On Thu, 22.07.10 18:16, Stephen John Smoogen (smooge@xxxxxxxxx) wrote: > > On Thu, Jul 22, 2010 at 18:06, Chris Adams <cmadams@xxxxxxxxxx> wrote: > > Once upon a time, Lennart Poettering <mzerqung@xxxxxxxxxxx> said: > >> Same with systemd. If you use "systemctl status foo.service" the output > >> is human readable. If it is "systemctl show foo.service" it is computer > >> parsable. Just a slightly different command of the systemctl tool. > > > > Again: this is confusing! There should be one (and only one) command to > > show information. It should accept arguments to modify that output, > > e.g. default to brief info, -v gets a little more info, -vv gets all > > kinds of info, -p to get "parseable" output (or -f for "formatted"), > > etc. > > > > Having "status" and "show" give the same info in different formats will > > always be confusing. People won't remember which is which (because the > > works mean similar things in this context) and will run the wrong one > > for what they want about 50% of the time (which will just be > > frustrating). > > > > To reword this into a slightly less cranky more useful comment: > > My usual dealing with this is that you want one command and an > optional flag to change output. > > systemctl status # awk readable > systemctl status --human # human readable > > The general rule to remember is that the person you are going to deal > with is going to be tired, grumpy, and pissed off. Commands need to be > simple to remember and easy to script first. Note again that the two commands show different data. "status" actually shows only what the name suggests, i.e. the runtime status. OTOH "show" allows you to introspect the unit in much more detailed ways, allows you to query and show every tiny property. For example you can use it to query the PID file of a unit, or the RLIMIT_NOFILE resource limit configured for the unit, or the capability bounding set, or whether the unit was started with a private, poly-instantiated /tmp, or, or, or. Anyway, it would be really cool if could discuss this when everybody actually had a look on the tool, and played around with this. Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel
