https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91437

            Bug ID: 91437
           Summary: Problem with multi-line test outputs on
                    x86_64-w64-mingw32
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: testsuite
          Assignee: unassigned at gcc dot gnu.org
          Reporter: tkoenig at gcc dot gnu.org
  Target Milestone: ---

https://gcc.gnu.org/ml/gcc-testresults/2019-08/msg00909.html shows quite
a few bugs on x86_64-w64-mingw32. However, many of these appear to
be false positives.

Some analysis of the original data at

https://cloud.emrich-ebersheim.de/index.php/s/g9D245XdCW6GD5W

shows that some of these "failures" seem to be an artifact.
For example, looking at the file gfortran.log.gz under

https://cloud.emrich-ebersheim.de/index.php/s/g9D245XdCW6GD5W?path=%2F9.1.1-rev.274208%2Fgcc%2Ftestsuite%2Fgfortran

one finds

D:/opt/devel/gnu/src/gcc-mingw-w64/gcc-9.1.1/gcc/testsuite/gfortran.dg/do_subscript_1.f90:6:7:
Warning: (1)
D:/opt/devel/gnu/src/gcc-mingw-w64/gcc-9.1.1/gcc/testsuite/gfortran.dg/do_subscript_1.f90:5:46:
Warning: Array reference at (1) out of bounds (-1 < 1) in loop beginning at (2)
D:/opt/devel/gnu/src/gcc-mingw-w64/gcc-9.1.1/gcc/testsuite/gfortran.dg/do_subscript_1.f90:9:7:
Warning: (1)
D:/opt/devel/gnu/src/gcc-mingw-w64/gcc-9.1.1/gcc/testsuite/gfortran.dg/do_subscript_1.f90:8:46:
Warning: Array reference at (1) out of bounds (4 > 3) in loop beginning at (2)

[..]

PASS: gfortran.dg/do_subscript_1.f90   -O   (test for warnings, line 5)
FAIL: gfortran.dg/do_subscript_1.f90   -O   (test for warnings, line 6)
PASS: gfortran.dg/do_subscript_1.f90   -O   (test for warnings, line 8)
FAIL: gfortran.dg/do_subscript_1.f90   -O   (test for warnings, line 9)

[...]

D:/opt/devel/gnu/src/gcc-mingw-w64/gcc-9.1.1/gcc/testsuite/gfortran.dg/do_subscript_1.f90:6:7:
Warning: (1)
D:/opt/devel/gnu/src/gcc-mingw-w64/gcc-9.1.1/gcc/testsuite/gfortran.dg/do_subscript_1.f90:9:7:
Warning: (1)

[...]

so the problem appears to be with the regexp of the test, the output of
the compiler, or dejagnu.

I think if this kind of problem was fixed, the test tesults would look
_much_ cleaner.

Reply via email to