Re: g77/F90 date_and_time() intrinsic

Re: g77/F90 date_and_time() intrinsic

[N8TM]

If you're using f90, you won't be getting the same version of date_and_time
which you get with g77 alone.  I bear some responsibility for the latter,
having persuaded Dave Love to include it in libU77.  The g77 version, of
course, doesn't deal with key words, but if it has other problems, I'll try to
fix them.

Re: g77/F90 date_and_time() intrinsic

[Dave Love]

>>>>> "Tim" == Tim Prince <N8TM@aol.com> writes:

 Tim> If you're using f90, you won't be getting the same version of
 Tim> date_and_time which you get with g77 alone.  I bear some
 Tim> responsibility for the latter, having persuaded Dave Love to
 Tim> include it in libU77.  The g77 version, of course, doesn't deal
 Tim> with key words, but if it has other problems, I'll try to fix
 Tim> them.

Of course it doesn't have keywords with g77.  I'm afarid I don't
undertsand the problem.  What's wrong with it since I fixed the thinko
Kevin reported?  [FWIW, I didn't need persuading to include it, but it
was sidelined pending portability issues and probably still doesn't
provide millisecond data on many non-GNU systems where it might.]

Re: g77/F90 date_and_time() intrinsic

[N8TM]

If you're using f90, you won't be getting the same version of date_and_time
which you get with g77 alone.  I bear some responsibility for the latter,
having persuaded Dave Love to include it in libU77.  The g77 version, of
course, doesn't deal with key words, but if it has other problems, I'll try to
fix them.

Re: g77/F90 date_and_time() intrinsic

[Dave Love]

>>>>> "Tim" == Tim Prince <N8TM@aol.com> writes:

 Tim> If you're using f90, you won't be getting the same version of
 Tim> date_and_time which you get with g77 alone.  I bear some
 Tim> responsibility for the latter, having persuaded Dave Love to
 Tim> include it in libU77.  The g77 version, of course, doesn't deal
 Tim> with key words, but if it has other problems, I'll try to fix
 Tim> them.

Of course it doesn't have keywords with g77.  I'm afarid I don't
undertsand the problem.  What's wrong with it since I fixed the thinko
Kevin reported?  [FWIW, I didn't need persuading to include it, but it
was sidelined pending portability issues and probably still doesn't
provide millisecond data on many non-GNU systems where it might.]

g77/F90 date_and_time() intrinsic

[Kevin Maguire]

Hi

Consider the following: (egcs-1.1.1 Linux/x86)

% cat egcs-timer.f
       character*8 dt
       character*10 tm
       character*5 zn
       integer tt(8)
       CALL Date_and_Time(dt,tm,zn,tt)
       print *,dt
       print *,tm
       print *,zn
       print *,tt
       end
% g77 egcs-timer.f
% ./a.out 
 19990115
 111938.953
 +0000
 1999 1 15 0 11 19 38 953109
                      ^^^^^^
% ./a.out
 19990115
 112017.207
 +0000
 1999 1 15 0 11 20 17 207858
                      ^^^^^^

The F90 standard says something like:

VALUE(8) : the milliseconds, in the range 0 to 999, or -HUGE(0) if there
           is no clock.

There is a clock, and I see in datetime_.c

  vals[7] = 0;                  /* no STDC way to get this */
  /* GNUish way; maybe use `ftime' on other systems. */
#if HAVE_GETTIMEOFDAY
  {
    struct timeval tp;
    struct timezone tzp;
    if (! gettimeofday (&tp, &tzp))
      vals[7] = tp.tv_usec;
  }
#endif

My man page says for gettimeofday:

       int gettimeofday(struct timeval *tv, struct timezone *tz);
       #include <sys/time.h>
       #include <unistd.h>

       gettimeofday  and  settimeofday  can set the time as well as a timezone.
       tv is a timeval struct, as specified  in /usr/include/sys/time.h:

       struct timeval {
                      long tv_sec;        /* seconds */
                      long tv_usec;  /* microseconds */
                      };

So don't we need to convert this to milliseconds to satisfy the
standard.

Incidentally the egcs/g77 info doc says the same thing:

---
VALUES
     The year, month of the year, day of the month, time difference in
     minutes from UTC, hour of the day, minutes of the hour and
     milliseconds of the second in successive values of the array.

   On systems where a millisecond timer isn't available, the millisecond
value is returned as zero.
---

Cheers
Kevin
-- 
   /\___/\    ==============================================
   | o o |    Kevin Maguire          Computational Scientist
  __\_^_/__   DCI                  CLRC Daresbury Laboratory
 (__/   \__)  e-mail:                     K.Maguire@dl.ac.uk
  _|  :  |_   Tel: 01925 603221            Fax: 01925 603634
 (__\___/__)  ==============================================