On 11/ 2/10 04:21 PM, Tim Foster wrote:
On Tue, 2010-11-02 at 15:46 -0700, Shawn Walker wrote:
On 10/31/10 09:20 PM, Tim Foster wrote:
http://cr.opensolaris.org/~timf/history-be-webrev/
...
src/client.py:
line 221: why not -d for date instead of -e? or are option values
sometimes not dates? I have no particular preference here, just
wondering what alternatives were considered.
I picked '-e' for 'event' or 'entry', though strictly speaking, yes, we
really mean timestamp or date. (I started out with -e, mistakenly
thinking there was a 1:1 correspondence between timestamp and event)
'-d' feels like it should be a debug flag, though perhaps '-t' would be
a better choice - fmdump(1) uses this, accepting a scary number of
different date formats (which I'd really like to avoid adding!)
I don't mind changing the flag I'm currently using to -t or any better
suggestions?
I'm indifferent. I'll defer to Danek on this point.
During the course of this work, I was half-wondering whether assigning a
uuid to each history event would be useful, allowing users to select
exactly one history event from the command line if they needed it:
ultimately that felt a bit like overkill, though it would have made
writing the tests a bit easier.
I think timestamp will ultimately work out ok; it's pretty rare that you
have lots of events happening in the same second.
...
Otherwise, this looks very thorough. Do you have any sample output?
Sure - I would you like some added to the man page? Otherwise, there's
Nah, that's not necessary, and I think it's just a maintenance burden.
some below. I'll update the webrev once we get consensus on the -e
flag name.
Two random things:
* as long as long output still shows reason by default, I can live with
the default output not showing it
* verify the client's behaviour if you remove the /var/pkg/history
directory
-Shawn
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
