Food for thought: automatic testing based on specified routing information

Routing systems often allow users to specify various parameters, paths, etc... that will cause a request to be routed to a given handler. Using this information, tests can be synthesized without additional user input:

* validation tests that use the specification as a basis for mutation (fuzzing)
* generation of sequences of calls to different endpoints using parameters that match the specification, remembering headers set by the response along the way etc...
* running through sequences of HTTP commands, but repeating certain commands without taking the responses into account etc... quickly finding heisenbugs relating to non-idempotent APIs that generally plague modern webdev

None of the above things require any additional user information to find crashing bugs relatively efficiently. If we additionally support things like pre/post conditions that check that an API behaved according to some given property, we can dramatically reduce the effort required to write tests of a wide range of complexity levels.

If a web framework finds tons of your bugs without you spending time writing tests, or lets you write low-bias model tests without writing hundreds of lines of boilerplate, that's a nice incentive to choose the framework over others.

We should build tools that make it easy to know when we have achieved success using them.