1 Revision history for Perl extension Time::HiRes.
4 - one should not mix u?alarm and sleep (the tests modified
5 by 1.65, #12 and #13, hung in Solaris), now we just busy
6 loop executing an empty block
7 - in the documentation underline the unspecificity of mixing
12 - regenerate ppport.h with Devel::PPPort 3.03,
13 now the MY_CXT_CLONE is defined in ppport.h,
14 we no more need to do that.
16 - the test #12 would often hang in sigsuspend() (at least that's
17 where Mac OS X' ktrace shows it hanging). With the sleep()s
18 changed to sleep(1)s, the tests still pass but no hang after
19 a few hundred repeats.
22 - Win32 and any ithread build: ppport.h didn't define
23 MY_CXT_CLONE, which seems to be a Time-HiResism.
26 - Skip testing if under PERL_CORE and Time::HiRes has not
27 been Configured (from Marcus Holland-Moritz, core change
29 - Use ppport.h generated by Devel::PPPort 3.01,
30 allowing cutting away our own portability code.
31 - Don't use $ENV{PERL_CORE} for < 5.6.0.
32 - Don't use "for my $i" for <= 5.003.
33 - Don't use Pause() for <= 5.003.
34 - Can't use newSVpvf for <= 5.003.
35 (most of the changes from Marcus)
38 - Win32: reset reading from the performance counters every
39 five minutes to better track wall clock time (thanks to
40 PC timers being often quite bad), should help long-running
44 - Win32: Patch from Steve Hay
45 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
46 to [perl #30755] reported by Nigel Sandever
48 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
49 <w32api/windows.h> APIs are available. Cygwin testing by
50 Yitzchak Scott-Thoennes.
52 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
53 after that keep using -lrt, patch from Alan Burlison,
54 bug reported in [cpan #7165]
57 - Change the Win32 recalibration limit to 0.5 seconds and tweak
58 the documentation to blather less about the gory details of the
59 Win32 implementation and more about the complications in general
60 of meddling with the system clock.
63 - Document the 1.57 change better.
66 - Win32/Cygwin/MinGW: if the performance counter drifts by more
67 than two seconds from the system clock (due to ntp adjustments,
68 for example), recalibrate our internal counter: from Jan Dubois,
69 based on [cpan #5933] by Jerry D. Hedden.
72 - Give a clearer message if the tests timeout (perl change #22253)
73 - Don't use /tmp or its moral equivalents (perl bug #15036,
77 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
78 instead of VC-specific i64 suffix)
81 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
84 - Windows: higher resolution time() by using the Windows
85 performance counter API, from Jan Dubois and Anton Shcherbinin.
86 The exact new higher resolution depends on the hardware,
87 but it should be quite a bit better than using the basic
91 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
92 the subtest 18. No known analysis nor fix, but added
93 an alarm (that requires fork() and alarm()) to the test.
96 - doc tweaks from mjd (perl change #20456)
97 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
100 - add a message (for non-core builds) to Makefile.PL about
101 the LC_ALL=C workaround
102 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
103 - the nanosleep() probe was broken (perl change #20061)
104 - use existence instead of definedness for feature probes
106 - MPE/iX tweak (perl change #20042)
107 - do not use HAS_NANOSLEEP (perl change #19898)
110 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
111 - OS/2 can always mix subsecond sleeps with signals
112 (part of perl change #19789)
115 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
119 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
120 - unnecessary (nanosleep is in libc anyway)
121 - harmful (-lrt slows down execution)
122 - incompatible (with many distributions' pthreads)
125 - do not create files in blib directories under core
126 (perl change #19160, from rgs)
127 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
130 - guarantee that $xdefine in HiRes.t is always defined
131 (perl change #19109, from IlyaZ)
132 - a cleaner way to detect PERL_CORE (perl change #19111,
136 - add hints/irix.pl to turn off overly POSIX flags that
137 cause hide struct timespec to be hidden (and compilation
138 to fail) (bleadperl change #19085)
139 - documentation tweaks
142 - add c:/temp to the list of temp directories to probe
143 so that cygwin (and win*?) builds are happy. This was
144 needed at least in my cygwin 1.3.20/w2k setup.
147 - modernize the constants code (from Nicholas Clark)
150 - At some point the ability to figure our the correct incdir
151 for EXTERN.h (either a core perl build, or an installed perl)
152 had broken (which lead into all test compiles failing with
153 a core perl build, but thanks to the robustness of Makefile.PL
154 nothing of this was visible). The brokenness seemed to be
155 caused by $ENV{PERL_CORE} not being on for core builds?
156 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
157 right, and both styles of build should work again.
160 - Nicholas Clark noticed that the my_catdir() emulation function
161 was broken (which means that we didn't really work for Perls
163 - inspired by fixing the above made the whole Makefile.PL -w
165 - tightened up the Makefile.PL output, less whitespace
168 - fix from Craig Berry for better building in VMS with PERL_CORE
171 - no functional changes
172 - move lib/Time/HiRes.pm as Hires.pm
173 - libraries scanning was slightly broken (always scanned
174 for a library even when $Config{libs} already had it)
177 - Ray Zimmerman ran into a race condition in Mac OS X.
178 A 0.01-second alarm fired before the test expected.
179 The test first slept indefinitely (blocking for signals)
180 and only after that tested for the signal having been sent.
181 Since the signal had already been sent, the test #12 never
182 completed. The solution: test first, then block.
183 - default to being silent on all probing attempts, set the
184 environment variable VERBOSE to a true value to see the
185 details (the probing command and the possible errors)
188 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
189 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
190 cannot be upgraded from CPAN to override the 5.8.0 version
192 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
193 because the include directories did not adjust themselves
194 if $ENV{PERL_CORE} (Hugo van der Sanden)
195 - add documentation about the restart of select() under alarm()
198 - small documentation tweaks
202 - better VMS operation (Craig Berry)
205 - our time machine is accelerating: now works with Perl 5.004_01
206 (tried with 5.003_07 and 5.002 but I get segmentation faults
207 from running the Makefile.PL with those in Tru64 4.0D)
210 - backward compatibility (pre-5.6.0) tweaks:
211 - no XSLoader in 5.00503, use DynaLoader instead
212 - no SvPV_nolen, either
213 - no PerlProc_pause(), either
214 - now tested with 5.00404 and 5.00503
215 - Makefile.PL requires 5.00404 (no more 5.002)
216 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
217 (this means that one can mix subsecond sleeps with alarms)
218 - because of nanosleep we probe for -lrt and -lposix4
219 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
220 is available by exportable constants Time::HiRes::d_func
221 (since older Perl do not have them in %Config, and even
222 5.8.0 does not probe for nanosleep)
225 - backward compatibility (pre-5.6.1) tweaks:
226 - define NV if no NVTYPE
227 - define IVdf if needed (note: the Devel::PPPort
228 in 5.8.0 does not try hard hard enough since
229 the IVSIZE might not be defined)
230 - define NVgf if needed
231 - grab the typemap from 5.8.0 for the NV stuff
234 - release 1.29_02 as 1.30
238 - fix a silly unclosed comment typo in HiRes.xs
239 - document and export REALTIME_REALPROF (Solaris)
243 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
244 (need to patch this also in Perl 5.[89])
245 - remove CVS revision log from HiRes.xs
249 The following numbered patches refer to the Perl 5.7 changes,
250 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
252 - 17558: Add #!./perl to the .t
253 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
254 - 16198: political correctness, from Simon Cozens
255 - 15857: doc tweaks, from Jarkko Hietaniemi
256 - 15593: optimization in .xs, from Paul Green
257 - 14892: pod fixes, from Robin Barker
258 - 14100: VOS fixes, from Paul Green
259 - 13422: XS segfault, from Marc Lehmann
260 - 13378: whether select() gets restarted on signals, depends
261 - 13354: timing constraints, again, from Andy Dougherty
262 - 13278: can't do subecond alarms with ualarm;
263 break out early if alarms do not seem to be working
264 - 13266: test relaxation (cygwin gets lower hires
265 times than lores ones)
266 - 12846: protect against high load, from Jarkko Hietaniemi
267 - 12837: HiRes.t VMS tweak, from Craig A. Berry
268 - 12797: HiRes.t VMS tweak, from Charles Lane
269 - 12769: HiRes.t VMS tweak, from Craig A. Berry
270 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
271 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
272 - 12692: alarm() ain't gonna work if ualarm() ain't,
273 from Gurusamy Sarathy
274 - 12680: minor VMS tweak, from Charles Lane
275 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
276 - 12609: croak on negative time, from Jarkko Hietaniemi
277 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
278 - 12594: MacOS Classic timeofday, from Chris Nandor
279 - 12473: allow for more than one second for sleep() and usleep()
280 - 12458: test tuning, relax timing constraints,
281 from Jarkko Hietaniemi
282 - 12449: make sleep() and usleep() to return the number
283 of seconds and microseconds actually slept (analogously
284 with the builtin sleep()), also make usleep() croak if
285 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
286 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
287 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
288 - 12196: use ftime() on Win32, from Artur Bergman
289 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
290 - 12105: use GetSystemTime() on Win32, from Artur Bergman
291 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
292 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
293 - 11797: problem in HiRes.t, from John P. Linderman
294 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
295 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
296 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
297 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
298 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
299 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
301 - 10942: MPE/IX test tweaks, from Mark Bixby
302 - 10784: unnecessary pod2man calls, from Andy Dougherty
303 - 10354: ext/ + -Wall, from Doug MacEachern
304 - 10320: fix the BOOT section to call myU2time correctly
305 - 10317: correct casting for AIX< from H. Merijn Brand
306 - 10119: document that the core time() may be rounding, not truncating
307 - 10118: test fix, from John Peacock
308 - 9988: long =item, from Robin Barker
309 - 9714: correct test output
310 - 9708: test also the scalar aspect of getitimer()
311 - 9705: Add interval timers (setitimer, getitimer)
312 - 9692: do not require at least 5.005 using XS
314 The following changes were made on top of the changes
315 made for Time::HiRes during the Perl 5.7 development
316 cycle that culminated in the release of Perl 5.8.0.
318 - add "require 5.005" to the Makefile.PL
319 - remove the REVISION section (CVS log) from HiRes.pm
320 - add jhi's copyright alongside Douglas'
321 - move HiRes.pm to lib/Time/
323 - modify HiRes.t to use $ENV{PERL_CORE}
324 - modify the original Time::HiRes version 1.20 Makefile.PL
325 to work both with Perl 5.8.0 and the new code with pre-5.8.0
326 Perls (tried with 5.6.1)
327 - tiny tweaks and updates in README and TODO
328 - bump the VERSION to 1.29
330 1.20 Wed Feb 24 21:30 1999
331 - make our usleep and ualarm substitutes into hrt_usleep
332 and hrt_ualarm. This helps static links of Perl with other
333 packages that also have usleep, etc. From
334 Ilya Zakharevich <ilya@math.ohio-state.edu>
335 - add C API stuff. From Joshua Pritikin
336 <joshua.pritikin@db.com>
337 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
338 - hopefully correct "-lc" fix for SCO.
341 1.19 Tue Sep 29 22:30 1998
342 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
343 <seb@stian.demon.co.uk>
344 - change GIMME_V to GIMME to help people with older versions of
346 - fix Win32 version of gettimeofday(). It didn't affect anything,
347 but it confuses people reading the code when the return value
348 is backwards (0 is success).
349 - fix Makefile.PL (more) so that detection of gettimeofday is
352 1.18 Mon Jul 6 22:40 1998
353 - add usleep() for Win32.
354 - fix Makefile.PL to fix reported HP/UX feature where unresolved
355 externals still cause an executable to be generated (though no
356 x bit set). Thanks to David Kozinn for report and explanation.
357 Problems with the fix are mine :)
359 1.17 Wed Jul 1 20:10 1998
360 - fix setitimer calls so microseconds is not more than 1000000.
361 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
362 - make Win32. We only get gettimeofday (the select hack doesn't
363 seem to work on my Win95 system).
364 - fix test 4 on 01test.t. add test to see if time() and
365 Time::HiRes::time() are close.
367 1.16 Wed Nov 12 21:05 1997
368 - add missing EXTEND in new gettimeofday scalar code.
370 1.15 Mon Nov 10 21:30 1997
371 - HiRes.pm: update pod. Provided by Gisle Aas.
372 - HiRes.xs: if gettimeofday() called in scalar context, do
373 something more useful than before. Provided by Gisle Aas.
374 - README: tell of xsubpp '-nolinenumber' woes. thanks to
375 Edward Henigin <ed@texas.net> for pointing out the problem.
377 1.14 Wed Nov 5 9:40 1997
378 - Makefile.PL: look for setitimer
379 - HiRes.xs: if missing ualarm, but we have setitimer, make up
380 our own setitimer. These were provided by Gisle Aas.
382 1.13 Tue Nov 4 23:30 1997
383 - Makefile.PL: fix autodetect mechanism to do try linking in addition
384 to just compiling; should fix Linux build problem. Fix was provided
387 1.12 Sun Oct 12 12:00:00 1997
388 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
389 you may need to comment this back out if you have an older xsubpp.
390 - HiRes.xs: set PROTOTYPES: DISABLE
392 1.11 Fri Sep 05 16:00:00 1997
394 Had some line commented out that shouldn't have been (testing
397 Previous version was corrupted.
399 1.10 Thu May 22 20:20:00 1997
400 - HiRes.xs, HiRes.pm, t/*:
401 - only compile what we have OS support for (or can
403 - only test what we compiled
404 - gross improvement to the test suite
406 This work was all done by Roderick Schertler
407 <roderick@argon.org>. If you run Linux or
408 one of the other ualarm-less platoforms, and you like this
409 module, let Roderick know; without him, it still wouldn't
410 be working on those boxes...
411 - Makefile.PL: figure out what routines the OS has and
412 only build what we need. These bits were written by Jarkko
413 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
415 1.02 Mon Dec 30 08:00:00 1996
416 - HiRes.pm: update documentation to say what to do when missing
417 ualarm() and friends.
418 - README: update to warn that ualarm() and friends need to exist
420 1.01 Fri Oct 17 08:00:00 1996
421 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
422 - HiRes.pm: put blank line between __END__ and =head1 so that
425 1.00 Tue Sep 03 13:00:00 1996
426 - original version; created by h2xs 1.16