Peter Memishian wrote: > > >Consider the output for show-linkprop -c. There are quotes around all > > >values. While this provides consistency among the format of all possible > > >values of all possible properties (as some values may contain spaces), > > >it's not obvious to parse. At least not for me. How do I easily strip > > >the " characters from around a value in a shell script? > > > > > >The context is that I use the same syntax for "show-iptun -p", and I have > > >a need to get access to all of the IP tunnel names. The output of dladm > > >show-iptun -p, for example, is > > > > Given that -p is being used for parseable output here, I'd like to > > suggest that ipmpstat be changed and the roles of -p/-P reversed > > so that -p has the same behaviour for both. > >Sorry, I'm afraid of you've lost me. However: > > 1. ipmpstat -p outputs probe information. We could change that > to -P, but that would create greater usability issues because > the other four output modes are -a -i -g and -t (address > interface, group, and target), all use lowercase. We could > change them all to be uppercase, but that would mean every > use of ipmpstat would need an extra keypress (the shift key) > which seems obnoxious. >
If ZFS and SMF can integrate and introduce new commands that have "long option names" (disable/enable/sharenfs/etc) then I'd be tempted to think that the extra keypress argument holds very little water when designing the CLI today. A well thought out design for new layer 2 tools would ensure that -p for both dladm and ipmpstat meant the same thing. You might want to take another look at the way you've designed the CLI for ipmpstat and think about a couple of things: - does it need to have long option names like dladm, etc? or conversely, why doesn't it need to? - if there are mutually exclusive options (such as -a/-i/-g/-t/-p), should they be individual options or sub-options? (e.g -x probe) - how the behaviour of various command line options for ipmpstat can be made to have similar behaviour to the same letters elsewhere (less confusion.) btw, the presentation of -a/-i/-g/-t/-p in the document does not suggest the design you're alluding to here and I'm somewhat surprised that -t and -p are seperate - they appear to be different sides of the same coin (which might suggest -t/-T.) Darren
