Shouldly VS Verify

On a side note, something I would highly recommend NOT doing is using the built in assertion types for any of the test adapters. Without a doubt the hardest part of switching unit test frameworks is having to fix all your assertions which is why we use 3rd party assertions. The built-in assertions also tend to not be very feature rich and don't have the most helpful messages. We personally use FluentAssertions, but there are other options such as Shoudly or Should. I highly recommend picking one of them over the built in assertions. You will thank yourself later :)

Shouldly - More "fluent" way of writing assertions that I tend to like more personally. Example:

I use NUnit and Shoudly.

Robbie Rogers came out in the MLS like a decade ago

Bogus - For creating fake data Verify - Snapshot testing for .NET MELT - For testing ILogger usage Stryker - Mutation Testing for .NET TestContainers - run docker programmatically in integration tests

Are you familiar with "snapshot testing" tools such as Verify that store expected output in files. It's still unit testing.

We can use Verify to perform snapshot testing and check for the correct output of the program. In order to make things easier and simplify working with process output capturing and invocation, I used CliWrap.

You will have to write extra code to set up a code base to get it started (there's always a large initial cost in getting things set up, and you'll be writing code that helps get the state of your application setup), but I can assure you that our team paid the initial tax and the only reason our tests change now is because of requirements changes (and maybe sometimes because the testing tools we use like Verify have some breaking changes in behavior when we upgrade). Otherwise, it helps us identify issues in our code, particularly when we do library upgrades or change to a different library. Again, our tests do not change when we completely reimplement anything, just when the external contract changes. We just get to refactor/reimplement and have confidence that the old behavior stays the same. And then you get to hook up a benchmark to your tests and, if your reason for refactoring was performance reasons, you can show that it was effective.

I assume this means you can take something like this, combine it with Snapshot/Approval testing (link to a library I have used), and then you have some quick-to-generate tests that help guard against regressions (even visual ones) by say: