gcc.dg/cpp/_Pragma3.c seems broken...

[David Daney]

With svn r117549 bootstrapped on mipsel-none-linux-gnu:

The test gcc.dg/cpp/_Pragma3.c basically checks whether _Pragma3.c is 
not younger than the file mi1c.h in the same directory.  The test fails 
with excess errors if this is not the case:

.
.
.
Executing on host: /home/build/gcc-build/gcc/xgcc 
-B/home/build/gcc-build/gcc/ 
/home/build/gcc/gcc/testsuite/gcc.dg/cpp/_Pragma3.c    -ansi 
-pedantic-errors -fno-show-column -E  -o _Pragma3.i    (timeout = 300)
/home/build/gcc/gcc/testsuite/gcc.dg/cpp/_Pragma3.c:11: warning: current 
file is older than mi1c.h
output is:
/home/build/gcc/gcc/testsuite/gcc.dg/cpp/_Pragma3.c:11: warning: current 
file is older than mi1c.h

FAIL: gcc.dg/cpp/_Pragma3.c (test for excess errors)
Excess errors:
/home/build/gcc/gcc/testsuite/gcc.dg/cpp/_Pragma3.c:11: warning: current 
file is older than mi1c.h
.
.
.


After doing a svn checkout I have:

$ ls --full-time _Pragma3.c mi1c.h
-rw-r--r--  1 root root 309 2006-10-06 23:51:13.000000000 -0700 _Pragma3.c
-rw-r--r--  1 root root 214 2006-10-06 23:51:14.000000000 -0700 mi1c.h


Am I missing something here, or is the success of this test really based 
on the ability of svn to get both of these files within the same second?

If so that seems bad.

David Daney