1 Revision history for Perl extension Time::HiRes.
4 - Win32: reset reading from the performance counters every
5 five minutes to better track wall clock time (thanks to
6 PC timers being often quite bad), should help long-running
10 - Win32: Patch from Steve Hay
11 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
12 to [perl #30755] reported by Nigel Sandever
14 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
15 <w32api/windows.h> APIs are available. Cygwin testing by
16 Yitzchak Scott-Thoennes.
18 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
19 after that keep using -lrt, patch from Alan Burlison,
20 bug reported in [cpan #7165]
23 - Change the Win32 recalibration limit to 0.5 seconds and tweak
24 the documentation to blather less about the gory details of the
25 Win32 implementation and more about the complications in general
26 of meddling with the system clock.
29 - Document the 1.57 change better.
32 - Win32/Cygwin/MinGW: if the performance counter drifts by more
33 than two seconds from the system clock (due to ntp adjustments,
34 for example), recalibrate our internal counter: from Jan Dubois,
35 based on [cpan #5933] by Jerry D. Hedden.
38 - Give a clearer message if the tests timeout (perl change #22253)
39 - Don't use /tmp or its moral equivalents (perl bug #15036,
43 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
44 instead of VC-specific i64 suffix)
47 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
50 - Windows: higher resolution time() by using the Windows
51 performance counter API, from Jan Dubois and Anton Shcherbinin.
52 The exact new higher resolution depends on the hardware,
53 but it should be quite a bit better than using the basic
57 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
58 the subtest 18. No known analysis nor fix, but added
59 an alarm (that requires fork() and alarm()) to the test.
62 - doc tweaks from mjd (perl change #20456)
63 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
66 - add a message (for non-core builds) to Makefile.PL about
67 the LC_ALL=C workaround
68 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
69 - the nanosleep() probe was broken (perl change #20061)
70 - use existence instead of definedness for feature probes
72 - MPE/iX tweak (perl change #20042)
73 - do not use HAS_NANOSLEEP (perl change #19898)
76 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
77 - OS/2 can always mix subsecond sleeps with signals
78 (part of perl change #19789)
81 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
85 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
86 - unnecessary (nanosleep is in libc anyway)
87 - harmful (-lrt slows down execution)
88 - incompatible (with many distributions' pthreads)
91 - do not create files in blib directories under core
92 (perl change #19160, from rgs)
93 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
96 - guarantee that $xdefine in HiRes.t is always defined
97 (perl change #19109, from IlyaZ)
98 - a cleaner way to detect PERL_CORE (perl change #19111,
102 - add hints/irix.pl to turn off overly POSIX flags that
103 cause hide struct timespec to be hidden (and compilation
104 to fail) (bleadperl change #19085)
105 - documentation tweaks
108 - add c:/temp to the list of temp directories to probe
109 so that cygwin (and win*?) builds are happy. This was
110 needed at least in my cygwin 1.3.20/w2k setup.
113 - modernize the constants code (from Nicholas Clark)
116 - At some point the ability to figure our the correct incdir
117 for EXTERN.h (either a core perl build, or an installed perl)
118 had broken (which lead into all test compiles failing with
119 a core perl build, but thanks to the robustness of Makefile.PL
120 nothing of this was visible). The brokenness seemed to be
121 caused by $ENV{PERL_CORE} not being on for core builds?
122 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
123 right, and both styles of build should work again.
126 - Nicholas Clark noticed that the my_catdir() emulation function
127 was broken (which means that we didn't really work for Perls
129 - inspired by fixing the above made the whole Makefile.PL -w
131 - tightened up the Makefile.PL output, less whitespace
134 - fix from Craig Berry for better building in VMS with PERL_CORE
137 - no functional changes
138 - move lib/Time/HiRes.pm as Hires.pm
139 - libraries scanning was slightly broken (always scanned
140 for a library even when $Config{libs} already had it)
143 - Ray Zimmerman ran into a race condition in Mac OS X.
144 A 0.01-second alarm fired before the test expected.
145 The test first slept indefinitely (blocking for signals)
146 and only after that tested for the signal having been sent.
147 Since the signal had already been sent, the test #12 never
148 completed. The solution: test first, then block.
149 - default to being silent on all probing attempts, set the
150 environment variable VERBOSE to a true value to see the
151 details (the probing command and the possible errors)
154 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
155 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
156 cannot be upgraded from CPAN to override the 5.8.0 version
158 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
159 because the include directories did not adjust themselves
160 if $ENV{PERL_CORE} (Hugo van der Sanden)
161 - add documentation about the restart of select() under alarm()
164 - small documentation tweaks
168 - better VMS operation (Craig Berry)
171 - our time machine is accelerating: now works with Perl 5.004_01
172 (tried with 5.003_07 and 5.002 but I get segmentation faults
173 from running the Makefile.PL with those in Tru64 4.0D)
176 - backward compatibility (pre-5.6.0) tweaks:
177 - no XSLoader in 5.00503, use DynaLoader instead
178 - no SvPV_nolen, either
179 - no PerlProc_pause(), either
180 - now tested with 5.00404 and 5.00503
181 - Makefile.PL requires 5.00404 (no more 5.002)
182 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
183 (this means that one can mix subsecond sleeps with alarms)
184 - because of nanosleep we probe for -lrt and -lposix4
185 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
186 is available by exportable constants Time::HiRes::d_func
187 (since older Perl do not have them in %Config, and even
188 5.8.0 does not probe for nanosleep)
191 - backward compatibility (pre-5.6.1) tweaks:
192 - define NV if no NVTYPE
193 - define IVdf if needed (note: the Devel::PPPort
194 in 5.8.0 does not try hard hard enough since
195 the IVSIZE might not be defined)
196 - define NVgf if needed
197 - grab the typemap from 5.8.0 for the NV stuff
200 - release 1.29_02 as 1.30
204 - fix a silly unclosed comment typo in HiRes.xs
205 - document and export REALTIME_REALPROF (Solaris)
209 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
210 (need to patch this also in Perl 5.[89])
211 - remove CVS revision log from HiRes.xs
215 The following numbered patches refer to the Perl 5.7 changes,
216 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
218 - 17558: Add #!./perl to the .t
219 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
220 - 16198: political correctness, from Simon Cozens
221 - 15857: doc tweaks, from Jarkko Hietaniemi
222 - 15593: optimization in .xs, from Paul Green
223 - 14892: pod fixes, from Robin Barker
224 - 14100: VOS fixes, from Paul Green
225 - 13422: XS segfault, from Marc Lehmann
226 - 13378: whether select() gets restarted on signals, depends
227 - 13354: timing constraints, again, from Andy Dougherty
228 - 13278: can't do subecond alarms with ualarm;
229 break out early if alarms do not seem to be working
230 - 13266: test relaxation (cygwin gets lower hires
231 times than lores ones)
232 - 12846: protect against high load, from Jarkko Hietaniemi
233 - 12837: HiRes.t VMS tweak, from Craig A. Berry
234 - 12797: HiRes.t VMS tweak, from Charles Lane
235 - 12769: HiRes.t VMS tweak, from Craig A. Berry
236 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
237 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
238 - 12692: alarm() ain't gonna work if ualarm() ain't,
239 from Gurusamy Sarathy
240 - 12680: minor VMS tweak, from Charles Lane
241 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
242 - 12609: croak on negative time, from Jarkko Hietaniemi
243 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
244 - 12594: MacOS Classic timeofday, from Chris Nandor
245 - 12473: allow for more than one second for sleep() and usleep()
246 - 12458: test tuning, relax timing constraints,
247 from Jarkko Hietaniemi
248 - 12449: make sleep() and usleep() to return the number
249 of seconds and microseconds actually slept (analogously
250 with the builtin sleep()), also make usleep() croak if
251 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
252 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
253 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
254 - 12196: use ftime() on Win32, from Artur Bergman
255 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
256 - 12105: use GetSystemTime() on Win32, from Artur Bergman
257 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
258 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
259 - 11797: problem in HiRes.t, from John P. Linderman
260 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
261 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
262 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
263 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
264 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
265 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
267 - 10942: MPE/IX test tweaks, from Mark Bixby
268 - 10784: unnecessary pod2man calls, from Andy Dougherty
269 - 10354: ext/ + -Wall, from Doug MacEachern
270 - 10320: fix the BOOT section to call myU2time correctly
271 - 10317: correct casting for AIX< from H. Merijn Brand
272 - 10119: document that the core time() may be rounding, not truncating
273 - 10118: test fix, from John Peacock
274 - 9988: long =item, from Robin Barker
275 - 9714: correct test output
276 - 9708: test also the scalar aspect of getitimer()
277 - 9705: Add interval timers (setitimer, getitimer)
278 - 9692: do not require at least 5.005 using XS
280 The following changes were made on top of the changes
281 made for Time::HiRes during the Perl 5.7 development
282 cycle that culminated in the release of Perl 5.8.0.
284 - add "require 5.005" to the Makefile.PL
285 - remove the REVISION section (CVS log) from HiRes.pm
286 - add jhi's copyright alongside Douglas'
287 - move HiRes.pm to lib/Time/
289 - modify HiRes.t to use $ENV{PERL_CORE}
290 - modify the original Time::HiRes version 1.20 Makefile.PL
291 to work both with Perl 5.8.0 and the new code with pre-5.8.0
292 Perls (tried with 5.6.1)
293 - tiny tweaks and updates in README and TODO
294 - bump the VERSION to 1.29
296 1.20 Wed Feb 24 21:30 1999
297 - make our usleep and ualarm substitutes into hrt_usleep
298 and hrt_ualarm. This helps static links of Perl with other
299 packages that also have usleep, etc. From
300 Ilya Zakharevich <ilya@math.ohio-state.edu>
301 - add C API stuff. From Joshua Pritikin
302 <joshua.pritikin@db.com>
303 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
304 - hopefully correct "-lc" fix for SCO.
307 1.19 Tue Sep 29 22:30 1998
308 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
309 <seb@stian.demon.co.uk>
310 - change GIMME_V to GIMME to help people with older versions of
312 - fix Win32 version of gettimeofday(). It didn't affect anything,
313 but it confuses people reading the code when the return value
314 is backwards (0 is success).
315 - fix Makefile.PL (more) so that detection of gettimeofday is
318 1.18 Mon Jul 6 22:40 1998
319 - add usleep() for Win32.
320 - fix Makefile.PL to fix reported HP/UX feature where unresolved
321 externals still cause an executable to be generated (though no
322 x bit set). Thanks to David Kozinn for report and explanation.
323 Problems with the fix are mine :)
325 1.17 Wed Jul 1 20:10 1998
326 - fix setitimer calls so microseconds is not more than 1000000.
327 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
328 - make Win32. We only get gettimeofday (the select hack doesn't
329 seem to work on my Win95 system).
330 - fix test 4 on 01test.t. add test to see if time() and
331 Time::HiRes::time() are close.
333 1.16 Wed Nov 12 21:05 1997
334 - add missing EXTEND in new gettimeofday scalar code.
336 1.15 Mon Nov 10 21:30 1997
337 - HiRes.pm: update pod. Provided by Gisle Aas.
338 - HiRes.xs: if gettimeofday() called in scalar context, do
339 something more useful than before. Provided by Gisle Aas.
340 - README: tell of xsubpp '-nolinenumber' woes. thanks to
341 Edward Henigin <ed@texas.net> for pointing out the problem.
343 1.14 Wed Nov 5 9:40 1997
344 - Makefile.PL: look for setitimer
345 - HiRes.xs: if missing ualarm, but we have setitimer, make up
346 our own setitimer. These were provided by Gisle Aas.
348 1.13 Tue Nov 4 23:30 1997
349 - Makefile.PL: fix autodetect mechanism to do try linking in addition
350 to just compiling; should fix Linux build problem. Fix was provided
353 1.12 Sun Oct 12 12:00:00 1997
354 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
355 you may need to comment this back out if you have an older xsubpp.
356 - HiRes.xs: set PROTOTYPES: DISABLE
358 1.11 Fri Sep 05 16:00:00 1997
360 Had some line commented out that shouldn't have been (testing
363 Previous version was corrupted.
365 1.10 Thu May 22 20:20:00 1997
366 - HiRes.xs, HiRes.pm, t/*:
367 - only compile what we have OS support for (or can
369 - only test what we compiled
370 - gross improvement to the test suite
372 This work was all done by Roderick Schertler
373 <roderick@argon.org>. If you run Linux or
374 one of the other ualarm-less platoforms, and you like this
375 module, let Roderick know; without him, it still wouldn't
376 be working on those boxes...
377 - Makefile.PL: figure out what routines the OS has and
378 only build what we need. These bits were written by Jarkko
379 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
381 1.02 Mon Dec 30 08:00:00 1996
382 - HiRes.pm: update documentation to say what to do when missing
383 ualarm() and friends.
384 - README: update to warn that ualarm() and friends need to exist
386 1.01 Fri Oct 17 08:00:00 1996
387 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
388 - HiRes.pm: put blank line between __END__ and =head1 so that
391 1.00 Tue Sep 03 13:00:00 1996
392 - original version; created by h2xs 1.16