1 Revision history for Perl extension Time::HiRes.
4 - Change the Win32 recalibration limit to 0.5 seconds and tweak
5 the documentation to blather less about the gory details of the
6 Win32 implementation and more about the complications in general
7 of meddling with the system clock.
10 - Document the 1.57 change better.
13 - Win32/Cygwin/MinGW: if the performance counter drifts by more
14 than two seconds from the system clock (due to ntp adjustments,
15 for example), recalibrate our internal counter: from Jan Dubois,
16 based on [cpan #5933] by Jerry D. Hedden.
19 - Give a clearer message if the tests timeout (perl change #22253)
20 - Don't use /tmp or its moral equivalents (perl bug #15036,
24 - Windows: ming32 patch from Mike Pomraning (use Perl's Const64()
25 instead of VC-specific i64 suffix)
28 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
31 - Windows: higher resolution time() by using the Windows
32 performance counter API, from Jan Dubois and Anton Shcherbinin.
33 The exact new higher resolution depends on the hardware,
34 but it should be quite a bit better than using the basic
38 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
39 the subtest 18. No known analysis nor fix, but added
40 an alarm (that requires fork() and alarm()) to the test.
43 - doc tweaks from mjd (perl change #20456)
44 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
47 - add a message (for non-core builds) to Makefile.PL about
48 the LC_ALL=C workaround
49 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
50 - the nanosleep() probe was broken (perl change #20061)
51 - use existence instead of definedness for feature probes
53 - MPE/iX tweak (perl change #20042)
54 - do not use HAS_NANOSLEEP (perl change #19898)
57 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
58 - OS/2 can always mix subsecond sleeps with signals
59 (part of perl change #19789)
62 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
66 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
67 - unnecessary (nanosleep is in libc anyway)
68 - harmful (-lrt slows down execution)
69 - incompatible (with many distributions' pthreads)
72 - do not create files in blib directories under core
73 (perl change #19160, from rgs)
74 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
77 - guarantee that $xdefine in HiRes.t is always defined
78 (perl change #19109, from IlyaZ)
79 - a cleaner way to detect PERL_CORE (perl change #19111,
83 - add hints/irix.pl to turn off overly POSIX flags that
84 cause hide struct timespec to be hidden (and compilation
85 to fail) (bleadperl change #19085)
86 - documentation tweaks
89 - add c:/temp to the list of temp directories to probe
90 so that cygwin (and win*?) builds are happy. This was
91 needed at least in my cygwin 1.3.20/w2k setup.
94 - modernize the constants code (from Nicholas Clark)
97 - At some point the ability to figure our the correct incdir
98 for EXTERN.h (either a core perl build, or an installed perl)
99 had broken (which lead into all test compiles failing with
100 a core perl build, but thanks to the robustness of Makefile.PL
101 nothing of this was visible). The brokenness seemed to be
102 caused by $ENV{PERL_CORE} not being on for core builds?
103 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
104 right, and both styles of build should work again.
107 - Nicholas Clark noticed that the my_catdir() emulation function
108 was broken (which means that we didn't really work for Perls
110 - inspired by fixing the above made the whole Makefile.PL -w
112 - tightened up the Makefile.PL output, less whitespace
115 - fix from Craig Berry for better building in VMS with PERL_CORE
118 - no functional changes
119 - move lib/Time/HiRes.pm as Hires.pm
120 - libraries scanning was slightly broken (always scanned
121 for a library even when $Config{libs} already had it)
124 - Ray Zimmerman ran into a race condition in Mac OS X.
125 A 0.01-second alarm fired before the test expected.
126 The test first slept indefinitely (blocking for signals)
127 and only after that tested for the signal having been sent.
128 Since the signal had already been sent, the test #12 never
129 completed. The solution: test first, then block.
130 - default to being silent on all probing attempts, set the
131 environment variable VERBOSE to a true value to see the
132 details (the probing command and the possible errors)
135 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
136 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
137 cannot be upgraded from CPAN to override the 5.8.0 version
139 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
140 because the include directories did not adjust themselves
141 if $ENV{PERL_CORE} (Hugo van der Sanden)
142 - add documentation about the restart of select() under alarm()
145 - small documentation tweaks
149 - better VMS operation (Craig Berry)
152 - our time machine is accelerating: now works with Perl 5.004_01
153 (tried with 5.003_07 and 5.002 but I get segmentation faults
154 from running the Makefile.PL with those in Tru64 4.0D)
157 - backward compatibility (pre-5.6.0) tweaks:
158 - no XSLoader in 5.00503, use DynaLoader instead
159 - no SvPV_nolen, either
160 - no PerlProc_pause(), either
161 - now tested with 5.00404 and 5.00503
162 - Makefile.PL requires 5.00404 (no more 5.002)
163 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
164 (this means that one can mix subsecond sleeps with alarms)
165 - because of nanosleep we probe for -lrt and -lposix4
166 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
167 is available by exportable constants Time::HiRes::d_func
168 (since older Perl do not have them in %Config, and even
169 5.8.0 does not probe for nanosleep)
172 - backward compatibility (pre-5.6.1) tweaks:
173 - define NV if no NVTYPE
174 - define IVdf if needed (note: the Devel::PPPort
175 in 5.8.0 does not try hard hard enough since
176 the IVSIZE might not be defined)
177 - define NVgf if needed
178 - grab the typemap from 5.8.0 for the NV stuff
181 - release 1.29_02 as 1.30
185 - fix a silly unclosed comment typo in HiRes.xs
186 - document and export REALTIME_REALPROF (Solaris)
190 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
191 (need to patch this also in Perl 5.[89])
192 - remove CVS revision log from HiRes.xs
196 The following numbered patches refer to the Perl 5.7 changes,
197 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
199 - 17558: Add #!./perl to the .t
200 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
201 - 16198: political correctness, from Simon Cozens
202 - 15857: doc tweaks, from Jarkko Hietaniemi
203 - 15593: optimization in .xs, from Paul Green
204 - 14892: pod fixes, from Robin Barker
205 - 14100: VOS fixes, from Paul Green
206 - 13422: XS segfault, from Marc Lehmann
207 - 13378: whether select() gets restarted on signals, depends
208 - 13354: timing constraints, again, from Andy Dougherty
209 - 13278: can't do subecond alarms with ualarm;
210 break out early if alarms do not seem to be working
211 - 13266: test relaxation (cygwin gets lower hires
212 times than lores ones)
213 - 12846: protect against high load, from Jarkko Hietaniemi
214 - 12837: HiRes.t VMS tweak, from Craig A. Berry
215 - 12797: HiRes.t VMS tweak, from Charles Lane
216 - 12769: HiRes.t VMS tweak, from Craig A. Berry
217 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
218 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
219 - 12692: alarm() ain't gonna work if ualarm() ain't,
220 from Gurusamy Sarathy
221 - 12680: minor VMS tweak, from Charles Lane
222 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
223 - 12609: croak on negative time, from Jarkko Hietaniemi
224 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
225 - 12594: MacOS Classic timeofday, from Chris Nandor
226 - 12473: allow for more than one second for sleep() and usleep()
227 - 12458: test tuning, relax timing constraints,
228 from Jarkko Hietaniemi
229 - 12449: make sleep() and usleep() to return the number
230 of seconds and microseconds actually slept (analogously
231 with the builtin sleep()), also make usleep() croak if
232 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
233 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
234 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
235 - 12196: use ftime() on Win32, from Artur Bergman
236 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
237 - 12105: use GetSystemTime() on Win32, from Artur Bergman
238 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
239 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
240 - 11797: problem in HiRes.t, from John P. Linderman
241 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
242 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
243 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
244 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
245 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
246 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
248 - 10942: MPE/IX test tweaks, from Mark Bixby
249 - 10784: unnecessary pod2man calls, from Andy Dougherty
250 - 10354: ext/ + -Wall, from Doug MacEachern
251 - 10320: fix the BOOT section to call myU2time correctly
252 - 10317: correct casting for AIX< from H. Merijn Brand
253 - 10119: document that the core time() may be rounding, not truncating
254 - 10118: test fix, from John Peacock
255 - 9988: long =item, from Robin Barker
256 - 9714: correct test output
257 - 9708: test also the scalar aspect of getitimer()
258 - 9705: Add interval timers (setitimer, getitimer)
259 - 9692: do not require at least 5.005 using XS
261 The following changes were made on top of the changes
262 made for Time::HiRes during the Perl 5.7 development
263 cycle that culminated in the release of Perl 5.8.0.
265 - add "require 5.005" to the Makefile.PL
266 - remove the REVISION section (CVS log) from HiRes.pm
267 - add jhi's copyright alongside Douglas'
268 - move HiRes.pm to lib/Time/
270 - modify HiRes.t to use $ENV{PERL_CORE}
271 - modify the original Time::HiRes version 1.20 Makefile.PL
272 to work both with Perl 5.8.0 and the new code with pre-5.8.0
273 Perls (tried with 5.6.1)
274 - tiny tweaks and updates in README and TODO
275 - bump the VERSION to 1.29
277 1.20 Wed Feb 24 21:30 1999
278 - make our usleep and ualarm substitutes into hrt_usleep
279 and hrt_ualarm. This helps static links of Perl with other
280 packages that also have usleep, etc. From
281 Ilya Zakharevich <ilya@math.ohio-state.edu>
282 - add C API stuff. From Joshua Pritikin
283 <joshua.pritikin@db.com>
284 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
285 - hopefully correct "-lc" fix for SCO.
288 1.19 Tue Sep 29 22:30 1998
289 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
290 <seb@stian.demon.co.uk>
291 - change GIMME_V to GIMME to help people with older versions of
293 - fix Win32 version of gettimeofday(). It didn't affect anything,
294 but it confuses people reading the code when the return value
295 is backwards (0 is success).
296 - fix Makefile.PL (more) so that detection of gettimeofday is
299 1.18 Mon Jul 6 22:40 1998
300 - add usleep() for Win32.
301 - fix Makefile.PL to fix reported HP/UX feature where unresolved
302 externals still cause an executable to be generated (though no
303 x bit set). Thanks to David Kozinn for report and explanation.
304 Problems with the fix are mine :)
306 1.17 Wed Jul 1 20:10 1998
307 - fix setitimer calls so microseconds is not more than 1000000.
308 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
309 - make Win32. We only get gettimeofday (the select hack doesn't
310 seem to work on my Win95 system).
311 - fix test 4 on 01test.t. add test to see if time() and
312 Time::HiRes::time() are close.
314 1.16 Wed Nov 12 21:05 1997
315 - add missing EXTEND in new gettimeofday scalar code.
317 1.15 Mon Nov 10 21:30 1997
318 - HiRes.pm: update pod. Provided by Gisle Aas.
319 - HiRes.xs: if gettimeofday() called in scalar context, do
320 something more useful than before. Provided by Gisle Aas.
321 - README: tell of xsubpp '-nolinenumber' woes. thanks to
322 Edward Henigin <ed@texas.net> for pointing out the problem.
324 1.14 Wed Nov 5 9:40 1997
325 - Makefile.PL: look for setitimer
326 - HiRes.xs: if missing ualarm, but we have setitimer, make up
327 our own setitimer. These were provided by Gisle Aas.
329 1.13 Tue Nov 4 23:30 1997
330 - Makefile.PL: fix autodetect mechanism to do try linking in addition
331 to just compiling; should fix Linux build problem. Fix was provided
334 1.12 Sun Oct 12 12:00:00 1997
335 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
336 you may need to comment this back out if you have an older xsubpp.
337 - HiRes.xs: set PROTOTYPES: DISABLE
339 1.11 Fri Sep 05 16:00:00 1997
341 Had some line commented out that shouldn't have been (testing
344 Previous version was corrupted.
346 1.10 Thu May 22 20:20:00 1997
347 - HiRes.xs, HiRes.pm, t/*:
348 - only compile what we have OS support for (or can
350 - only test what we compiled
351 - gross improvement to the test suite
353 This work was all done by Roderick Schertler
354 <roderick@argon.org>. If you run Linux or
355 one of the other ualarm-less platoforms, and you like this
356 module, let Roderick know; without him, it still wouldn't
357 be working on those boxes...
358 - Makefile.PL: figure out what routines the OS has and
359 only build what we need. These bits were written by Jarkko
360 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
362 1.02 Mon Dec 30 08:00:00 1996
363 - HiRes.pm: update documentation to say what to do when missing
364 ualarm() and friends.
365 - README: update to warn that ualarm() and friends need to exist
367 1.01 Fri Oct 17 08:00:00 1996
368 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
369 - HiRes.pm: put blank line between __END__ and =head1 so that
372 1.00 Tue Sep 03 13:00:00 1996
373 - original version; created by h2xs 1.16