STR: * log in to the developer hub * go to Payment Keys, https://marketplace.firefox.com/developers/in-app-keys/ * press Generate Keys * note that the page says Application Key (Not yet generated.) Screenshot: http://cl.ly/image/2m362y0p1w33 This prevents developers from setting up in-app payments.
Now I get: http://sentry.mktmon.services.phx1.mozilla.com/mkt/marketplacefirefoxcom/group/1490/ mkt.developers.views_payments in in_app_keys MultipleObjectsReturned: get() returned more than one UserInappKey
I just tried this and I was able to see the generated keys if I refreshed the page (I didn't try and generate a second time). Seems the UI should be refreshing and it isn't for some reason. Also it clearly shouldn't be possible to generate more than one if that's not expected.
I think its this commit: https://github.com/mozilla/zamboni/commit/3f44e85e0fbfa8eac1dcb505cba0eb9841bedbfe#diff-5b4589fe747c55cf59dc46e9700fefa6L252 Reverting that and it worked fine for me locally.
> This prevents developers from setting up in-app payments. Actually that's not true, real in-app payment keys are stored in solitude. This is stopping developers from testing in-app payments.
The `uncached` manager was removed because it was confusing to have 2 different ways to disable caching, and also caused bugs because it wasn't following the magic `objects` does, at least on the `Addon` class (it wasn't filtering on addon type correctly, and wasn't filtering out soft-deleted addons) `no_cache()` should just work. Can we come up with a unit test exhibiting the issue ? Worth noting, we are trying to upgrade django-cache-machine in bug 883477 (there is a PR up) and that might fix some weird issues.
Took a look at the problem, it turns out it was due to a bug in django-cache-machine 0.6 : it caches count()s even if you call no_cache(). So if you didn't wait 60 seconds you'd get the cached result (which was "no key") bug 883477 is fixed, so we should now be using django-cache-machine 0.8 on -dev, which should fix this issue. Can you reproduce it now on -dev ?
Works for me on -dev, thanks Mathieu.