[Cosmo-dev] 0.7 server dogfooding

[Katie Capps Parlante]

One of the risks that Sheila brought up in a conversation yesterday 
(which she didn't get a chance to write up) was about dogfooding 0.7 
server dashboard features (and roundtrip scenarios between server and 
desktop).

I think she and Mimi have surfaced a lot of harder-to-find-but-important 
bugs from day to day use collaborating on a ppd collection over a longer 
period of time. (That's what dogfooding is all about, of course). The QA 
team is doing this as well.

They've been doing this using desktop checkpoints and the production 
hub, so we've been getting mileage on desktop features, EIM, and 
everything server up to 0.6.1.1. While we do have qacosmo for doing 
testing the 0.7 server, we don't get a chance to do the same dogfood 
mileage on it for finding those detailed dashboard bugs.

Is it worth doing something to mitigate this risk?

- iirc, the qacosmo doesn't always need to be wiped from update to 
update. At some point should we have some people trying to "dogfood" 0.7 
qacosmo? (of course, the data would be lost eventually so its not 
exactly dogfooding, but at least try testing scenarios that span a few 
days/sessions)

- have a protocol of reloading "real" data each time qacosmo is wiped? 
(office calendar, some others)

Any other ideas?

Cheers,
Katie