1 Revision history for Perl extension Time::HiRes.
3 1.69 - actually run a test for nanosleep since e.g. in AIX 4.2
4 it seems that one can link in nanosleep() but then calling
5 it fails instantly and sets errno to ENOSYS (Not implemented).
6 This may be fixable in the AIX case by figuring out the right
7 (realtime POSIX?) libs and whatnot, but in the general case
8 running a real test case is better. (Of course, this change
9 will no doubt run into portability problems because of the
11 (from José Auguste-tienne)
12 - support XSLoader also since it's much faster
14 - add SEE ALSO (BSD::Resource and Time::TAI64)
17 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
18 error suspected), but miraculously it still worked since the
19 doubling took place below the __END__ token
20 - undef Pause() before defining it to avoid redefinition warnings
21 during compilation in case perl.h had already defined Pause()
22 (part of perl change #24271)
26 - (internal) don't ignore the return value of gettimeofday()
27 - (external) return undef or an empty if the C gettimeofday() fails
28 (affects Time::HiRes gettimeofday() and the hires time())
32 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
33 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
34 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
35 - 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
36 has become rather unlikely
39 - one should not mix u?alarm and sleep (the tests modified
40 by 1.65, #12 and #13, hung in Solaris), now we just busy
41 loop executing an empty block
42 - in the documentation underline the unspecificity of mixing
44 - small spelling fixes
47 - regenerate ppport.h with Devel::PPPort 3.03,
48 now the MY_CXT_CLONE is defined in ppport.h,
49 we no more need to do that.
51 - the test #12 would often hang in sigsuspend() (at least that's
52 where Mac OS X' ktrace shows it hanging). With the sleep()s
53 changed to sleep(1)s, the tests still pass but no hang after
54 a few hundred repeats.
57 - Win32 and any ithread build: ppport.h didn't define
58 MY_CXT_CLONE, which seems to be a Time-HiResism.
61 - Skip testing if under PERL_CORE and Time::HiRes has not
62 been Configured (from Marcus Holland-Moritz, core change
64 - Use ppport.h generated by Devel::PPPort 3.01,
65 allowing cutting away our own portability code.
66 - Don't use $ENV{PERL_CORE} for < 5.6.0.
67 - Don't use "for my $i" for <= 5.003.
68 - Don't use Pause() for <= 5.003.
69 - Can't use newSVpvf for <= 5.003.
70 (most of the changes from Marcus)
73 - Win32: reset reading from the performance counters every
74 five minutes to better track wall clock time (thanks to
75 PC timers being often quite bad), should help long-running
79 - Win32: Patch from Steve Hay
80 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
81 to [perl #30755] reported by Nigel Sandever
83 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
84 <w32api/windows.h> APIs are available. Cygwin testing by
85 Yitzchak Scott-Thoennes.
87 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
88 after that keep using -lrt, patch from Alan Burlison,
89 bug reported in [cpan #7165]
92 - Change the Win32 recalibration limit to 0.5 seconds and tweak
93 the documentation to blather less about the gory details of the
94 Win32 implementation and more about the complications in general
95 of meddling with the system clock.
98 - Document the 1.57 change better.
101 - Win32/Cygwin/MinGW: if the performance counter drifts by more
102 than two seconds from the system clock (due to ntp adjustments,
103 for example), recalibrate our internal counter: from Jan Dubois,
104 based on [cpan #5933] by Jerry D. Hedden.
107 - Give a clearer message if the tests timeout (perl change #22253)
108 - Don't use /tmp or its moral equivalents (perl bug #15036,
112 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
113 instead of VC-specific i64 suffix)
116 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
119 - Windows: higher resolution time() by using the Windows
120 performance counter API, from Jan Dubois and Anton Shcherbinin.
121 The exact new higher resolution depends on the hardware,
122 but it should be quite a bit better than using the basic
126 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
127 the subtest 18. No known analysis nor fix, but added
128 an alarm (that requires fork() and alarm()) to the test.
131 - doc tweaks from mjd (perl change #20456)
132 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
135 - add a message (for non-core builds) to Makefile.PL about
136 the LC_ALL=C workaround
137 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
138 - the nanosleep() probe was broken (perl change #20061)
139 - use existence instead of definedness for feature probes
141 - MPE/iX tweak (perl change #20042)
142 - do not use HAS_NANOSLEEP (perl change #19898)
145 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
146 - OS/2 can always mix subsecond sleeps with signals
147 (part of perl change #19789)
150 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
154 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
155 - unnecessary (nanosleep is in libc anyway)
156 - harmful (-lrt slows down execution)
157 - incompatible (with many distributions' pthreads)
160 - do not create files in blib directories under core
161 (perl change #19160, from rgs)
162 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
165 - guarantee that $xdefine in HiRes.t is always defined
166 (perl change #19109, from IlyaZ)
167 - a cleaner way to detect PERL_CORE (perl change #19111,
171 - add hints/irix.pl to turn off overly POSIX flags that
172 cause hide struct timespec to be hidden (and compilation
173 to fail) (bleadperl change #19085)
174 - documentation tweaks
177 - add c:/temp to the list of temp directories to probe
178 so that cygwin (and win*?) builds are happy. This was
179 needed at least in my cygwin 1.3.20/w2k setup.
182 - modernize the constants code (from Nicholas Clark)
185 - At some point the ability to figure our the correct incdir
186 for EXTERN.h (either a core perl build, or an installed perl)
187 had broken (which lead into all test compiles failing with
188 a core perl build, but thanks to the robustness of Makefile.PL
189 nothing of this was visible). The brokenness seemed to be
190 caused by $ENV{PERL_CORE} not being on for core builds?
191 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
192 right, and both styles of build should work again.
195 - Nicholas Clark noticed that the my_catdir() emulation function
196 was broken (which means that we didn't really work for Perls
198 - inspired by fixing the above made the whole Makefile.PL -w
200 - tightened up the Makefile.PL output, less whitespace
203 - fix from Craig Berry for better building in VMS with PERL_CORE
206 - no functional changes
207 - move lib/Time/HiRes.pm as Hires.pm
208 - libraries scanning was slightly broken (always scanned
209 for a library even when $Config{libs} already had it)
212 - Ray Zimmerman ran into a race condition in Mac OS X.
213 A 0.01-second alarm fired before the test expected.
214 The test first slept indefinitely (blocking for signals)
215 and only after that tested for the signal having been sent.
216 Since the signal had already been sent, the test #12 never
217 completed. The solution: test first, then block.
218 - default to being silent on all probing attempts, set the
219 environment variable VERBOSE to a true value to see the
220 details (the probing command and the possible errors)
223 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
224 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
225 cannot be upgraded from CPAN to override the 5.8.0 version
227 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
228 because the include directories did not adjust themselves
229 if $ENV{PERL_CORE} (Hugo van der Sanden)
230 - add documentation about the restart of select() under alarm()
233 - small documentation tweaks
237 - better VMS operation (Craig Berry)
240 - our time machine is accelerating: now works with Perl 5.004_01
241 (tried with 5.003_07 and 5.002 but I get segmentation faults
242 from running the Makefile.PL with those in Tru64 4.0D)
245 - backward compatibility (pre-5.6.0) tweaks:
246 - no XSLoader in 5.00503, use DynaLoader instead
247 - no SvPV_nolen, either
248 - no PerlProc_pause(), either
249 - now tested with 5.00404 and 5.00503
250 - Makefile.PL requires 5.00404 (no more 5.002)
251 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
252 (this means that one can mix subsecond sleeps with alarms)
253 - because of nanosleep we probe for -lrt and -lposix4
254 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
255 is available by exportable constants Time::HiRes::d_func
256 (since older Perl do not have them in %Config, and even
257 5.8.0 does not probe for nanosleep)
260 - backward compatibility (pre-5.6.1) tweaks:
261 - define NV if no NVTYPE
262 - define IVdf if needed (note: the Devel::PPPort
263 in 5.8.0 does not try hard hard enough since
264 the IVSIZE might not be defined)
265 - define NVgf if needed
266 - grab the typemap from 5.8.0 for the NV stuff
269 - release 1.29_02 as 1.30
273 - fix a silly unclosed comment typo in HiRes.xs
274 - document and export REALTIME_REALPROF (Solaris)
278 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
279 (need to patch this also in Perl 5.[89])
280 - remove CVS revision log from HiRes.xs
284 The following numbered patches refer to the Perl 5.7 changes,
285 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
287 - 17558: Add #!./perl to the .t
288 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
289 - 16198: political correctness, from Simon Cozens
290 - 15857: doc tweaks, from Jarkko Hietaniemi
291 - 15593: optimization in .xs, from Paul Green
292 - 14892: pod fixes, from Robin Barker
293 - 14100: VOS fixes, from Paul Green
294 - 13422: XS segfault, from Marc Lehmann
295 - 13378: whether select() gets restarted on signals, depends
296 - 13354: timing constraints, again, from Andy Dougherty
297 - 13278: can't do subecond alarms with ualarm;
298 break out early if alarms do not seem to be working
299 - 13266: test relaxation (cygwin gets lower hires
300 times than lores ones)
301 - 12846: protect against high load, from Jarkko Hietaniemi
302 - 12837: HiRes.t VMS tweak, from Craig A. Berry
303 - 12797: HiRes.t VMS tweak, from Charles Lane
304 - 12769: HiRes.t VMS tweak, from Craig A. Berry
305 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
306 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
307 - 12692: alarm() ain't gonna work if ualarm() ain't,
308 from Gurusamy Sarathy
309 - 12680: minor VMS tweak, from Charles Lane
310 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
311 - 12609: croak on negative time, from Jarkko Hietaniemi
312 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
313 - 12594: MacOS Classic timeofday, from Chris Nandor
314 - 12473: allow for more than one second for sleep() and usleep()
315 - 12458: test tuning, relax timing constraints,
316 from Jarkko Hietaniemi
317 - 12449: make sleep() and usleep() to return the number
318 of seconds and microseconds actually slept (analogously
319 with the builtin sleep()), also make usleep() croak if
320 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
321 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
322 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
323 - 12196: use ftime() on Win32, from Artur Bergman
324 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
325 - 12105: use GetSystemTime() on Win32, from Artur Bergman
326 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
327 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
328 - 11797: problem in HiRes.t, from John P. Linderman
329 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
330 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
331 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
332 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
333 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
334 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
336 - 10942: MPE/IX test tweaks, from Mark Bixby
337 - 10784: unnecessary pod2man calls, from Andy Dougherty
338 - 10354: ext/ + -Wall, from Doug MacEachern
339 - 10320: fix the BOOT section to call myU2time correctly
340 - 10317: correct casting for AIX< from H. Merijn Brand
341 - 10119: document that the core time() may be rounding, not truncating
342 - 10118: test fix, from John Peacock
343 - 9988: long =item, from Robin Barker
344 - 9714: correct test output
345 - 9708: test also the scalar aspect of getitimer()
346 - 9705: Add interval timers (setitimer, getitimer)
347 - 9692: do not require at least 5.005 using XS
349 The following changes were made on top of the changes
350 made for Time::HiRes during the Perl 5.7 development
351 cycle that culminated in the release of Perl 5.8.0.
353 - add "require 5.005" to the Makefile.PL
354 - remove the REVISION section (CVS log) from HiRes.pm
355 - add jhi's copyright alongside Douglas'
356 - move HiRes.pm to lib/Time/
358 - modify HiRes.t to use $ENV{PERL_CORE}
359 - modify the original Time::HiRes version 1.20 Makefile.PL
360 to work both with Perl 5.8.0 and the new code with pre-5.8.0
361 Perls (tried with 5.6.1)
362 - tiny tweaks and updates in README and TODO
363 - bump the VERSION to 1.29
365 1.20 Wed Feb 24 21:30 1999
366 - make our usleep and ualarm substitutes into hrt_usleep
367 and hrt_ualarm. This helps static links of Perl with other
368 packages that also have usleep, etc. From
369 Ilya Zakharevich <ilya@math.ohio-state.edu>
370 - add C API stuff. From Joshua Pritikin
371 <joshua.pritikin@db.com>
372 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
373 - hopefully correct "-lc" fix for SCO.
376 1.19 Tue Sep 29 22:30 1998
377 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
378 <seb@stian.demon.co.uk>
379 - change GIMME_V to GIMME to help people with older versions of
381 - fix Win32 version of gettimeofday(). It didn't affect anything,
382 but it confuses people reading the code when the return value
383 is backwards (0 is success).
384 - fix Makefile.PL (more) so that detection of gettimeofday is
387 1.18 Mon Jul 6 22:40 1998
388 - add usleep() for Win32.
389 - fix Makefile.PL to fix reported HP/UX feature where unresolved
390 externals still cause an executable to be generated (though no
391 x bit set). Thanks to David Kozinn for report and explanation.
392 Problems with the fix are mine :)
394 1.17 Wed Jul 1 20:10 1998
395 - fix setitimer calls so microseconds is not more than 1000000.
396 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
397 - make Win32. We only get gettimeofday (the select hack doesn't
398 seem to work on my Win95 system).
399 - fix test 4 on 01test.t. add test to see if time() and
400 Time::HiRes::time() are close.
402 1.16 Wed Nov 12 21:05 1997
403 - add missing EXTEND in new gettimeofday scalar code.
405 1.15 Mon Nov 10 21:30 1997
406 - HiRes.pm: update pod. Provided by Gisle Aas.
407 - HiRes.xs: if gettimeofday() called in scalar context, do
408 something more useful than before. Provided by Gisle Aas.
409 - README: tell of xsubpp '-nolinenumber' woes. thanks to
410 Edward Henigin <ed@texas.net> for pointing out the problem.
412 1.14 Wed Nov 5 9:40 1997
413 - Makefile.PL: look for setitimer
414 - HiRes.xs: if missing ualarm, but we have setitimer, make up
415 our own setitimer. These were provided by Gisle Aas.
417 1.13 Tue Nov 4 23:30 1997
418 - Makefile.PL: fix autodetect mechanism to do try linking in addition
419 to just compiling; should fix Linux build problem. Fix was provided
422 1.12 Sun Oct 12 12:00:00 1997
423 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
424 you may need to comment this back out if you have an older xsubpp.
425 - HiRes.xs: set PROTOTYPES: DISABLE
427 1.11 Fri Sep 05 16:00:00 1997
429 Had some line commented out that shouldn't have been (testing
432 Previous version was corrupted.
434 1.10 Thu May 22 20:20:00 1997
435 - HiRes.xs, HiRes.pm, t/*:
436 - only compile what we have OS support for (or can
438 - only test what we compiled
439 - gross improvement to the test suite
441 This work was all done by Roderick Schertler
442 <roderick@argon.org>. If you run Linux or
443 one of the other ualarm-less platoforms, and you like this
444 module, let Roderick know; without him, it still wouldn't
445 be working on those boxes...
446 - Makefile.PL: figure out what routines the OS has and
447 only build what we need. These bits were written by Jarkko
448 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
450 1.02 Mon Dec 30 08:00:00 1996
451 - HiRes.pm: update documentation to say what to do when missing
452 ualarm() and friends.
453 - README: update to warn that ualarm() and friends need to exist
455 1.01 Fri Oct 17 08:00:00 1996
456 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
457 - HiRes.pm: put blank line between __END__ and =head1 so that
460 1.00 Tue Sep 03 13:00:00 1996
461 - original version; created by h2xs 1.16