Re: Need a small VMS test

Michael G Schwern wrote:
> John E. Malmberg wrote:
>> Default:
>>
> Thanks, that looks all correct.  If you could try the latest version of the
> code that would be great.  It makes it more accurate for systems which have
> silly failure points, like Y10K.
> http://code.google.com/p/y2038/source/browse/trunk/bin/check_max.c

EAGLE> cc check_max
EAGLE> link check_max
EAGLE> run check_max
               gmtime max 4294967295
            localtime max 4294967295
               gmtime min 0
            localtime min 0

EAGLE> cc check_max/define=__SIGNED_INT_TIME_T
EAGLE> link check_max
EAGLE> run check_max
               gmtime max 2147483647
            localtime max 2147483647
               gmtime min 3221225472
            localtime min 3221225472

> 
>> VMS keeps time internally as a 64 bit signed integer all platforms,
>> usually in local time.  The base time is November 17, 1958
> 
> What do localtime() and gmtime() think time_t = 0 is?
> 
> #include <time.h>
> #include <stdio.h>
> 
> int main(void)
> {
>     struct tm *date;
>     time_t time = 0;
>     date = gmtime(&time);
> 
>     printf("# %04d.%02d.%02d %02d:%02d:%02d\n",
>         date->tm_year + 1900,
>         date->tm_mon  + 1,
>         date->tm_mday,
> 
>         date->tm_hour,
>         date->tm_min,
>         date->tm_sec
>     );
> 
>     return 0;
> }

EAGLE> cc check_time_0.c
EAGLE> link check_time_0
EAGLE> run check_time_0
# 1970.01.01 00:00:00

Substituting localtime for gmtime gives this for my timezone.
# 1969.12.31 18:00:00

For VMS, most of the routines that generate a time_t value are already 
handled by wrapper routines.  It is possible that at some time in the 
future, even if the C library does not support it, Perl on VMS could 
support a 64 bit time_t type.

-John
[email protected]
Personal Opinion Only