On Wed, Oct 22, 2008 at 09:07:18PM -0700, Danek Duvall wrote:

> > Maybe I'm just asking for a few RFEs (smarter column selection, inverse
> > dependency listing, help that enumerates possible -t and -o values),
> > though.
> 
> Those are all good RFEs (search actually covers the reverse dependencies,
> but not at all intuitively), but some ideas on how they might work are
> greatly appreciated.

I'm afraid I haven't thought hard enough about it to be able to help
much. I can throw out random opinions with the best of them though.

> Given that -o values can be absolutely anything, what
> should we suggest?  -t is much more constrained, but at least those are
> completely documented in the man page, and so verbose help seems less
> useful.

I disagree - one the nicest things about zfs(1) is that you almost never
need the man page. Interactive help is so much easier.

> As for smarter column listing, what would you suggest?  Perhaps we could
> have a default column listing for each action type -- "-t depend" could
> imply "-o fmri"

Yes, please - each -t option should have a default set.

> -t link,depend?

Default to -m if more than one -t is specified  with non-matching
default sets, seems like the only sensible way to display disparate
objects. The current output looks funny anyway.

>  Should the default output (no -t option) have all types or just
>  actions with paths?

I think the current choice is the right one.

> I'm thinking that perhaps we should have a -l (long) option akin to ls that
> gives you a multicolumn output by default, giving similar output to ls for
> filesystem objects, and both fmri and type for dependencies, etc.
> 
> And perhaps a -a option that tells it to list all types?  Maybe prefixing
> each line with the action type?

Perhaps -m is really enough.

> What about if -t isn't specified, but -o is, then only actions with
> attributes named in -o are printed -- so you can do pkg contents -o path.

That would be a nice enhancement.

I do think that dependencies are worthy of special treatment, however,
to the extent of deserving pkg depends and pkg depends-on.

regards
john
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to