On 11/09/2012 05:14 PM, Paul Eggert wrote:
On 11/09/2012 01:50 AM, Bernhard Voelker wrote:
What about simplifying the first s/... to eliminate all blanks?
Yes, could do that, if someone has the energy....
I also noticed that `sed -i` was used,
so I fixed that and made the simplificiation as suggested above, with...
commit 6bc77b3ae940b8afff101a1484e8f04d2cba1f56
Author: Pádraig Brady <[email protected]>
Date: Thu Nov 22 10:07:25 2012 +0000
tests: don't rely on sed -i being available
* tests/df/df-output.sh: sed --in-place is not generally available.
Also as suggested by Bernhard Voelker, simplify one of the
sed expressions a little, as removing all spaces suffices here.
diff --git a/tests/df/df-output.sh b/tests/df/df-output.sh
index f586334..0aee5f2 100644
--- a/tests/df/df-output.sh
+++ b/tests/df/df-output.sh
@@ -27,28 +27,28 @@ Try 'df --help' for more information.
EOF
df -i --output '.' 2>out && fail=1
-sed -i 's/ -i / OPT /' out
-compare exp out || fail=1
+sed 's/ -i / OPT /' out > out2
+compare exp out2 || fail=1
df --output -i '.' 2>out && fail=1
-sed -i 's/ -i / OPT /' out
-compare exp out || fail=1
+sed 's/ -i / OPT /' out > out2
+compare exp out2 || fail=1
df -P --output '.' 2>out && fail=1
-sed -i 's/ -P / OPT /' out
-compare exp out || fail=1
+sed 's/ -P / OPT /' out > out2
+compare exp out2 || fail=1
df --output -P '.' 2>out && fail=1
-sed -i 's/ -P / OPT /' out
-compare exp out || fail=1
+sed 's/ -P / OPT /' out > out2
+compare exp out2 || fail=1
df -T --output '.' 2>out && fail=1
-sed -i 's/ -T / OPT /' out
-compare exp out || fail=1
+sed 's/ -T / OPT /' out > out2
+compare exp out2 || fail=1
df --output -T '.' 2>out && fail=1
-sed -i 's/ -T / OPT /' out
-compare exp out || fail=1
+sed 's/ -T / OPT /' out > out2
+compare exp out2 || fail=1
# Ensure that each field is only used once for the --output argument.
cat <<\EOF > exp || framework_failure_
@@ -93,8 +93,7 @@ EOF
df -B1K --output=size '.' >out || fail=1
sed -e '1 {
- s/ [ ]*/ /g
- s/^ //
+ s/ //
q
}' out > out2
compare exp out2 || fail=1
