the Perl 5.7 changes,
367 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
369 - 17558: Add #!./perl to the .t
370 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
371 - 16198: political correctness, from Simon Cozens
372 - 15857: doc tweaks, from Jarkko Hietaniemi
373 - 15593: optimization in .xs, from Paul Green
374 - 14892: pod fixes, from Robin Barker
375 - 14100: VOS fixes, from Paul Green
376 - 13422: XS segfault, from Marc Lehmann
377 - 13378: whether select() gets restarted on signals, depends
378 - 13354: timing constraints, again, from Andy Dougherty
379 - 13278: can't do subecond alarms with ualarm;
380 break out early if alarms do not seem to be working
381 - 13266: test relaxation (cygwin gets lower hires
382 times than lores ones)
383 - 12846: protect against high load, from Jarkko Hietaniemi
384 - 12837: HiRes.t VMS tweak, from Craig A. Berry
385 - 12797: HiRes.t VMS tweak, from Charles Lane
386 - 12769: HiRes.t VMS tweak, from Craig A. Berry
387 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
388 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
389 - 12692: alarm() ain't gonna work if ualarm() ain't,
390 from Gurusamy Sarathy
391 - 12680: minor VMS tweak, from Charles Lane
392 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
393 - 12609: croak on negative time, from Jarkko Hietaniemi
394 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
395 - 12594: MacOS Classic timeofday, from Chris Nandor
396 - 12473: allow for more than one second for sleep() and usleep()
397 - 12458: test tuning, relax timing constraints,
398 from Jarkko Hietaniemi
399 - 12449: make sleep() and usleep() to return the number
400 of seconds and microseconds actually slept (analogously
401 with the builtin sleep()), also make usleep() croak if
402 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
403 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
404 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
405 - 12196: use ftime() on Win32, from Artur Bergman
406 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
407 - 12105: use GetSystemTime() on Win32, from Artur Bergman
408 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
409 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
410 - 11797: problem in HiRes.t, from John P. Linderman
411 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
412 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
413 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
414 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
415 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
416 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
418 - 10942: MPE/IX test tweaks, from Mark Bixby
419 - 10784: unnecessary pod2man calls, from Andy Dougherty
420 - 10354: ext/ + -Wall, from Doug MacEachern
421 - 10320: fix the BOOT section to call myU2time correctly
422 - 10317: correct casting for AIX< from H. Merijn Brand
423 - 10119: document that the core time() may be rounding, not truncating
424 - 10118: test fix, from John Peacock
425 - 9988: long =item, from Robin Barker
426 - 9714: correct test output
427 - 9708: test also the scalar aspect of getitimer()
428 - 9705: Add interval timers (setitimer, getitimer)
429 - 9692: do not require at least 5.005 using XS
431 The following changes were made on top of the changes
432 made for Time::HiRes during the Perl 5.7 development
433 cycle that culminated in the release of Perl 5.8.0.
435 - add "require 5.005" to the Makefile.PL
436 - remove the REVISION section (CVS log) from HiRes.pm
437 - add jhi's copyright alongside Douglas'
438 - move HiRes.pm to lib/Time/
440 - modify HiRes.t to use $ENV{PERL_CORE}
441 - modify the original Time::HiRes version 1.20 Makefile.PL
442 to work both with Perl 5.8.0 and the new code with pre-5.8.0
443 Perls (tried with 5.6.1)
444 - tiny tweaks and updates in README and TODO
445 - bump the VERSION to 1.29
447 1.20 Wed Feb 24 21:30 1999
448 - make our usleep and ualarm substitutes into hrt_usleep
449 and hrt_ualarm. This helps static links of Perl with other
450 packages that also have usleep, etc. From
451 Ilya Zakharevich <ilya@math.ohio-state.edu>
452 - add C API stuff. From Joshua Pritikin
453 <joshua.pritikin@db.com>
454 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
455 - hopefully correct "-lc" fix for SCO.
458 1.20 adds a platform neutral set of C accessible routines if you are
459 running 5.005+. All other changes are packaging changes and build
460 fixes(?) for statically linked Perl, SCO, and VMS.
462 1.19 Tue Sep 29 22:30 1998
463 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
464 <seb@stian.demon.co.uk>
465 - change GIMME_V to GIMME to help people with older versions of
467 - fix Win32 version of gettimeofday(). It didn't affect anything,
468 but it confuses people reading the code when the return value
469 is backwards (0 is success).
470 - fix Makefile.PL (more) so that detection of gettimeofday is
473 1.19 has better VMS support.
475 1.18 Mon Jul 6 22:40 1998
476 - add usleep() for Win32.
477 - fix Makefile.PL to fix reported HP/UX feature where unresolved
478 externals still cause an executable to be generated (though no
479 x bit set). Thanks to David Kozinn for report and explanation.
480 Problems with the fix are mine :)
482 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
483 Probably buggy. I'm sure I'll hear.
485 1.17 Wed Jul 1 20:10 1998
486 - fix setitimer calls so microseconds is not more than 1000000.
487 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
488 - make Win32. We only get gettimeofday (the select hack doesn't
489 seem to work on my Win95 system).
490 - fix test 4 on 01test.t. add test to see if time() and
491 Time::HiRes::time() are close.
493 1.16 Wed Nov 12 21:05 1997
494 - add missing EXTEND in new gettimeofday scalar code.
496 1.16+ should be closer to building out of the box on Linux. Thanks
497 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
499 If your underlying operating system doesn't implement ualarm(), then
500 a fake using setitimer() will be made. If the OS is missing usleep(),
501 a fake one using select() will be made. If a fake can't be made for
502 either ualarm() or usleep(), then the corresponding Perl function will
503 not be available. If the OS is missing gettimeofday(), you will get
504 unresolved externals, either at link- or run-time.
506 This is an improvement; the package used to not even build if
507 you were missing any of these bits. Roderick Schertler
509 <roderick@argon.org> did all the conditional compilation stuff,
510 look at HiRes.pm and the test suites; it's good educational reading.
512 1.15 Mon Nov 10 21:30 1997
513 - HiRes.pm: update pod. Provided by Gisle Aas.
514 - HiRes.xs: if gettimeofday() called in scalar context, do
515 something more useful than before. Provided by Gisle Aas.
516 - README: tell of xsubpp '-nolinenumber' woes. thanks to
517 Edward Henigin <ed@texas.net> for pointing out the problem.
519 1.14 Wed Nov 5 9:40 1997
520 - Makefile.PL: look for setitimer
521 - HiRes.xs: if missing ualarm, but we have setitimer, make up
522 our own setitimer. These were provided by Gisle Aas.
524 1.13 Tue Nov 4 23:30 1997
525 - Makefile.PL: fix autodetect mechanism to do try linking in addition
526 to just compiling; should fix Linux build problem. Fix was provided
529 1.12 Sun Oct 12 12:00:00 1997
530 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
531 you may need to comment this back out if you have an older xsubpp.
532 - HiRes.xs: set PROTOTYPES: DISABLE
534 1.11 Fri Sep 05 16:00:00 1997
536 Had some line commented out that shouldn't have been (testing
539 Previous version was corrupted.
541 1.10 Thu May 22 20:20:00 1997
542 - HiRes.xs, HiRes.pm, t/*:
543 - only compile what we have OS support for (or can
545 - only test what we compiled
546 - gross improvement to the test suite
548 This work was all done by Roderick Schertler
549 <roderick@argon.org>. If you run Linux or
550 one of the other ualarm-less platoforms, and you like this
551 module, let Roderick know; without him, it still wouldn't
552 be working on those boxes...
553 - Makefile.PL: figure out what routines the OS has and
554 only build what we need. These bits were written by Jarkko
555 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
557 1.02 Mon Dec 30 08:00:00 1996
558 - HiRes.pm: update documentation to say what to do when missing
559 ualarm() and friends.
560 - README: update to warn that ualarm() and friends need to exist
562 1.01 Fri Oct 17 08:00:00 1996
563 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
564 - HiRes.pm: put blank line between __END__ and =head1 so that
567 1.00 Tue Sep 03 13:00:00 1996
568 - original version; created by h2xs 1.16