Currently the test framework immediately returns results for http calls, that's not how things will happen in production.
Having a configurable delay in the fixtures would help, a default of 100ms would probably be OK?
It would help indicate where too many http calls are being made for production (even though they're snappy in the tests).
I don't want to add 1/10 or 2/10 of a second to lots of tests in order to
catch this. Maybe the dummy API should raise an error if too many HTTP
calls are made?
Currently the test framework immediately returns results for http calls, that's not how things will happen in production.
Having a configurable delay in the fixtures would help, a default of 100ms would probably be OK? It would help indicate where too many http calls are being made for production (even though they're snappy in the tests).