On Tue, Apr 16, 2013 at 06:26:38AM -0600, Eric Blake wrote:
> On 04/16/2013 03:48 AM, Kevin Wolf wrote:
> > $QEMU_PROG happens to be 'qemu' in my setup, so this sed command
> > replaces a bit too much. Restrict it to the start of the line and to
> > when it's followed by a colon, i.e. the form used by error messages.
> >
> > Signed-off-by: Kevin Wolf <kw...@redhat.com>
> > ---
> > tests/qemu-iotests/common.filter | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/tests/qemu-iotests/common.filter
> > b/tests/qemu-iotests/common.filter
> > index bc5f250..a7f889a 100644
> > --- a/tests/qemu-iotests/common.filter
> > +++ b/tests/qemu-iotests/common.filter
> > @@ -155,7 +155,7 @@ _filter_qemu_io()
> > # replace occurrences of QEMU_PROG with "qemu"
> > _filter_qemu()
> > {
> > - sed -e "s#$(basename $QEMU_PROG)#QEMU_PROG#g"
> > + sed -e "s#^$(basename $QEMU_PROG):#QEMU_PROG:#g"
>
> Why spawn a basename process, when you can use shell to do the same?
> Also, the g modifier is worthless once you have a ^ anchor, since sed
> can only replace one ^ per line. And you don't need sed's -e option for
> a single script.
sed -e is the convention in qemu-iotests.
Dropping the 'g' would be nice.
The problem with the POSIX shell string replacement is that the syntax
is horrible. I can never remember what ${%}, ${%%}, ${#} and
%{##} do. $(basename $QEMU_PROG) is clear (although it doesn't handle
spaces in the filename!).
Reviewed-by: Stefan Hajnoczi <stefa...@redhat.com>
