1 Revision history for Perl extension Time::HiRes.
3 1.72 - going back to the 1.68 loader setup (using DynaLoader)
4 since too many weird things starting breaking
5 - fix a typo in José Auguste-Etienne's name
7 1.71 - a thinko in the nanosleep() detection
8 - move more changes stuff from the README to Changes
9 - add -w to the Makefile.PL
11 1.70 - oops in 1.69 about @ISA (not affecting anything but silly)
12 - add copyright 2005 to HiRes.pm
13 - add copyright and license to HiRes.xs
14 - add copyrights 2003, 2004, 2005 to README
16 1.69 - actually run a test for nanosleep
17 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
18 it seems that one can link in nanosleep() but then calling
19 it fails instantly and sets errno to ENOSYS (Not implemented).
20 This may be fixable in the AIX case by figuring out the right
21 (realtime POSIX?) libs and whatnot, but in the general case
22 running a real test case is better. (Of course, this change
23 will no doubt run into portability problems because of the
24 execution step...) Note that because of hysterical raisins
25 most Perls do NOT have $Config{d_nanosleep} (scanning for
26 it by Configure would in many platforms require linking in
27 things like -lrt, which would in many platforms be a bad idea
29 (from José Auguste-Etienne)
30 - support XSLoader also since it's much faster
32 - add SEE ALSO (BSD::Resource and Time::TAI64)
35 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
36 error suspected), but miraculously it still worked since the
37 doubling took place below the __END__ token
38 - undef Pause() before defining it to avoid redefinition warnings
39 during compilation in case perl.h had already defined Pause()
40 (part of perl change #24271)
44 - (internal) don't ignore the return value of gettimeofday()
45 - (external) return undef or an empty if the C gettimeofday() fails
46 (affects Time::HiRes gettimeofday() and the hires time())
50 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
51 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
52 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
53 - 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
54 has become rather unlikely
57 - one should not mix u?alarm and sleep (the tests modified
58 by 1.65, #12 and #13, hung in Solaris), now we just busy
59 loop executing an empty block
60 - in the documentation underline the unspecificity of mixing
62 - small spelling fixes
65 - regenerate ppport.h with Devel::PPPort 3.03,
66 now the MY_CXT_CLONE is defined in ppport.h,
67 we no more need to do that.
69 - the test #12 would often hang in sigsuspend() (at least that's
70 where Mac OS X' ktrace shows it hanging). With the sleep()s
71 changed to sleep(1)s, the tests still pass but no hang after
72 a few hundred repeats.
75 - Win32 and any ithread build: ppport.h didn't define
76 MY_CXT_CLONE, which seems to be a Time-HiResism.
79 - Skip testing if under PERL_CORE and Time::HiRes has not
80 been Configured (from Marcus Holland-Moritz, core change
82 - Use ppport.h generated by Devel::PPPort 3.01,
83 allowing cutting away our own portability code.
84 - Don't use $ENV{PERL_CORE} for < 5.6.0.
85 - Don't use "for my $i" for <= 5.003.
86 - Don't use Pause() for <= 5.003.
87 - Can't use newSVpvf for <= 5.003.
88 (most of the changes from Marcus)
91 - Win32: reset reading from the performance counters every
92 five minutes to better track wall clock time (thanks to
93 PC timers being often quite bad), should help long-running
97 - Win32: Patch from Steve Hay
98 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
99 to [perl #30755] reported by Nigel Sandever
101 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
102 <w32api/windows.h> APIs are available. Cygwin testing by
103 Yitzchak Scott-Thoennes.
105 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
106 after that keep using -lrt, patch from Alan Burlison,
107 bug reported in [cpan #7165]
110 - Change the Win32 recalibration limit to 0.5 seconds and tweak
111 the documentation to blather less about the gory details of the
112 Win32 implementation and more about the complications in general
113 of meddling with the system clock.
116 - Document the 1.57 change better.
119 - Win32/Cygwin/MinGW: if the performance counter drifts by more
120 than two seconds from the system clock (due to ntp adjustments,
121 for example), recalibrate our internal counter: from Jan Dubois,
122 based on [cpan #5933] by Jerry D. Hedden.
125 - Give a clearer message if the tests timeout (perl change #22253)
126 - Don't use /tmp or its moral equivalents (perl bug #15036,
130 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
131 instead of VC-specific i64 suffix)
134 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
137 - Windows: higher resolution time() by using the Windows
138 performance counter API, from Jan Dubois and Anton Shcherbinin.
139 The exact new higher resolution depends on the hardware,
140 but it should be quite a bit better than using the basic
144 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
145 the subtest 18. No known analysis nor fix, but added
146 an alarm (that requires fork() and alarm()) to the test.
149 - doc tweaks from mjd (perl change #20456)
150 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
153 - add a message (for non-core builds) to Makefile.PL about
154 the LC_ALL=C workaround
155 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
156 - the nanosleep() probe was broken (perl change #20061)
157 - use existence instead of definedness for feature probes
159 - MPE/iX tweak (perl change #20042)
160 - do not use HAS_NANOSLEEP (perl change #19898)
163 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
164 - OS/2 can always mix subsecond sleeps with signals
165 (part of perl change #19789)
168 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
172 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
173 - unnecessary (nanosleep is in libc anyway)
174 - harmful (-lrt slows down execution)
175 - incompatible (with many distributions' pthreads)
178 - do not create files in blib directories under core
179 (perl change #19160, from rgs)
180 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
183 - guarantee that $xdefine in HiRes.t is always defined
184 (perl change #19109, from IlyaZ)
185 - a cleaner way to detect PERL_CORE (perl change #19111,
189 - add hints/irix.pl to turn off overly POSIX flags that
190 cause hide struct timespec to be hidden (and compilation
191 to fail) (bleadperl change #19085)
192 - documentation tweaks
195 - add c:/temp to the list of temp directories to probe
196 so that cygwin (and win*?) builds are happy. This was
197 needed at least in my cygwin 1.3.20/w2k setup.
200 - modernize the constants code (from Nicholas Clark)
203 - At some point the ability to figure our the correct incdir
204 for EXTERN.h (either a core perl build, or an installed perl)
205 had broken (which lead into all test compiles failing with
206 a core perl build, but thanks to the robustness of Makefile.PL
207 nothing of this was visible). The brokenness seemed to be
208 caused by $ENV{PERL_CORE} not being on for core builds?
209 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
210 right, and both styles of build should work again.
213 - Nicholas Clark noticed that the my_catdir() emulation function
214 was broken (which means that we didn't really work for Perls
216 - inspired by fixing the above made the whole Makefile.PL -w
218 - tightened up the Makefile.PL output, less whitespace
221 - fix from Craig Berry for better building in VMS with PERL_CORE
224 - no functional changes
225 - move lib/Time/HiRes.pm as Hires.pm
226 - libraries scanning was slightly broken (always scanned
227 for a library even when $Config{libs} already had it)
230 - Ray Zimmerman ran into a race condition in Mac OS X.
231 A 0.01-second alarm fired before the test expected.
232 The test first slept indefinitely (blocking for signals)
233 and only after that tested for the signal having been sent.
234 Since the signal had already been sent, the test #12 never
235 completed. The solution: test first, then block.
236 - default to being silent on all probing attempts, set the
237 environment variable VERBOSE to a true value to see the
238 details (the probing command and the possible errors)
241 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
242 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
243 cannot be upgraded from CPAN to override the 5.8.0 version
245 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
246 because the include directories did not adjust themselves
247 if $ENV{PERL_CORE} (Hugo van der Sanden)
248 - add documentation about the restart of select() under alarm()
251 - small documentation tweaks
255 - better VMS operation (Craig Berry)
258 - our time machine is accelerating: now works with Perl 5.004_01
259 (tried with 5.003_07 and 5.002 but I get segmentation faults
260 from running the Makefile.PL with those in Tru64 4.0D)
263 - backward compatibility (pre-5.6.0) tweaks:
264 - no XSLoader in 5.00503, use DynaLoader instead
265 - no SvPV_nolen, either
266 - no PerlProc_pause(), either
267 - now tested with 5.00404 and 5.00503
268 - Makefile.PL requires 5.00404 (no more 5.002)
269 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
270 (this means that one can mix subsecond sleeps with alarms)
271 - because of nanosleep we probe for -lrt and -lposix4
272 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
273 is available by exportable constants Time::HiRes::d_func
274 (since older Perl do not have them in %Config, and even
275 5.8.0 does not probe for nanosleep)
278 - backward compatibility (pre-5.6.1) tweaks:
279 - define NV if no NVTYPE
280 - define IVdf if needed (note: the Devel::PPPort
281 in 5.8.0 does not try hard hard enough since
282 the IVSIZE might not be defined)
283 - define NVgf if needed
284 - grab the typemap from 5.8.0 for the NV stuff
286 1.31 and 1.32 add more backward compatibility (now all the way
287 back to Perl 5.00404), and using nanosleep() (if available) for
292 - release 1.29_02 as 1.30
294 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
295 development cycle. Most notably portability across platforms has been
296 enhanced, and the interval timers (setitimer, getitimer) have been
297 added. Note that the version of Time::HiRes that is included in Perl
298 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
299 version. Note also that in 1.30 Wegscheid turns over the maintenance
300 to Jarkko Hietaniemi.
304 - fix a silly unclosed comment typo in HiRes.xs
305 - document and export REALTIME_REALPROF (Solaris)
309 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
310 (need to patch this also in Perl 5.[89])
311 - remove CVS revision log from HiRes.xs
315 The following numbered patches refer to the Perl 5.7 changes,
316 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
318 - 17558: Add #!./perl to the .t
319 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
320 - 16198: political correctness, from Simon Cozens
321 - 15857: doc tweaks, from Jarkko Hietaniemi
322 - 15593: optimization in .xs, from Paul Green
323 - 14892: pod fixes, from Robin Barker
324 - 14100: VOS fixes, from Paul Green
325 - 13422: XS segfault, from Marc Lehmann
326 - 13378: whether select() gets restarted on signals, depends
327 - 13354: timing constraints, again, from Andy Dougherty
328 - 13278: can't do subecond alarms with ualarm;
329 break out early if alarms do not seem to be working
330 - 13266: test relaxation (cygwin gets lower hires
331 times than lores ones)
332 - 12846: protect against high load, from Jarkko Hietaniemi
333 - 12837: HiRes.t VMS tweak, from Craig A. Berry
334 - 12797: HiRes.t VMS tweak, from Charles Lane
335 - 12769: HiRes.t VMS tweak, from Craig A. Berry
336 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
337 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
338 - 12692: alarm() ain't gonna work if ualarm() ain't,
339 from Gurusamy Sarathy
340 - 12680: minor VMS tweak, from Charles Lane
341 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
342 - 12609: croak on negative time, from Jarkko Hietaniemi
343 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
344 - 12594: MacOS Classic timeofday, from Chris Nandor
345 - 12473: allow for more than one second for sleep() and usleep()
346 - 12458: test tuning, relax timing constraints,
347 from Jarkko Hietaniemi
348 - 12449: make sleep() and usleep() to return the number
349 of seconds and microseconds actually slept (analogously
350 with the builtin sleep()), also make usleep() croak if
351 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
352 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
353 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
354 - 12196: use ftime() on Win32, from Artur Bergman
355 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
356 - 12105: use GetSystemTime() on Win32, from Artur Bergman
357 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
358 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
359 - 11797: problem in HiRes.t, from John P. Linderman
360 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
361 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
362 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
363 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
364 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
365 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
367 - 10942: MPE/IX test tweaks, from Mark Bixby
368 - 10784: unnecessary pod2man calls, from Andy Dougherty
369 - 10354: ext/ + -Wall, from Doug MacEachern
370 - 10320: fix the BOOT section to call myU2time correctly
371 - 10317: correct casting for AIX< from H. Merijn Brand
372 - 10119: document that the core time() may be rounding, not truncating
373 - 10118: test fix, from John Peacock
374 - 9988: long =item, from Robin Barker
375 - 9714: correct test output
376 - 9708: test also the scalar aspect of getitimer()
377 - 9705: Add interval timers (setitimer, getitimer)
378 - 9692: do not require at least 5.005 using XS
380 The following changes were made on top of the changes
381 made for Time::HiRes during the Perl 5.7 development
382 cycle that culminated in the release of Perl 5.8.0.
384 - add "require 5.005" to the Makefile.PL
385 - remove the REVISION section (CVS log) from HiRes.pm
386 - add jhi's copyright alongside Douglas'
387 - move HiRes.pm to lib/Time/
389 - modify HiRes.t to use $ENV{PERL_CORE}
390 - modify the original Time::HiRes version 1.20 Makefile.PL
391 to work both with Perl 5.8.0 and the new code with pre-5.8.0
392 Perls (tried with 5.6.1)
393 - tiny tweaks and updates in README and TODO
394 - bump the VERSION to 1.29
396 1.20 Wed Feb 24 21:30 1999
397 - make our usleep and ualarm substitutes into hrt_usleep
398 and hrt_ualarm. This helps static links of Perl with other
399 packages that also have usleep, etc. From
400 Ilya Zakharevich <ilya@math.ohio-state.edu>
401 - add C API stuff. From Joshua Pritikin
402 <joshua.pritikin@db.com>
403 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
404 - hopefully correct "-lc" fix for SCO.
407 1.20 adds a platform neutral set of C accessible routines if you are
408 running 5.005+. All other changes are packaging changes and build
409 fixes(?) for statically linked Perl, SCO, and VMS.
411 1.19 Tue Sep 29 22:30 1998
412 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
413 <seb@stian.demon.co.uk>
414 - change GIMME_V to GIMME to help people with older versions of
416 - fix Win32 version of gettimeofday(). It didn't affect anything,
417 but it confuses people reading the code when the return value
418 is backwards (0 is success).
419 - fix Makefile.PL (more) so that detection of gettimeofday is
422 1.19 has better VMS support.
424 1.18 Mon Jul 6 22:40 1998
425 - add usleep() for Win32.
426 - fix Makefile.PL to fix reported HP/UX feature where unresolved
427 externals still cause an executable to be generated (though no
428 x bit set). Thanks to David Kozinn for report and explanation.
429 Problems with the fix are mine :)
431 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
432 Probably buggy. I'm sure I'll hear.
434 1.17 Wed Jul 1 20:10 1998
435 - fix setitimer calls so microseconds is not more than 1000000.
436 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
437 - make Win32. We only get gettimeofday (the select hack doesn't
438 seem to work on my Win95 system).
439 - fix test 4 on 01test.t. add test to see if time() and
440 Time::HiRes::time() are close.
442 1.16 Wed Nov 12 21:05 1997
443 - add missing EXTEND in new gettimeofday scalar code.
445 1.16+ should be closer to building out of the box on Linux. Thanks
446 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
448 If your underlying operating system doesn't implement ualarm(), then
449 a fake using setitimer() will be made. If the OS is missing usleep(),
450 a fake one using select() will be made. If a fake can't be made for
451 either ualarm() or usleep(), then the corresponding Perl function will
452 not be available. If the OS is missing gettimeofday(), you will get
453 unresolved externals, either at link- or run-time.
455 This is an improvement; the package used to not even build if
456 you were missing any of these bits. Roderick Schertler
458 <roderick@argon.org> did all the conditional compilation stuff,
459 look at HiRes.pm and the test suites; it's good educational reading.
461 1.15 Mon Nov 10 21:30 1997
462 - HiRes.pm: update pod. Provided by Gisle Aas.
463 - HiRes.xs: if gettimeofday() called in scalar context, do
464 something more useful than before. Provided by Gisle Aas.
465 - README: tell of xsubpp '-nolinenumber' woes. thanks to
466 Edward Henigin <ed@texas.net> for pointing out the problem.
468 1.14 Wed Nov 5 9:40 1997
469 - Makefile.PL: look for setitimer
470 - HiRes.xs: if missing ualarm, but we have setitimer, make up
471 our own setitimer. These were provided by Gisle Aas.
473 1.13 Tue Nov 4 23:30 1997
474 - Makefile.PL: fix autodetect mechanism to do try linking in addition
475 to just compiling; should fix Linux build problem. Fix was provided
478 1.12 Sun Oct 12 12:00:00 1997
479 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
480 you may need to comment this back out if you have an older xsubpp.
481 - HiRes.xs: set PROTOTYPES: DISABLE
483 1.11 Fri Sep 05 16:00:00 1997
485 Had some line commented out that shouldn't have been (testing
488 Previous version was corrupted.
490 1.10 Thu May 22 20:20:00 1997
491 - HiRes.xs, HiRes.pm, t/*:
492 - only compile what we have OS support for (or can
494 - only test what we compiled
495 - gross improvement to the test suite
497 This work was all done by Roderick Schertler
498 <roderick@argon.org>. If you run Linux or
499 one of the other ualarm-less platoforms, and you like this
500 module, let Roderick know; without him, it still wouldn't
501 be working on those boxes...
502 - Makefile.PL: figure out what routines the OS has and
503 only build what we need. These bits were written by Jarkko
504 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
506 1.02 Mon Dec 30 08:00:00 1996
507 - HiRes.pm: update documentation to say what to do when missing
508 ualarm() and friends.
509 - README: update to warn that ualarm() and friends need to exist
511 1.01 Fri Oct 17 08:00:00 1996
512 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
513 - HiRes.pm: put blank line between __END__ and =head1 so that
516 1.00 Tue Sep 03 13:00:00 1996
517 - original version; created by h2xs 1.16