dots in charm names (shelr.tv)

[Gustavo Niemeyer]

On Tue, Mar 26, 2013 at 4:01 PM, Kapil Thangavelu
<kapil.thangavelu at canonical.com> wrote:
> On Tue, Mar 26, 2013 at 11:44 AM, Gustavo Niemeyer
> <gustavo.niemeyer at canonical.com> wrote:
>> No, the best way to ensure this divergence doesn't happen is to check
>> the store to see if a charm exists. This is very trivial to do, and
>> very effective too. Anything else will open the door for headaches.
>
>> The charm browser does do this check and stores it against the charm
> although its not currently limiting visibility on this alone. It should do
> that though (bug 1160527).

Thanks, it should really do this check. It has already caused multiple
problems in the past where people get confused by the divergence
between both views of the world.

> The issue is that we have no feedback mechanism
> for users from the store, so at times it can be opaque as to what the actual
> issue is with the appearance in the charm store, unless that rule is also

Agreed, this is an issue, and it's at the top of my TODO. This is
orthogonal to the problem we're discussing, though. I'll solve it by
introducing a proper "juju publish" command.

> encoded into the charm-lint tool, or into the review queue/lint reports on
> the browser. The mechanism i suggested was to avoid this divergence of rules
> by incorporating the store processing directly into the feedback mechanisms
> we have for charm authors.

That's definitely a good idea, +1. That said, it doesn't solve the
original problem, because there are again two different places that
check to see if a charm is valid or not. We need a single
authoritative place being verified when we want to tell if a charm
exists, and that place exists and is trivial to query.


gustavo @ http://niemeyer.net