1 Revision history for Perl extension Time::HiRes.
4 - has_symbol() was wrong since e.g. ITIMER_VIRTUAL is exported
5 via @EXPORT_OK even when it is not available. This is heinous.
6 @EXPORT_OK should be determined at Makefile.PL time.
7 - be more lenient is testing clock_gettime(): allow more slop,
8 and retry up to three times, sleeping a random nap between
10 - human months are one-based (noticed by Anton Berezin)
13 - CLOCK_REALTIME is an enum value (of the clockid_t enum)
14 in HP-UX (and might be so elsewhere, too), debugged by
16 - include const-c.inc as late as possible (from Randy Kobes,
17 [rt.cpan.org #15552] to avoid undefined usleep() on Win32
20 - try to be more robust and consistent in the detection of
21 CLOCK_REALTIME and ITIMER_VIRTUAL in HiRes.t: the proper
26 eval 'import Time::HiRes qw($symbol)';
27 return 0 unless $@ eq '';
28 return exists ${"Time::HiRes::$symbol"};
35 in the test. All these moves are needed because
37 1) one cannot directly do eval 'Time::HiRes::FOO_BAR'
38 because FOO_BAR might have a true value of zero
39 (or in the general case an empty string or even undef)
41 2) In case FOO_BAR is not available in this platform,
42 &FOO_BAR avoids the bareword warning
44 - wait more (1.5 seconds instead of 0.1) for the CLOCK_REALTIME test
45 but expect the 'customary' slop of 0.20 instead of 0.25
46 - fixed inside a comment HAS_POLL -> TIME_HIRES_NANOSLEEP
47 - at the end of HiRest.t tell how close we were to termination
50 - Gisle noticed a mistake (using HAS_NANOSLEEP) in 1.79
53 - try nanosleep for emulating usleep -- may help in some weird
54 embedded realtime places which have nanosleep but neither usleep
55 nor select nor poll (doesn't have to be weird embedded realtime
56 place, though -- in many places usleep is nanosleep anyway)
57 - try poll for emulating usleep -- this may help some obscure/old
58 SVR4 places that have neither usleep nor select
59 - a redundant test guard in HiRes.t
62 - ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
63 in the general case fail already at 'use' phase is suboptimal)
64 - fixes to the documentation of clock_gettime() and clock_getres()
67 - add support for the POSIX clock_gettime() and clock_getres(),
68 if available, either as library calls or as syscalls
69 - be more defensive about missing functionality: break out
70 early (during 'use') if no e.g. clock_getres() is available,
71 and protect our back by trapping those cases also in HiRes.xs
72 - the test added in 1.76 could cause an endless loop e.g. in Solaris,
73 due to mixing of sleep() and alarm() (bad programmer, no cookie!)
76 - testing for nanosleep had wrong logic which caused nanosleep
77 to become undefined for e.g. Mac OS X
78 - added a test for a core dump that was introduced by Perl 5.8.0
79 safe signals and was fixed for the time of 5.8.1 (one report of
80 the core dump was [perl #20920]), the test skipped pre-5.8.1.
81 - *cough* s/unanosleep/nanosleep/g; *cough*
84 - installation patch from Gisle Aas: in Perls 5.8.x and later
85 use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
88 - [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
89 (the workaround is not to use rel2abs, should not be necessary)
90 - [cpan #14642] U2time wrongly exported on the C API
91 (patch supplied by the reporter, SALVA@cpan.org)
92 - add release dates to Changes
95 - Time::HiRes::nanosleep support for Solaris [PATCH]
96 (POSIX::uname() not available if building with core perl,
97 from Gisle Aas, via perl5-porters, perl change #25295)
100 - going back to the 1.68 loader setup (using DynaLoader)
101 since too many weird things starting breaking
102 - fix a typo in José Auguste-Etienne's name
105 - a thinko in the nanosleep() detection
106 - move more changes stuff from the README to Changes
107 - add -w to the Makefile.PL
110 - oops in 1.69 about @ISA (not affecting anything but silly)
111 - add copyright 2005 to HiRes.pm
112 - add copyright and license to HiRes.xs
113 - add copyrights 2003, 2004, 2005 to README
116 - actually run a test for nanosleep
117 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
118 it seems that one can link in nanosleep() but then calling
119 it fails instantly and sets errno to ENOSYS (Not implemented).
120 This may be fixable in the AIX case by figuring out the right
121 (realtime POSIX?) libs and whatnot, but in the general case
122 running a real test case is better. (Of course, this change
123 will no doubt run into portability problems because of the
124 execution step...) Note that because of hysterical raisins
125 most Perls do NOT have $Config{d_nanosleep} (scanning for
126 it by Configure would in many platforms require linking in
127 things like -lrt, which would in many platforms be a bad idea
129 (from José Auguste-Etienne)
130 - support XSLoader also since it's much faster
131 (from Alexey Tourbin)
132 - add SEE ALSO (BSD::Resource and Time::TAI64)
135 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
136 error suspected), but miraculously it still worked since the
137 doubling took place below the __END__ token
138 - undef Pause() before defining it to avoid redefinition warnings
139 during compilation in case perl.h had already defined Pause()
140 (part of perl change #24271)
144 - (internal) don't ignore the return value of gettimeofday()
145 - (external) return undef or an empty if the C gettimeofday() fails
146 (affects Time::HiRes gettimeofday() and the hires time())
150 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
151 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
152 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
153 - close also [rt.cpan.org #5933] "Time::HiRes::time does not
154 pick up time adjustments like ntp" since ever reproducing it
155 (and therefore verifying a possible fix) in the same environment
156 has become rather unlikely
159 - one should not mix u?alarm and sleep (the tests modified
160 by 1.65, #12 and #13, hung in Solaris), now we just busy
161 loop executing an empty block
162 - in the documentation underline the unspecificity of mixing
164 - small spelling fixes
167 - regenerate ppport.h with Devel::PPPort 3.03,
168 now the MY_CXT_CLONE is defined in ppport.h,
169 we no more need to do that.
171 - the test #12 would often hang in sigsuspend() (at least that's
172 where Mac OS X' ktrace shows it hanging). With the sleep()s
173 changed to sleep(1)s, the tests still pass but no hang after
174 a few hundred repeats.
177 - Win32 and any ithread build: ppport.h didn't define
178 MY_CXT_CLONE, which seems to be a Time-HiResism.
181 - Skip testing if under PERL_CORE and Time::HiRes has not
182 been Configured (from Marcus Holland-Moritz, core change
184 - Use ppport.h generated by Devel::PPPort 3.01,
185 allowing cutting away our own portability code.
186 - Don't use $ENV{PERL_CORE} for < 5.6.0.
187 - Don't use "for my $i" for <= 5.003.
188 - Don't use Pause() for <= 5.003.
189 - Can't use newSVpvf for <= 5.003.
190 (most of the changes from Marcus)
193 - Win32: reset reading from the performance counters every
194 five minutes to better track wall clock time (thanks to
195 PC timers being often quite bad), should help long-running
199 - Win32: Patch from Steve Hay
200 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
201 to [perl #30755] reported by Nigel Sandever
203 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
204 <w32api/windows.h> APIs are available. Cygwin testing by
205 Yitzchak Scott-Thoennes.
207 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
208 after that keep using -lrt, patch from Alan Burlison,
209 bug reported in [cpan #7165]
212 - Change the Win32 recalibration limit to 0.5 seconds and tweak
213 the documentation to blather less about the gory details of the
214 Win32 implementation and more about the complications in general
215 of meddling with the system clock.
218 - Document the 1.57 change better.
221 - Win32/Cygwin/MinGW: if the performance counter drifts by more
222 than two seconds from the system clock (due to ntp adjustments,
223 for example), recalibrate our internal counter: from Jan Dubois,
224 based on [cpan #5933] by Jerry D. Hedden.
227 - Give a clearer message if the tests timeout (perl change #22253)
228 - Don't use /tmp or its moral equivalents (perl bug #15036,
232 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
233 instead of VC-specific i64 suffix)
236 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
239 - Windows: higher resolution time() by using the Windows
240 performance counter API, from Jan Dubois and Anton Shcherbinin.
241 The exact new higher resolution depends on the hardware,
242 but it should be quite a bit better than using the basic
246 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
247 the subtest 18. No known analysis nor fix, but added
248 an alarm (that requires fork() and alarm()) to the test.
251 - doc tweaks from mjd (perl change #20456)
252 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
255 - add a message (for non-core builds) to Makefile.PL about
256 the LC_ALL=C workaround
257 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
258 - the nanosleep() probe was broken (perl change #20061)
259 - use existence instead of definedness for feature probes
261 - MPE/iX tweak (perl change #20042)
262 - do not use HAS_NANOSLEEP (perl change #19898)
265 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
266 - OS/2 can always mix subsecond sleeps with signals
267 (part of perl change #19789)
270 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
274 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
275 - unnecessary (nanosleep is in libc anyway)
276 - harmful (-lrt slows down execution)
277 - incompatible (with many distributions' pthreads)
280 - do not create files in blib directories under core
281 (perl change #19160, from rgs)
282 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
285 - guarantee that $xdefine in HiRes.t is always defined
286 (perl change #19109, from IlyaZ)
287 - a cleaner way to detect PERL_CORE (perl change #19111,
291 - add hints/irix.pl to turn off overly POSIX flags that
292 cause hide struct timespec to be hidden (and compilation
293 to fail) (bleadperl change #19085)
294 - documentation tweaks
297 - add c:/temp to the list of temp directories to probe
298 so that cygwin (and win*?) builds are happy. This was
299 needed at least in my cygwin 1.3.20/w2k setup.
302 - modernize the constants code (from Nicholas Clark)
305 - At some point the ability to figure our the correct incdir
306 for EXTERN.h (either a core perl build, or an installed perl)
307 had broken (which lead into all test compiles failing with
308 a core perl build, but thanks to the robustness of Makefile.PL
309 nothing of this was visible). The brokenness seemed to be
310 caused by $ENV{PERL_CORE} not being on for core builds?
311 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
312 right, and both styles of build should work again.
315 - Nicholas Clark noticed that the my_catdir() emulation function
316 was broken (which means that we didn't really work for Perls
318 - inspired by fixing the above made the whole Makefile.PL -w
320 - tightened up the Makefile.PL output, less whitespace
323 - fix from Craig Berry for better building in VMS with PERL_CORE
326 - no functional changes
327 - move lib/Time/HiRes.pm as Hires.pm
328 - libraries scanning was slightly broken (always scanned
329 for a library even when $Config{libs} already had it)
332 - Ray Zimmerman ran into a race condition in Mac OS X.
333 A 0.01-second alarm fired before the test expected.
334 The test first slept indefinitely (blocking for signals)
335 and only after that tested for the signal having been sent.
336 Since the signal had already been sent, the test #12 never
337 completed. The solution: test first, then block.
338 - default to being silent on all probing attempts, set the
339 environment variable VERBOSE to a true value to see the
340 details (the probing command and the possible errors)
343 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
344 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
345 cannot be upgraded from CPAN to override the 5.8.0 version
347 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
348 because the include directories did not adjust themselves
349 if $ENV{PERL_CORE} (Hugo van der Sanden)
350 - add documentation about the restart of select() under alarm()
353 - small documentation tweaks
357 - better VMS operation (Craig Berry)
360 - our time machine is accelerating: now works with Perl 5.004_01
361 (tried with 5.003_07 and 5.002 but I get segmentation faults
362 from running the Makefile.PL with those in Tru64 4.0D)
365 - backward compatibility (pre-5.6.0) tweaks:
366 - no XSLoader in 5.00503, use DynaLoader instead
367 - no SvPV_nolen, either
368 - no PerlProc_pause(), either
369 - now tested with 5.00404 and 5.00503
370 - Makefile.PL requires 5.00404 (no more 5.002)
371 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
372 (this means that one can mix subsecond sleeps with alarms)
373 - because of nanosleep we probe for -lrt and -lposix4
374 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
375 is available by exportable constants Time::HiRes::d_func
376 (since older Perl do not have them in %Config, and even
377 5.8.0 does not probe for nanosleep)
380 - backward compatibility (pre-5.6.1) tweaks:
381 - define NV if no NVTYPE
382 - define IVdf if needed (note: the Devel::PPPort
383 in 5.8.0 does not try hard hard enough since
384 the IVSIZE might not be defined)
385 - define NVgf if needed
386 - grab the typemap from 5.8.0 for the NV stuff
388 1.31 and 1.32 add more backward compatibility (now all the way
389 back to Perl 5.00404), and using nanosleep() (if available) for
394 - release 1.29_02 as 1.30
396 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
397 development cycle. Most notably portability across platforms has been
398 enhanced, and the interval timers (setitimer, getitimer) have been
399 added. Note that the version of Time::HiRes that is included in Perl
400 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
401 version. Note also that in 1.30 Wegscheid turns over the maintenance
402 to Jarkko Hietaniemi.
406 - fix a silly unclosed comment typo in HiRes.xs
407 - document and export REALTIME_REALPROF (Solaris)
411 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
412 (need to patch this also in Perl 5.[89])
413 - remove CVS revision log from HiRes.xs
417 The following numbered patches refer to the Perl 5.7 changes,
418 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
420 - 17558: Add #!./perl to the .t
421 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
422 - 16198: political correctness, from Simon Cozens
423 - 15857: doc tweaks, from Jarkko Hietaniemi
424 - 15593: optimization in .xs, from Paul Green
425 - 14892: pod fixes, from Robin Barker
426 - 14100: VOS fixes, from Paul Green
427 - 13422: XS segfault, from Marc Lehmann
428 - 13378: whether select() gets restarted on signals, depends
429 - 13354: timing constraints, again, from Andy Dougherty
430 - 13278: can't do subecond alarms with ualarm;
431 break out early if alarms do not seem to be working
432 - 13266: test relaxation (cygwin gets lower hires
433 times than lores ones)
434 - 12846: protect against high load, from Jarkko Hietaniemi
435 - 12837: HiRes.t VMS tweak, from Craig A. Berry
436 - 12797: HiRes.t VMS tweak, from Charles Lane
437 - 12769: HiRes.t VMS tweak, from Craig A. Berry
438 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
439 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
440 - 12692: alarm() ain't gonna work if ualarm() ain't,
441 from Gurusamy Sarathy
442 - 12680: minor VMS tweak, from Charles Lane
443 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
444 - 12609: croak on negative time, from Jarkko Hietaniemi
445 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
446 - 12594: MacOS Classic timeofday, from Chris Nandor
447 - 12473: allow for more than one second for sleep() and usleep()
448 - 12458: test tuning, relax timing constraints,
449 from Jarkko Hietaniemi
450 - 12449: make sleep() and usleep() to return the number
451 of seconds and microseconds actually slept (analogously
452 with the builtin sleep()), also make usleep() croak if
453 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
454 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
455 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
456 - 12196: use ftime() on Win32, from Artur Bergman
457 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
458 - 12105: use GetSystemTime() on Win32, from Artur Bergman
459 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
460 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
461 - 11797: problem in HiRes.t, from John P. Linderman
462 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
463 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
464 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
465 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
466 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
467 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
469 - 10942: MPE/IX test tweaks, from Mark Bixby
470 - 10784: unnecessary pod2man calls, from Andy Dougherty
471 - 10354: ext/ + -Wall, from Doug MacEachern
472 - 10320: fix the BOOT section to call myU2time correctly
473 - 10317: correct casting for AIX< from H. Merijn Brand
474 - 10119: document that the core time() may be rounding, not truncating
475 - 10118: test fix, from John Peacock
476 - 9988: long =item, from Robin Barker
477 - 9714: correct test output
478 - 9708: test also the scalar aspect of getitimer()
479 - 9705: Add interval timers (setitimer, getitimer)
480 - 9692: do not require at least 5.005 using XS
482 The following changes were made on top of the changes
483 made for Time::HiRes during the Perl 5.7 development
484 cycle that culminated in the release of Perl 5.8.0.
486 - add "require 5.005" to the Makefile.PL
487 - remove the REVISION section (CVS log) from HiRes.pm
488 - add jhi's copyright alongside Douglas'
489 - move HiRes.pm to lib/Time/
491 - modify HiRes.t to use $ENV{PERL_CORE}
492 - modify the original Time::HiRes version 1.20 Makefile.PL
493 to work both with Perl 5.8.0 and the new code with pre-5.8.0
494 Perls (tried with 5.6.1)
495 - tiny tweaks and updates in README and TODO
496 - bump the VERSION to 1.29
498 1.20 Wed Feb 24 21:30 1999
499 - make our usleep and ualarm substitutes into hrt_usleep
500 and hrt_ualarm. This helps static links of Perl with other
501 packages that also have usleep, etc. From
502 Ilya Zakharevich <ilya@math.ohio-state.edu>
503 - add C API stuff. From Joshua Pritikin
504 <joshua.pritikin@db.com>
505 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
506 - hopefully correct "-lc" fix for SCO.
509 1.20 adds a platform neutral set of C accessible routines if you are
510 running 5.005+. All other changes are packaging changes and build
511 fixes(?) for statically linked Perl, SCO, and VMS.
513 1.19 Tue Sep 29 22:30 1998
514 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
515 <seb@stian.demon.co.uk>
516 - change GIMME_V to GIMME to help people with older versions of
518 - fix Win32 version of gettimeofday(). It didn't affect anything,
519 but it confuses people reading the code when the return value
520 is backwards (0 is success).
521 - fix Makefile.PL (more) so that detection of gettimeofday is
524 1.19 has better VMS support.
526 1.18 Mon Jul 6 22:40 1998
527 - add usleep() for Win32.
528 - fix Makefile.PL to fix reported HP/UX feature where unresolved
529 externals still cause an executable to be generated (though no
530 x bit set). Thanks to David Kozinn for report and explanation.
531 Problems with the fix are mine :)
533 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
534 Probably buggy. I'm sure I'll hear.
536 1.17 Wed Jul 1 20:10 1998
537 - fix setitimer calls so microseconds is not more than 1000000.
538 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
539 - make Win32. We only get gettimeofday (the select hack doesn't
540 seem to work on my Win95 system).
541 - fix test 4 on 01test.t. add test to see if time() and
542 Time::HiRes::time() are close.
544 1.16 Wed Nov 12 21:05 1997
545 - add missing EXTEND in new gettimeofday scalar code.
547 1.16+ should be closer to building out of the box on Linux. Thanks
548 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
550 If your underlying operating system doesn't implement ualarm(), then
551 a fake using setitimer() will be made. If the OS is missing usleep(),
552 a fake one using select() will be made. If a fake can't be made for
553 either ualarm() or usleep(), then the corresponding Perl function will
554 not be available. If the OS is missing gettimeofday(), you will get
555 unresolved externals, either at link- or run-time.
557 This is an improvement; the package used to not even build if
558 you were missing any of these bits. Roderick Schertler
560 <roderick@argon.org> did all the conditional compilation stuff,
561 look at HiRes.pm and the test suites; it's good educational reading.
563 1.15 Mon Nov 10 21:30 1997
564 - HiRes.pm: update pod. Provided by Gisle Aas.
565 - HiRes.xs: if gettimeofday() called in scalar context, do
566 something more useful than before. Provided by Gisle Aas.
567 - README: tell of xsubpp '-nolinenumber' woes. thanks to
568 Edward Henigin <ed@texas.net> for pointing out the problem.
570 1.14 Wed Nov 5 9:40 1997
571 - Makefile.PL: look for setitimer
572 - HiRes.xs: if missing ualarm, but we have setitimer, make up
573 our own setitimer. These were provided by Gisle Aas.
575 1.13 Tue Nov 4 23:30 1997
576 - Makefile.PL: fix autodetect mechanism to do try linking in addition
577 to just compiling; should fix Linux build problem. Fix was provided
580 1.12 Sun Oct 12 12:00:00 1997
581 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
582 you may need to comment this back out if you have an older xsubpp.
583 - HiRes.xs: set PROTOTYPES: DISABLE
585 1.11 Fri Sep 05 16:00:00 1997
587 Had some line commented out that shouldn't have been (testing
590 Previous version was corrupted.
592 1.10 Thu May 22 20:20:00 1997
593 - HiRes.xs, HiRes.pm, t/*:
594 - only compile what we have OS support for (or can
596 - only test what we compiled
597 - gross improvement to the test suite
599 This work was all done by Roderick Schertler
600 <roderick@argon.org>. If you run Linux or
601 one of the other ualarm-less platoforms, and you like this
602 module, let Roderick know; without him, it still wouldn't
603 be working on those boxes...
604 - Makefile.PL: figure out what routines the OS has and
605 only build what we need. These bits were written by Jarkko
606 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
608 1.02 Mon Dec 30 08:00:00 1996
609 - HiRes.pm: update documentation to say what to do when missing
610 ualarm() and friends.
611 - README: update to warn that ualarm() and friends need to exist
613 1.01 Fri Oct 17 08:00:00 1996
614 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
615 - HiRes.pm: put blank line between __END__ and =head1 so that
618 1.00 Tue Sep 03 13:00:00 1996
619 - original version; created by h2xs 1.16