Failfast and DataConnectors

[Rod Widdowson]

> It's not just fail-fast, but how errors in general in a connector are
> exposed or not. That has the impact of affecting the fail-fast situation
> though.
> 
> But yes, I'm suggesting mainly just that this option be configured
> consistently with one property name instead of having every connector
> define its own and hoping they match.

I think I follow most of that, but I suggest we add it to the Agenda for
Friday since there is no urgency.

R