+1 I think this is well considered.
Resurrecting -z for symmetry and perspicuity, and then documenting that "-z" is equivalent to "-t $'\0'" and thus conflicts with the presence of another -t (since multiple -t is disallowed) $.02, Malcolm (usually lurking) >-----Original Message----- >From: [email protected] >[mailto:[email protected]] On Behalf Of Pádraig >Brady >Sent: Thursday, January 08, 2015 10:24 PM >To: Assaf Gordon >Cc: Markus Elfring; [email protected] >Subject: Re: Splitting search results from a "find -print0" > >On 09/01/15 04:13, Assaf Gordon wrote: >> On Jan 8, 2015, at 21:16, Pádraig Brady <[email protected]> wrote: >>> I made a few adjustments, as seen inline below. >> >> Thank you for the clean-up and fixes. Looks much better now. >> >>> The main change was the removal of the -z option as that's supported with >>> -t '\0'. >> >> I humbly do think that the '-z' is nice, add some symmetry with the other >> utilities which support '-z' for NUL line-termination. I >understand "split" is not strictly a line-based utility (more like >'record-based'), but in the use-cases when the separator is relevant, it is >commonly used for lines. >> So if one combines it with other gnu programs (e.g. >> find/xargs/sed/grep/sort/join/uniq) - it is '-z' almost for all of them. >> But this is >nit-picking, of course, and does add some bloat/redundancy. > >maybe > >> If not '-z', perhaps it's worth adding an explicit mention of the "-t '\0'" >> method ? >> at least for other programs, the man-page clearly mentions the words 'NUL' >> and 'zero' - giving inexperienced user a hint about what >to do. > >definitely. How about: > > -t, --separator=SEP > use SEP instead of newline as record separator. > use -t '\0' to specify the NUL (zero) character. > >thanks, >Pádraig.
