On Tue, 2010-11-09 at 13:39 -0800, Danek Duvall wrote:
> Tim Foster wrote:
> I don't know that multiple independent ranges seem useful to me, but if
> you've done the work, you've done the work.
>
> It's not clear what "e" stands for here. "Event"? If so, I'd expect to
> pass it a unique event identifier. If this just accepts timestamps (and
> ranges), then -t seems more natural.
Yeah, it was originally -e, thinking that there was a 1:1 correspondence
between timestamps and history events. I'll change to -t.
> > > My question for Tim is whether I can ask for events using some subset of
> > > the time specification -- if I ask for HH:MM, will I get all events that
> > > took place (or at least started) during that hour and minute, or is it a
> > > parsing error?
> >
> > That's currently a parsing error.
>
> Okay. That can always be added later on, anyway.
Sure.
> > I can go back to the drawing board if you think this would be a useful
> > addition, or you think overloading -e to take both discreet events and
> > ranges of events is bad? (I quite like it, I have to admit)
>
> Single events and ranges both seem quite useful to me, though distinct
> ranges don't, and I think a way to represent a time with a shorter string,
> similarly to fmdump (including "the past X hours/minutes/fortnights"),
> would be nice, but like I said, that can be added later on.
I think with -t we could certainly extend the keywords and formats it
accepts (currently just a datestamp and "now") I'll make the small
change to swith '-e' to '-t', test and putback.
cheers,
tim
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
