1 Revision history for Perl extension Time::HiRes.
3 1.71 - a thinko in the nanosleep() detection
4 - move more changes stuff from the README to Changes
5 - add -w to the Makefile.PL
7 1.70 - oops in 1.69 about @ISA (not affecting anything but silly)
8 - add copyright 2005 to HiRes.pm
9 - add copyright and license to HiRes.xs
10 - add copyrights 2003, 2004, 2005 to README
12 1.69 - actually run a test for nanosleep
13 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
14 it seems that one can link in nanosleep() but then calling
15 it fails instantly and sets errno to ENOSYS (Not implemented).
16 This may be fixable in the AIX case by figuring out the right
17 (realtime POSIX?) libs and whatnot, but in the general case
18 running a real test case is better. (Of course, this change
19 will no doubt run into portability problems because of the
21 (from José Auguste-tienne)
22 - support XSLoader also since it's much faster
24 - add SEE ALSO (BSD::Resource and Time::TAI64)
27 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
28 error suspected), but miraculously it still worked since the
29 doubling took place below the __END__ token
30 - undef Pause() before defining it to avoid redefinition warnings
31 during compilation in case perl.h had already defined Pause()
32 (part of perl change #24271)
36 - (internal) don't ignore the return value of gettimeofday()
37 - (external) return undef or an empty if the C gettimeofday() fails
38 (affects Time::HiRes gettimeofday() and the hires time())
42 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
43 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
44 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
45 - 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
46 has become rather unlikely
49 - one should not mix u?alarm and sleep (the tests modified
50 by 1.65, #12 and #13, hung in Solaris), now we just busy
51 loop executing an empty block
52 - in the documentation underline the unspecificity of mixing
54 - small spelling fixes
57 - regenerate ppport.h with Devel::PPPort 3.03,
58 now the MY_CXT_CLONE is defined in ppport.h,
59 we no more need to do that.
61 - the test #12 would often hang in sigsuspend() (at least that's
62 where Mac OS X' ktrace shows it hanging). With the sleep()s
63 changed to sleep(1)s, the tests still pass but no hang after
64 a few hundred repeats.
67 - Win32 and any ithread build: ppport.h didn't define
68 MY_CXT_CLONE, which seems to be a Time-HiResism.
71 - Skip testing if under PERL_CORE and Time::HiRes has not
72 been Configured (from Marcus Holland-Moritz, core change
74 - Use ppport.h generated by Devel::PPPort 3.01,
75 allowing cutting away our own portability code.
76 - Don't use $ENV{PERL_CORE} for < 5.6.0.
77 - Don't use "for my $i" for <= 5.003.
78 - Don't use Pause() for <= 5.003.
79 - Can't use newSVpvf for <= 5.003.
80 (most of the changes from Marcus)
83 - Win32: reset reading from the performance counters every
84 five minutes to better track wall clock time (thanks to
85 PC timers being often quite bad), should help long-running
89 - Win32: Patch from Steve Hay
90 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
91 to [perl #30755] reported by Nigel Sandever
93 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
94 <w32api/windows.h> APIs are available. Cygwin testing by
95 Yitzchak Scott-Thoennes.
97 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
98 after that keep using -lrt, patch from Alan Burlison,
99 bug reported in [cpan #7165]
102 - Change the Win32 recalibration limit to 0.5 seconds and tweak
103 the documentation to blather less about the gory details of the
104 Win32 implementation and more about the complications in general
105 of meddling with the system clock.
108 - Document the 1.57 change better.
111 - Win32/Cygwin/MinGW: if the performance counter drifts by more
112 than two seconds from the system clock (due to ntp adjustments,
113 for example), recalibrate our internal counter: from Jan Dubois,
114 based on [cpan #5933] by Jerry D. Hedden.
117 - Give a clearer message if the tests timeout (perl change #22253)
118 - Don't use /tmp or its moral equivalents (perl bug #15036,
122 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
123 instead of VC-specific i64 suffix)
126 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
129 - Windows: higher resolution time() by using the Windows
130 performance counter API, from Jan Dubois and Anton Shcherbinin.
131 The exact new higher resolution depends on the hardware,
132 but it should be quite a bit better than using the basic
136 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
137 the subtest 18. No known analysis nor fix, but added
138 an alarm (that requires fork() and alarm()) to the test.
141 - doc tweaks from mjd (perl change #20456)
142 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
145 - add a message (for non-core builds) to Makefile.PL about
146 the LC_ALL=C workaround
147 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
148 - the nanosleep() probe was broken (perl change #20061)
149 - use existence instead of definedness for feature probes
151 - MPE/iX tweak (perl change #20042)
152 - do not use HAS_NANOSLEEP (perl change #19898)
155 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
156 - OS/2 can always mix subsecond sleeps with signals
157 (part of perl change #19789)
160 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
164 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
165 - unnecessary (nanosleep is in libc anyway)
166 - harmful (-lrt slows down execution)
167 - incompatible (with many distributions' pthreads)
170 - do not create files in blib directories under core
171 (perl change #19160, from rgs)
172 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
175 - guarantee that $xdefine in HiRes.t is always defined
176 (perl change #19109, from IlyaZ)
177 - a cleaner way to detect PERL_CORE (perl change #19111,
181 - add hints/irix.pl to turn off overly POSIX flags that
182 cause hide struct timespec to be hidden (and compilation
183 to fail) (bleadperl change #19085)
184 - documentation tweaks
187 - add c:/temp to the list of temp directories to probe
188 so that cygwin (and win*?) builds are happy. This was
189 needed at least in my cygwin 1.3.20/w2k setup.
192 - modernize the constants code (from Nicholas Clark)
195 - At some point the ability to figure our the correct incdir
196 for EXTERN.h (either a core perl build, or an installed perl)
197 had broken (which lead into all test compiles failing with
198 a core perl build, but thanks to the robustness of Makefile.PL
199 nothing of this was visible). The brokenness seemed to be
200 caused by $ENV{PERL_CORE} not being on for core builds?
201 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
202 right, and both styles of build should work again.
205 - Nicholas Clark noticed that the my_catdir() emulation function
206 was broken (which means that we didn't really work for Perls
208 - inspired by fixing the above made the whole Makefile.PL -w
210 - tightened up the Makefile.PL output, less whitespace
213 - fix from Craig Berry for better building in VMS with PERL_CORE
216 - no functional changes
217 - move lib/Time/HiRes.pm as Hires.pm
218 - libraries scanning was slightly broken (always scanned
219 for a library even when $Config{libs} already had it)
222 - Ray Zimmerman ran into a race condition in Mac OS X.
223 A 0.01-second alarm fired before the test expected.
224 The test first slept indefinitely (blocking for signals)
225 and only after that tested for the signal having been sent.
226 Since the signal had already been sent, the test #12 never
227 completed. The solution: test first, then block.
228 - default to being silent on all probing attempts, set the
229 environment variable VERBOSE to a true value to see the
230 details (the probing command and the possible errors)
233 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
234 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
235 cannot be upgraded from CPAN to override the 5.8.0 version
237 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
238 because the include directories did not adjust themselves
239 if $ENV{PERL_CORE} (Hugo van der Sanden)
240 - add documentation about the restart of select() under alarm()
243 - small documentation tweaks
247 - better VMS operation (Craig Berry)
250 - our time machine is accelerating: now works with Perl 5.004_01
251 (tried with 5.003_07 and 5.002 but I get segmentation faults
252 from running the Makefile.PL with those in Tru64 4.0D)
255 - backward compatibility (pre-5.6.0) tweaks:
256 - no XSLoader in 5.00503, use DynaLoader instead
257 - no SvPV_nolen, either
258 - no PerlProc_pause(), either
259 - now tested with 5.00404 and 5.00503
260 - Makefile.PL requires 5.00404 (no more 5.002)
261 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
262 (this means that one can mix subsecond sleeps with alarms)
263 - because of nanosleep we probe for -lrt and -lposix4
264 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
265 is available by exportable constants Time::HiRes::d_func
266 (since older Perl do not have them in %Config, and even
267 5.8.0 does not probe for nanosleep)
270 - backward compatibility (pre-5.6.1) tweaks:
271 - define NV if no NVTYPE
272 - define IVdf if needed (note: the Devel::PPPort
273 in 5.8.0 does not try hard hard enough since
274 the IVSIZE might not be defined)
275 - define NVgf if needed
276 - grab the typemap from 5.8.0 for the NV stuff
278 1.31 and 1.32 add more backward compatibility (now all the way
279 back to Perl 5.00404), and using nanosleep() (if available) for
284 - release 1.29_02 as 1.30
286 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
287 development cycle. Most notably portability across platforms has been
288 enhanced, and the interval timers (setitimer, getitimer) have been
289 added. Note that the version of Time::HiRes that is included in Perl
290 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
291 version. Note also that in 1.30 Wegscheid turns over the maintenance
292 to Jarkko Hietaniemi.
296 - fix a silly unclosed comment typo in HiRes.xs
297 - document and export REALTIME_REALPROF (Solaris)
301 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
302 (need to patch this also in Perl 5.[89])
303 - remove CVS revision log from HiRes.xs
307 The following numbered patches refer to the Perl 5.7 changes,
308 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
310 - 17558: Add #!./perl to the .t
311 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
312 - 16198: political correctness, from Simon Cozens
313 - 15857: doc tweaks, from Jarkko Hietaniemi
314 - 15593: optimization in .xs, from Paul Green
315 - 14892: pod fixes, from Robin Barker
316 - 14100: VOS fixes, from Paul Green
317 - 13422: XS segfault, from Marc Lehmann
318 - 13378: whether select() gets restarted on signals, depends
319 - 13354: timing constraints, again, from Andy Dougherty
320 - 13278: can't do subecond alarms with ualarm;
321 break out early if alarms do not seem to be working
322 - 13266: test relaxation (cygwin gets lower hires
323 times than lores ones)
324 - 12846: protect against high load, from Jarkko Hietaniemi
325 - 12837: HiRes.t VMS tweak, from Craig A. Berry
326 - 12797: HiRes.t VMS tweak, from Charles Lane
327 - 12769: HiRes.t VMS tweak, from Craig A. Berry
328 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
329 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
330 - 12692: alarm() ain't gonna work if ualarm() ain't,
331 from Gurusamy Sarathy
332 - 12680: minor VMS tweak, from Charles Lane
333 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
334 - 12609: croak on negative time, from Jarkko Hietaniemi
335 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
336 - 12594: MacOS Classic timeofday, from Chris Nandor
337 - 12473: allow for more than one second for sleep() and usleep()
338 - 12458: test tuning, relax timing constraints,
339 from Jarkko Hietaniemi
340 - 12449: make sleep() and usleep() to return the number
341 of seconds and microseconds actually slept (analogously
342 with the builtin sleep()), also make usleep() croak if
343 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
344 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
345 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
346 - 12196: use ftime() on Win32, from Artur Bergman
347 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
348 - 12105: use GetSystemTime() on Win32, from Artur Bergman
349 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
350 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
351 - 11797: problem in HiRes.t, from John P. Linderman
352 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
353 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
354 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
355 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
356 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
357 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
359 - 10942: MPE/IX test tweaks, from Mark Bixby
360 - 10784: unnecessary pod2man calls, from Andy Dougherty
361 - 10354: ext/ + -Wall, from Doug MacEachern
362 - 10320: fix the BOOT section to call myU2time correctly
363 - 10317: correct casting for AIX< from H. Merijn Brand
364 - 10119: document that the core time() may be rounding, not truncating
365 - 10118: test fix, from John Peacock
366 - 9988: long =item, from Robin Barker
367 - 9714: correct test output
368 - 9708: test also the scalar aspect of getitimer()
369 - 9705: Add interval timers (setitimer, getitimer)
370 - 9692: do not require at least 5.005 using XS
372 The following changes were made on top of the changes
373 made for Time::HiRes during the Perl 5.7 development
374 cycle that culminated in the release of Perl 5.8.0.
376 - add "require 5.005" to the Makefile.PL
377 - remove the REVISION section (CVS log) from HiRes.pm
378 - add jhi's copyright alongside Douglas'
379 - move HiRes.pm to lib/Time/
381 - modify HiRes.t to use $ENV{PERL_CORE}
382 - modify the original Time::HiRes version 1.20 Makefile.PL
383 to work both with Perl 5.8.0 and the new code with pre-5.8.0
384 Perls (tried with 5.6.1)
385 - tiny tweaks and updates in README and TODO
386 - bump the VERSION to 1.29
388 1.20 Wed Feb 24 21:30 1999
389 - make our usleep and ualarm substitutes into hrt_usleep
390 and hrt_ualarm. This helps static links of Perl with other
391 packages that also have usleep, etc. From
392 Ilya Zakharevich <ilya@math.ohio-state.edu>
393 - add C API stuff. From Joshua Pritikin
394 <joshua.pritikin@db.com>
395 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
396 - hopefully correct "-lc" fix for SCO.
399 1.20 adds a platform neutral set of C accessible routines if you are
400 running 5.005+. All other changes are packaging changes and build
401 fixes(?) for statically linked Perl, SCO, and VMS.
403 1.19 Tue Sep 29 22:30 1998
404 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
405 <seb@stian.demon.co.uk>
406 - change GIMME_V to GIMME to help people with older versions of
408 - fix Win32 version of gettimeofday(). It didn't affect anything,
409 but it confuses people reading the code when the return value
410 is backwards (0 is success).
411 - fix Makefile.PL (more) so that detection of gettimeofday is
414 1.19 has better VMS support.
416 1.18 Mon Jul 6 22:40 1998
417 - add usleep() for Win32.
418 - fix Makefile.PL to fix reported HP/UX feature where unresolved
419 externals still cause an executable to be generated (though no
420 x bit set). Thanks to David Kozinn for report and explanation.
421 Problems with the fix are mine :)
423 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
424 Probably buggy. I'm sure I'll hear.
426 1.17 Wed Jul 1 20:10 1998
427 - fix setitimer calls so microseconds is not more than 1000000.
428 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
429 - make Win32. We only get gettimeofday (the select hack doesn't
430 seem to work on my Win95 system).
431 - fix test 4 on 01test.t. add test to see if time() and
432 Time::HiRes::time() are close.
434 1.16 Wed Nov 12 21:05 1997
435 - add missing EXTEND in new gettimeofday scalar code.
437 1.16+ should be closer to building out of the box on Linux. Thanks
438 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
440 If your underlying operating system doesn't implement ualarm(), then
441 a fake using setitimer() will be made. If the OS is missing usleep(),
442 a fake one using select() will be made. If a fake can't be made for
443 either ualarm() or usleep(), then the corresponding Perl function will
444 not be available. If the OS is missing gettimeofday(), you will get
445 unresolved externals, either at link- or run-time.
447 This is an improvement; the package used to not even build if
448 you were missing any of these bits. Roderick Schertler
450 <roderick@argon.org> did all the conditional compilation stuff,
451 look at HiRes.pm and the test suites; it's good educational reading.
453 1.15 Mon Nov 10 21:30 1997
454 - HiRes.pm: update pod. Provided by Gisle Aas.
455 - HiRes.xs: if gettimeofday() called in scalar context, do
456 something more useful than before. Provided by Gisle Aas.
457 - README: tell of xsubpp '-nolinenumber' woes. thanks to
458 Edward Henigin <ed@texas.net> for pointing out the problem.
460 1.14 Wed Nov 5 9:40 1997
461 - Makefile.PL: look for setitimer
462 - HiRes.xs: if missing ualarm, but we have setitimer, make up
463 our own setitimer. These were provided by Gisle Aas.
465 1.13 Tue Nov 4 23:30 1997
466 - Makefile.PL: fix autodetect mechanism to do try linking in addition
467 to just compiling; should fix Linux build problem. Fix was provided
470 1.12 Sun Oct 12 12:00:00 1997
471 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
472 you may need to comment this back out if you have an older xsubpp.
473 - HiRes.xs: set PROTOTYPES: DISABLE
475 1.11 Fri Sep 05 16:00:00 1997
477 Had some line commented out that shouldn't have been (testing
480 Previous version was corrupted.
482 1.10 Thu May 22 20:20:00 1997
483 - HiRes.xs, HiRes.pm, t/*:
484 - only compile what we have OS support for (or can
486 - only test what we compiled
487 - gross improvement to the test suite
489 This work was all done by Roderick Schertler
490 <roderick@argon.org>. If you run Linux or
491 one of the other ualarm-less platoforms, and you like this
492 module, let Roderick know; without him, it still wouldn't
493 be working on those boxes...
494 - Makefile.PL: figure out what routines the OS has and
495 only build what we need. These bits were written by Jarkko
496 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
498 1.02 Mon Dec 30 08:00:00 1996
499 - HiRes.pm: update documentation to say what to do when missing
500 ualarm() and friends.
501 - README: update to warn that ualarm() and friends need to exist
503 1.01 Fri Oct 17 08:00:00 1996
504 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
505 - HiRes.pm: put blank line between __END__ and =head1 so that
508 1.00 Tue Sep 03 13:00:00 1996
509 - original version; created by h2xs 1.16