On Tue, Sep 20, 2011 at 04:17:44PM +0100, Måns Rullgård wrote: > Måns Rullgård <[email protected]> writes: > > > Diego Biurrun <[email protected]> writes: > > > >> --- a/tests/fate.sh > >> +++ b/tests/fate.sh > >> @@ -21,7 +21,7 @@ test -d "$samples" || die "samples location not > >> specified" > >> > >> lock(){ > >> lock=$1/fate.lock > >> - (set -C; exec >$lock) 2>/dev/null || return > >> + (set -C; exec > $lock) 2> /dev/null || return > >> trap 'rm $lock' EXIT > >> } > >> > >> @@ -75,8 +75,8 @@ clean(){ > >> > >> report(){ > >> date=$(date -u +%Y%m%d%H%M%S) > >> - echo "fate:0:${date}:${slot}:${version}:$1:$2:${comment}" >report > >> - cat ${build}/config.fate ${build}/tests/data/fate/*.rep >>report > >> + echo "fate:0:${date}:${slot}:${version}:$1:$2:${comment}" > report > >> + cat ${build}/config.fate ${build}/tests/data/fate/*.rep >> report > >> test -n "$fate_recv" && $tar report *.log | gzip | $fate_recv > >> } > >> > >> @@ -99,14 +99,14 @@ test -d "$src" && update || checkout || die "Error > >> fetching source" > >> cd ${workdir} > >> > >> version=$(${src}/version.sh ${src}) > >> -test "$version" = "$(cat version-$slot 2>/dev/null)" && exit 0 > >> -echo ${version} >version-$slot > >> +test "$version" = "$(cat version-$slot 2> /dev/null)" && exit 0 > >> +echo ${version} > version-$slot > >> > >> rm -rf "${build}" *.log > >> mkdir -p ${build} > >> > >> -configure >configure.log 2>&1 || fail $? "error configuring" > >> -compile >compile.log 2>&1 || fail $? "error compiling" > >> -fate >test.log 2>&1 || fail $? "error testing" > >> +configure > configure.log 2>&1 || fail $? "error configuring" > >> +compile > compile.log 2>&1 || fail $? "error compiling" > >> +fate > test.log 2>&1 || fail $? "error testing" > >> report 0 success > >> clean > >> -- > > > > Rejected. I like it the way it is. > > To elaborate on why, it is consistent with >&1 and similar redirections > which may not have a space.
To also elaborate on why I proposed the patch in the first place: We use more generous spacing in configure (almost) consistently, except for 2>&1 and similar constructs, which I argue could be interpreted as a single operator. Diego _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
