On Thu, Oct 15, 2015 at 9:44 AM, Pádraig Brady <[email protected]> wrote: > On 15/10/15 17:18, Pádraig Brady wrote: >> On 15/10/15 16:13, Jim Meyering wrote: >>> Here's a small improvement: >> >>> # Strip that part off for the following comparison. >>> clean_rm_err_() >>> { >>> - sed "s/.*rm: /rm: /; \ >>> - s/\(rm: it is dangerous to operate recursively on\).*$/\1 '\/'/" >>> + sed 's,.*\(rm: it is dangerous to operate recursively on\).*$,\1 '"'/'," >>> } > > Actually I'm not sure the combined works as we need > to match against multiple lines starting with "rm:..." > See "exp" below in the script.
Good points. My primary reaction was to the use of double quotes and "/" as the delimiter. Adjusting for that yields this: - sed "s/.*rm: /rm: /; \ - s/\(rm: it is dangerous to operate recursively on\).*$/\1 '\/'/" + sed 's/.*rm: /rm: / + s,\(rm: it is dangerous to operate recursively on\).*$,\1 '"'/'," Worth it?
