On 08/28/2018 08:33 PM, Tim Graham wrote:
> Why is a new option needed? Why can't --keepdb be used for this use case?
Because that would be different behavior.
If you can convince me that using --keepdb is a solution, then I will
use your argument to say that --keepdb should be on always.
What I'm looking to offer users is to create and clean up objects in the
test DB, separate from create/drop test DB.
However, I want to do this in a way that doesn't weaken the existing
protection for "Oops, we're trying to use a test DB that already
exists... did you make a mistake? We don't want to trounce real data."
--
Curtis