How can mechanism error handling become more robust?
Here are some things to get the discussion started.
- The crowdmatch tests should implement a variant of 'dummyStripe' that returns failures for everything. This will mimic getting failure messages from Stripe. Then tests can use it to record/define how we handle errors.
- Some methods need to return error conditions, rather than handling them directly.
- When something goes wrong, how do we show it to the users? Let's have a discussion of all the things that could go wrong, and make decisions on them (priority on getting something workable versus having complete and final answers)
- Some errors can happen before stripe gets called, like if a user's payment info somehow disappears before we do a payment event.