Default user feedback

[William Reade]

On Tue, 2013-04-02 at 15:57 -0300, Gustavo Niemeyer wrote:
> On Tue, Apr 2, 2013 at 3:49 PM, William Reade
> <william.reade at canonical.com> wrote:
> > I think there are two distinct issues here. First off, on reflection, I
> > don't think we should need to put the package prefix on log messages in
> > "main" packages at all; all that does is clutter things up. But I don't
> > think we should change the general policy of including package context
> > in other log messages.
> 
> I'd do a quick data acquisition before trusting that feeling. How many
> of the log messages are ambiguous, in the sense that a quick "grep"
> can't spot its location, when in doubt?

That's not quite the use case I have in mind. I already have a pretty good
picture of the code in my head, but not to the extent that I can identify
the source of log messages by eye. Seeing the package context allows me to
build a coherent picture of what's going on in a log file *without* needing
to check the source.

> Gut feeling is that this catches about 100% of the cases, which means
> dumping that information is hurting the user experience for everybody
> in exchange for a cost that is very low on the developer side. That's
> an opinion, though, obviously.

FWIW, my opinion is that feedback and logging are very different contexts.
The formatting in John's first example is great as feedback but bad as
logging; in the second, vice versa. 

So, to dial it back a little: I propose that I add a "tweak logging" card:

1) Drop package badging from log calls in "main" packages
2) Drop the JUJU:... badging across the board

...which I think/hope everyone can agree moves us in a positive direction,
even if it doesn't perfectly match everyone's desires, and which is small
enough that someone might be able to pick it up in a spare half hour at EOD,
or something. Would anyone object to this?

Cheers
William

PS I still don't guarantee that it'll be done. But without a card, I can
almost guarantee that it won't ;).