tests: Don't use #n in sed scripts.

For whatever reason at least some versions of the MinGW/MSYS port of GNU
sed doesn't seem to handle scripts staring with #n.  No big deal, we can
just use the -n option explicitly instead.

diff --git a/tests/general.at b/tests/general.at
index c40d6b511e56b13008220ff92165eb18362d5232..a77134f04310086a04159df468bd6ffeeb1fa1ec 100644 (file)
--- a/tests/general.at
+++ b/tests/general.at
@@ -70,15 +70,14 @@ explain int
 
 # Program to generate expected output based on src/execute.gperf
 AT_DATA([expout.sed],
-[[#n
-/^%%$/,/^%%$/ {
+[[/^%%$/,/^%%$/ {
   s/^exit.*//
   s/,.*//p
 }
 $a\
 type int
 ]])
-sed -f expout.sed "$srcdir/src/execute.gperf" >expout
+sed -n -f expout.sed "$srcdir/src/execute.gperf" >expout
 
 # Program to filter the help output to extract the command list from "help"
 # without any descriptions.