diff options
author | Jim Meyering <jim@meyering.net> | 2004-02-08 10:00:38 +0000 |
---|---|---|
committer | Jim Meyering <jim@meyering.net> | 2004-02-08 10:00:38 +0000 |
commit | 400a1fc781b11dc00c47880138f43052b006c3da (patch) | |
tree | 30077fe30e6f7b645e7e96d4473ca434490eca0d | |
parent | 8c756b5ed6f71101f34565d25b5be4bf7c85ff26 (diff) | |
download | coreutils-400a1fc781b11dc00c47880138f43052b006c3da.tar.xz |
Avoid triggering a bug in OSF/Tru64's sed
that would cause an unwarranted test failure.
-rwxr-xr-x | tests/rm/rm3 | 10 | ||||
-rwxr-xr-x | tests/rm/rm5 | 10 |
2 files changed, 14 insertions, 6 deletions
diff --git a/tests/rm/rm3 b/tests/rm/rm3 index 018c4ae66..98b7ff9bf 100755 --- a/tests/rm/rm3 +++ b/tests/rm/rm3 @@ -49,9 +49,13 @@ EOF # Both of these should fail. rm -ir z < in > out 2>&1 || fail=1 -# The following loses with Tru64's sed, since that sed mistakenly -# tacks a newline onto the end of the file. -tr '?' '\n' < out |sed 's/^ //' |sort > o2 + +# Given input like `rm: ...? rm: ...? ' (no trailing newline), +# the `head...' part of the pipeline below removes the trailing space, so +# that sed doesn't have to deal with a line lacking a terminating newline. +# This avoids a bug whereby some vendor-provided (Tru64) versions of sed +# would mistakenly tack a newline onto the end of the output. +tr '?' '\n' < out | head --bytes=-1 | sed 's/^ //' |sort > o2 mv o2 out sort <<EOF > exp || fail=1 diff --git a/tests/rm/rm5 b/tests/rm/rm5 index 91715ae75..a1f3231af 100755 --- a/tests/rm/rm5 +++ b/tests/rm/rm5 @@ -39,9 +39,13 @@ fi fail=0 rm -ir d < in > out 2>&1 || fail=1 -# The following loses with Tru64's sed, since that sed mistakenly -# tacks a newline onto the end of the file. -tr '?' '\n' < out |sed 's/^ //' > o2 + +# Given input like `rm: ...? rm: ...? ' (no trailing newline), +# the `head...' part of the pipeline below removes the trailing space, so +# that sed doesn't have to deal with a line lacking a terminating newline. +# This avoids a bug whereby some vendor-provided (Tru64) versions of sed +# would mistakenly tack a newline onto the end of the output. +tr '?' '\n' < out | head --bytes=-1 | sed 's/^ //' > o2 mv o2 out # Make sure it's been removed. |