This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
[patch, fortran, testsuite] Replace secnds*.f 20-ms tolerance with bracketing
- From: Brooks Moses <brooks dot moses at codesourcery dot com>
- To: gcc-patches at gcc dot gnu dot org
- Cc: fortran at gcc dot gnu dot org
- Date: Fri, 09 Feb 2007 16:52:28 -0800
- Subject: [patch, fortran, testsuite] Replace secnds*.f 20-ms tolerance with bracketing
On my computer, on most regression checks, I usually get a random
failure in one of the secnds*.f test cases. The problem is that a pass
in the testcase requires that subsequent calls to secnds() and
date_and_time() return the same time to within a 20ms tolerance, and on
slower systems this arbitrary tolerance value is too small.
Rather than just increasing the tolerance, I replaced it with a bracket;
now, this calls secnds() before and after the date_and_time() call in
order to establish a range, and check that the result of date_and_time()
is within that range. This way, we know that any failures are genuine
out-of-order errors, rather than just out-of-tolerance glitches.
---------------------------------------------------------------
2006-02-09 Brooks Moses <brooks.moses@codesourcery.com>
* gfortran.dg/secnds.f: Replace time tolerance with a
before-and-after bracket.
* gfortran.dg/secnds-1.f: Likewise.
---------------------------------------------------------------
Tested on i686-pc-linux-gnu. Ok for trunk?
- Brooks
Index: gcc/testsuite/gfortran.dg/secnds-1.f
===================================================================
--- gcc/testsuite/gfortran.dg/secnds-1.f (revision 121705)
+++ gcc/testsuite/gfortran.dg/secnds-1.f (working copy)
@@ -1,24 +1,23 @@
C { dg-do run }
+C { dg-options "-ffloat-store" }
C Tests fix for PR29099 - SECNDS intrinsic wrong result with no delay.
-C Note1: The test uses +/-20ms accuracy in the check that
-C date_and_time and secnds give the same values.
C
C Contributed by Paul Thomas <pault@gcc.gnu.org>
C
character*20 dum1, dum2, dum3
- real t1, t2
+ real t1, t1a, t2, t2a
real dat1, dat2
- real dt
integer i, j, values(8)
- dt = 40e-3
t1 = secnds (0.0)
call date_and_time (dum1, dum2, dum3, values)
+ t1a = secnds (0.0)
dat1 = 0.001*real (values(8)) + real (values(7)) +
& 60.0*real (values(6)) + 3600.0* real (values(5))
- if (int ((dat1 - t1 + dt * 0.5) / dt) .ne. 0) call abort ()
+ if (((dat1 - t1) < 0.) .or. ((dat1 - t1) > (t1a - t1))) call abort ()
+ t2a = secnds (t1a)
call date_and_time (dum1, dum2, dum3, values)
+ t2 = secnds (t1)
dat2 = 0.001*real (values(8)) + real (values(7)) +
& 60.0*real (values(6)) + 3600.0* real (values(5))
- t2 = secnds (t1)
- if (int ((dat1-dat2 + t2 + dt * 0.5) / dt) .ne. 0.0) call abort ()
+ if (((dat2 - dat1) < t2a) .or. ((dat2 - dat1) > t2)) call abort ()
end
Index: gcc/testsuite/gfortran.dg/secnds.f
===================================================================
--- gcc/testsuite/gfortran.dg/secnds.f (revision 121705)
+++ gcc/testsuite/gfortran.dg/secnds.f (working copy)
@@ -1,29 +1,27 @@
C { dg-do run }
-C { dg-options "-O0" }
+C { dg-options "-O0 -ffloat-store" }
C Tests fix for PR14994 - SECNDS intrinsic not supported.
-C Note1: The test uses +/-20ms accuracy in the check that
-C date_and_time and secnds give the same values.
C
C Contributed by Paul Thomas <pault@gcc.gnu.org>
C
character*20 dum1, dum2, dum3
- real t1, t2
+ real t1, t1a, t2, t2a
real dat1, dat2
- real dt
integer i, j, values(8)
- dt = 40e-3
t1 = secnds (0.0)
call date_and_time (dum1, dum2, dum3, values)
+ t1a = secnds (0.0)
dat1 = 0.001*real (values(8)) + real (values(7)) +
& 60.0*real (values(6)) + 3600.0* real (values(5))
- if (int ((dat1 - t1 + dt * 0.5) / dt) .ne. 0) call abort ()
+ if (((dat1 - t1) < 0.) .or. ((dat1 - t1) > (t1a - t1))) call abort ()
do j=1,10000
do i=1,10000
end do
end do
+ t2a = secnds (t1a)
call date_and_time (dum1, dum2, dum3, values)
+ t2 = secnds (t1)
dat2 = 0.001*real (values(8)) + real (values(7)) +
& 60.0*real (values(6)) + 3600.0* real (values(5))
- t2 = secnds (t1)
- if (int ((dat1-dat2 + t2 + dt * 0.5) / dt) .ne. 0.0) call abort ()
+ if (((dat2 - dat1) < t2a) .or. ((dat2 - dat1) > t2)) call abort ()
end