1 Revision history for Perl extension Time::HiRes.
4 - Win32 and any ithread build: ppport.h didn't define
5 MY_CXT_CLONE, which seems to be a Time-HiResism.
8 - Skip testing if under PERL_CORE and Time::HiRes has not
9 been Configured (from Marcus Holland-Moritz, core change
11 - Use ppport.h generated by Devel::PPPort 3.01,
12 allowing cutting away our own portability code.
13 - Don't use $ENV{PERL_CORE} for < 5.6.0.
14 - Don't use "for my $i" for <= 5.003.
15 - Don't use Pause() for <= 5.003.
16 - Can't use newSVpvf for <= 5.003.
17 (most of the changes from Marcus)
20 - Win32: reset reading from the performance counters every
21 five minutes to better track wall clock time (thanks to
22 PC timers being often quite bad), should help long-running
26 - Win32: Patch from Steve Hay
27 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
28 to [perl #30755] reported by Nigel Sandever
30 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
31 <w32api/windows.h> APIs are available. Cygwin testing by
32 Yitzchak Scott-Thoennes.
34 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
35 after that keep using -lrt, patch from Alan Burlison,
36 bug reported in [cpan #7165]
39 - Change the Win32 recalibration limit to 0.5 seconds and tweak
40 the documentation to blather less about the gory details of the
41 Win32 implementation and more about the complications in general
42 of meddling with the system clock.
45 - Document the 1.57 change better.
48 - Win32/Cygwin/MinGW: if the performance counter drifts by more
49 than two seconds from the system clock (due to ntp adjustments,
50 for example), recalibrate our internal counter: from Jan Dubois,
51 based on [cpan #5933] by Jerry D. Hedden.
54 - Give a clearer message if the tests timeout (perl change #22253)
55 - Don't use /tmp or its moral equivalents (perl bug #15036,
59 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
60 instead of VC-specific i64 suffix)
63 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
66 - Windows: higher resolution time() by using the Windows
67 performance counter API, from Jan Dubois and Anton Shcherbinin.
68 The exact new higher resolution depends on the hardware,
69 but it should be quite a bit better than using the basic
73 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
74 the subtest 18. No known analysis nor fix, but added
75 an alarm (that requires fork() and alarm()) to the test.
78 - doc tweaks from mjd (perl change #20456)
79 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
82 - add a message (for non-core builds) to Makefile.PL about
83 the LC_ALL=C workaround
84 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
85 - the nanosleep() probe was broken (perl change #20061)
86 - use existence instead of definedness for feature probes
88 - MPE/iX tweak (perl change #20042)
89 - do not use HAS_NANOSLEEP (perl change #19898)
92 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
93 - OS/2 can always mix subsecond sleeps with signals
94 (part of perl change #19789)
97 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
101 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
102 - unnecessary (nanosleep is in libc anyway)
103 - harmful (-lrt slows down execution)
104 - incompatible (with many distributions' pthreads)
107 - do not create files in blib directories under core
108 (perl change #19160, from rgs)
109 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
112 - guarantee that $xdefine in HiRes.t is always defined
113 (perl change #19109, from IlyaZ)
114 - a cleaner way to detect PERL_CORE (perl change #19111,
118 - add hints/irix.pl to turn off overly POSIX flags that
119 cause hide struct timespec to be hidden (and compilation
120 to fail) (bleadperl change #19085)
121 - documentation tweaks
124 - add c:/temp to the list of temp directories to probe
125 so that cygwin (and win*?) builds are happy. This was
126 needed at least in my cygwin 1.3.20/w2k setup.
129 - modernize the constants code (from Nicholas Clark)
132 - At some point the ability to figure our the correct incdir
133 for EXTERN.h (either a core perl build, or an installed perl)
134 had broken (which lead into all test compiles failing with
135 a core perl build, but thanks to the robustness of Makefile.PL
136 nothing of this was visible). The brokenness seemed to be
137 caused by $ENV{PERL_CORE} not being on for core builds?
138 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
139 right, and both styles of build should work again.
142 - Nicholas Clark noticed that the my_catdir() emulation function
143 was broken (which means that we didn't really work for Perls
145 - inspired by fixing the above made the whole Makefile.PL -w
147 - tightened up the Makefile.PL output, less whitespace
150 - fix from Craig Berry for better building in VMS with PERL_CORE
153 - no functional changes
154 - move lib/Time/HiRes.pm as Hires.pm
155 - libraries scanning was slightly broken (always scanned
156 for a library even when $Config{libs} already had it)
159 - Ray Zimmerman ran into a race condition in Mac OS X.
160 A 0.01-second alarm fired before the test expected.
161 The test first slept indefinitely (blocking for signals)
162 and only after that tested for the signal having been sent.
163 Since the signal had already been sent, the test #12 never
164 completed. The solution: test first, then block.
165 - default to being silent on all probing attempts, set the
166 environment variable VERBOSE to a true value to see the
167 details (the probing command and the possible errors)
170 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
171 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
172 cannot be upgraded from CPAN to override the 5.8.0 version
174 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
175 because the include directories did not adjust themselves
176 if $ENV{PERL_CORE} (Hugo van der Sanden)
177 - add documentation about the restart of select() under alarm()
180 - small documentation tweaks
184 - better VMS operation (Craig Berry)
187 - our time machine is accelerating: now works with Perl 5.004_01
188 (tried with 5.003_07 and 5.002 but I get segmentation faults
189 from running the Makefile.PL with those in Tru64 4.0D)
192 - backward compatibility (pre-5.6.0) tweaks:
193 - no XSLoader in 5.00503, use DynaLoader instead
194 - no SvPV_nolen, either
195 - no PerlProc_pause(), either
196 - now tested with 5.00404 and 5.00503
197 - Makefile.PL requires 5.00404 (no more 5.002)
198 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
199 (this means that one can mix subsecond sleeps with alarms)
200 - because of nanosleep we probe for -lrt and -lposix4
201 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
202 is available by exportable constants Time::HiRes::d_func
203 (since older Perl do not have them in %Config, and even
204 5.8.0 does not probe for nanosleep)
207 - backward compatibility (pre-5.6.1) tweaks:
208 - define NV if no NVTYPE
209 - define IVdf if needed (note: the Devel::PPPort
210 in 5.8.0 does not try hard hard enough since
211 the IVSIZE might not be defined)
212 - define NVgf if needed
213 - grab the typemap from 5.8.0 for the NV stuff
216 - release 1.29_02 as 1.30
220 - fix a silly unclosed comment typo in HiRes.xs
221 - document and export REALTIME_REALPROF (Solaris)
225 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
226 (need to patch this also in Perl 5.[89])
227 - remove CVS revision log from HiRes.xs
231 The following numbered patches refer to the Perl 5.7 changes,
232 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
234 - 17558: Add #!./perl to the .t
235 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
236 - 16198: political correctness, from Simon Cozens
237 - 15857: doc tweaks, from Jarkko Hietaniemi
238 - 15593: optimization in .xs, from Paul Green
239 - 14892: pod fixes, from Robin Barker
240 - 14100: VOS fixes, from Paul Green
241 - 13422: XS segfault, from Marc Lehmann
242 - 13378: whether select() gets restarted on signals, depends
243 - 13354: timing constraints, again, from Andy Dougherty
244 - 13278: can't do subecond alarms with ualarm;
245 break out early if alarms do not seem to be working
246 - 13266: test relaxation (cygwin gets lower hires
247 times than lores ones)
248 - 12846: protect against high load, from Jarkko Hietaniemi
249 - 12837: HiRes.t VMS tweak, from Craig A. Berry
250 - 12797: HiRes.t VMS tweak, from Charles Lane
251 - 12769: HiRes.t VMS tweak, from Craig A. Berry
252 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
253 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
254 - 12692: alarm() ain't gonna work if ualarm() ain't,
255 from Gurusamy Sarathy
256 - 12680: minor VMS tweak, from Charles Lane
257 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
258 - 12609: croak on negative time, from Jarkko Hietaniemi
259 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
260 - 12594: MacOS Classic timeofday, from Chris Nandor
261 - 12473: allow for more than one second for sleep() and usleep()
262 - 12458: test tuning, relax timing constraints,
263 from Jarkko Hietaniemi
264 - 12449: make sleep() and usleep() to return the number
265 of seconds and microseconds actually slept (analogously
266 with the builtin sleep()), also make usleep() croak if
267 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
268 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
269 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
270 - 12196: use ftime() on Win32, from Artur Bergman
271 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
272 - 12105: use GetSystemTime() on Win32, from Artur Bergman
273 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
274 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
275 - 11797: problem in HiRes.t, from John P. Linderman
276 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
277 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
278 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
279 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
280 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
281 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
283 - 10942: MPE/IX test tweaks, from Mark Bixby
284 - 10784: unnecessary pod2man calls, from Andy Dougherty
285 - 10354: ext/ + -Wall, from Doug MacEachern
286 - 10320: fix the BOOT section to call myU2time correctly
287 - 10317: correct casting for AIX< from H. Merijn Brand
288 - 10119: document that the core time() may be rounding, not truncating
289 - 10118: test fix, from John Peacock
290 - 9988: long =item, from Robin Barker
291 - 9714: correct test output
292 - 9708: test also the scalar aspect of getitimer()
293 - 9705: Add interval timers (setitimer, getitimer)
294 - 9692: do not require at least 5.005 using XS
296 The following changes were made on top of the changes
297 made for Time::HiRes during the Perl 5.7 development
298 cycle that culminated in the release of Perl 5.8.0.
300 - add "require 5.005" to the Makefile.PL
301 - remove the REVISION section (CVS log) from HiRes.pm
302 - add jhi's copyright alongside Douglas'
303 - move HiRes.pm to lib/Time/
305 - modify HiRes.t to use $ENV{PERL_CORE}
306 - modify the original Time::HiRes version 1.20 Makefile.PL
307 to work both with Perl 5.8.0 and the new code with pre-5.8.0
308 Perls (tried with 5.6.1)
309 - tiny tweaks and updates in README and TODO
310 - bump the VERSION to 1.29
312 1.20 Wed Feb 24 21:30 1999
313 - make our usleep and ualarm substitutes into hrt_usleep
314 and hrt_ualarm. This helps static links of Perl with other
315 packages that also have usleep, etc. From
316 Ilya Zakharevich <ilya@math.ohio-state.edu>
317 - add C API stuff. From Joshua Pritikin
318 <joshua.pritikin@db.com>
319 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
320 - hopefully correct "-lc" fix for SCO.
323 1.19 Tue Sep 29 22:30 1998
324 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
325 <seb@stian.demon.co.uk>
326 - change GIMME_V to GIMME to help people with older versions of
328 - fix Win32 version of gettimeofday(). It didn't affect anything,
329 but it confuses people reading the code when the return value
330 is backwards (0 is success).
331 - fix Makefile.PL (more) so that detection of gettimeofday is
334 1.18 Mon Jul 6 22:40 1998
335 - add usleep() for Win32.
336 - fix Makefile.PL to fix reported HP/UX feature where unresolved
337 externals still cause an executable to be generated (though no
338 x bit set). Thanks to David Kozinn for report and explanation.
339 Problems with the fix are mine :)
341 1.17 Wed Jul 1 20:10 1998
342 - fix setitimer calls so microseconds is not more than 1000000.
343 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
344 - make Win32. We only get gettimeofday (the select hack doesn't
345 seem to work on my Win95 system).
346 - fix test 4 on 01test.t. add test to see if time() and
347 Time::HiRes::time() are close.
349 1.16 Wed Nov 12 21:05 1997
350 - add missing EXTEND in new gettimeofday scalar code.
352 1.15 Mon Nov 10 21:30 1997
353 - HiRes.pm: update pod. Provided by Gisle Aas.
354 - HiRes.xs: if gettimeofday() called in scalar context, do
355 something more useful than before. Provided by Gisle Aas.
356 - README: tell of xsubpp '-nolinenumber' woes. thanks to
357 Edward Henigin <ed@texas.net> for pointing out the problem.
359 1.14 Wed Nov 5 9:40 1997
360 - Makefile.PL: look for setitimer
361 - HiRes.xs: if missing ualarm, but we have setitimer, make up
362 our own setitimer. These were provided by Gisle Aas.
364 1.13 Tue Nov 4 23:30 1997
365 - Makefile.PL: fix autodetect mechanism to do try linking in addition
366 to just compiling; should fix Linux build problem. Fix was provided
369 1.12 Sun Oct 12 12:00:00 1997
370 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
371 you may need to comment this back out if you have an older xsubpp.
372 - HiRes.xs: set PROTOTYPES: DISABLE
374 1.11 Fri Sep 05 16:00:00 1997
376 Had some line commented out that shouldn't have been (testing
379 Previous version was corrupted.
381 1.10 Thu May 22 20:20:00 1997
382 - HiRes.xs, HiRes.pm, t/*:
383 - only compile what we have OS support for (or can
385 - only test what we compiled
386 - gross improvement to the test suite
388 This work was all done by Roderick Schertler
389 <roderick@argon.org>. If you run Linux or
390 one of the other ualarm-less platoforms, and you like this
391 module, let Roderick know; without him, it still wouldn't
392 be working on those boxes...
393 - Makefile.PL: figure out what routines the OS has and
394 only build what we need. These bits were written by Jarkko
395 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
397 1.02 Mon Dec 30 08:00:00 1996
398 - HiRes.pm: update documentation to say what to do when missing
399 ualarm() and friends.
400 - README: update to warn that ualarm() and friends need to exist
402 1.01 Fri Oct 17 08:00:00 1996
403 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
404 - HiRes.pm: put blank line between __END__ and =head1 so that
407 1.00 Tue Sep 03 13:00:00 1996
408 - original version; created by h2xs 1.16