[gfortran] Fix date_and_time intrinsic procedure

[Steve Kargl <sgk at troutmask dot apl dot washington dot edu>]

The attached patch fixes the date_and_time procedure for
two case.

(1) If a clock is not available the standard requires that
    date_and_time() return -HUGE(0) for all elements in
    VALUES(1:8).   We currently return HUGE(0).

(2) If a clock is available, but it lacks millisecond
    (or better) resolution we were setting VALUE(8) to
    HUGE(0).  All other VALUES() are correctly, set so
    we are return 2**31 milliseconds, which is out of
    range.  A much better choice is 0.

2005-01-24  Steven G. Kargl  <kargls@comcast.net>

      *  intrinsic/date_and_time.c: Fix conformance problems.

-- 
Steve

Attachment: date_and_time.c.diff
Description: Text document