Hey. I repeatedly have people saying what they really want is to write isolated functional tests and then feel confident that the mocks they are using are correct. The way many people are solving this is by doing a record/playback approach, where they run their tests in staging on a weekly basis, record the results, and use these as their mocks. But I thought it would be great to say "use Pact to define your mocks, run your functional tests, and then use the Pact mocks to verify your provider." But in the docs you recommend not doing this for reasons I don't fully understand - something about the mocks defining a contract that is too constrictive and which breaks even when the contract has not broken. Can you explain more? I am very tempted to start using Pact as the mocks for functional tests, but not if this is just a non-starter use of Pact. Thanks!

Great question. The main issue is that Pact is not a general purpose mocking tool, so often times the mocks aren’t extensive enough to be used in functional tests. A good example is boundary testing - you might want to test a UI that accepts a user’s name with many combinations of short, long and unicode characters. From a contract POV, this is not useful additional information to encode and enforce a provider test for, but if you did it, it could make provider verification more cumbersome.

Makes sense, thanks!