Great video! Thank you! Just a small tip: When creating a test list, start with happy path and end with edge/exceptional cases (not the other way around).
If there was authentication/authorization setup, would you put the mocks/overrides for the auth schemes in the factory class? And how would you pass different claims to it if you wanted to dynamically mock different claims. I'm not sure on the structure.i was thinking pass data to the base class, which calls a method on the factory class to override some services. I don't know if that's a clean way
Ideally, I wouldn't mock Authentication at all. I covered that in my course, for example. I'm using Keycloak for auth, and in functional tests I'm first getting the JWT before running the tests.
Oh okay that makes sense. Thanks. I read your article on resource based auth in CA. Interesting because I'd have usually put the auth logic in the api layer using policies of some sort So if you wanted to mock the authenticated user you would opt for mocking that during subcutaneous testing , basically sending a MediatR req?
Where does authentication / authorization get done? I am thinking the microservice does not have any security in it and then you would handle all that in a proxy like konghq or Apigee? This would allow you to do the testing without messy with security. Is this a correct statement?
Great vidéos . I use test collection in order to share test container and avoid to create to many database . And for each test I reset the DB with snapshot or respawn
Some call it behaviour tests, other call integration tests. I prefer the 2nd and it's the kind of tests I am writing for my webapps. I like it calling integration tests, because it tests the app from the webapi down to the database, the whole application logic in other words. For simple web apis or most of microservices it's enough, but for complex db logic I additionally write integration tests for my repositories
I am struggling to understand the difference between unit tests and functional tests , or are they the same ? I use to write unit tests but without docker, test-containers, web application factory. What is the benefit of this approach ?
They are not the same. Unit tests just test a simple unit - class/service/method. Functional tests actually execute the ASP.NET Core application and run the tests. Huge difference.
I handle authorization part with AuthenticationHandler by overriding HandleAuthenticateAsync to use claims, found that there is a pretty common way to handle authentication part, but couldn't find the way to provide different claims for each test. What will bi the proper way to do such a thing ?
Want to master Clean Architecture? Go here: bit.ly/3PupkOJ
Want to unlock Modular Monoliths? Go here: bit.ly/3SXlzSt
I was actually waiting for this video to drop. Thanks Milan.
Awesome, I hope it was helpful :)
Great video! Thank you! Just a small tip: When creating a test list, start with happy path and end with edge/exceptional cases (not the other way around).
Fair enough
I would start with failure paths and end with happy paths.
As long as you cover every scenario it doesn't really matter.
@@sevansoft it matters when you test drive your code
I love this kind of videos. Thanks Milan!!
You're very welcome!
If there was authentication/authorization setup, would you put the mocks/overrides for the auth schemes in the factory class?
And how would you pass different claims to it if you wanted to dynamically mock different claims. I'm not sure on the structure.i was thinking pass data to the base class, which calls a method on the factory class to override some services.
I don't know if that's a clean way
Ideally, I wouldn't mock Authentication at all. I covered that in my course, for example. I'm using Keycloak for auth, and in functional tests I'm first getting the JWT before running the tests.
Oh okay that makes sense. Thanks. I read your article on resource based auth in CA. Interesting because I'd have usually put the auth logic in the api layer using policies of some sort
So if you wanted to mock the authenticated user you would opt for mocking that during subcutaneous testing , basically sending a MediatR req?
Where does authentication / authorization get done? I am thinking the microservice does not have any security in it and then you would handle all that in a proxy like konghq or Apigee? This would allow you to do the testing without messy with security. Is this a correct statement?
No auth in this example. But we definitely want it in a real world scenario. I'll cover an example with Keycloak in the near future.
@@MilanJovanovicTech thanks
Great vidéos .
I use test collection in order to share test container and avoid to create to many database .
And for each test I reset the DB with snapshot or respawn
I just talked about that in a recent video I was recording, for running integration tests in CI
Some call it behaviour tests, other call integration tests. I prefer the 2nd and it's the kind of tests I am writing for my webapps. I like it calling integration tests, because it tests the app from the webapi down to the database, the whole application logic in other words. For simple web apis or most of microservices it's enough, but for complex db logic I additionally write integration tests for my repositories
I consider it an integration test as long as it's using real external services
Great content. Thanks!
My pleasure!
What theme are you using in VS?
Dark theme + ReSharper
I am struggling to understand the difference between unit tests and functional tests , or are they the same ? I use to write unit tests but without docker, test-containers, web application factory. What is the benefit of this approach ?
They are not the same. Unit tests just test a simple unit - class/service/method. Functional tests actually execute the ASP.NET Core application and run the tests. Huge difference.
Great video! Thanks!!
Glad you liked it!
I handle authorization part with AuthenticationHandler by overriding HandleAuthenticateAsync to use claims, found that there is a pretty common way to handle authentication part, but couldn't find the way to provide different claims for each test. What will bi the proper way to do such a thing ?
Why not create the appropriate token with claims before executing the test?
@@MilanJovanovicTechAnd how would you do this in an integration test that only tests a MediatR handler?
How to perform docker login with the TestContainers PostgreSql package?
Don't need to, you can connect to it from the test
Instead of using your CustomProblemDetails, could you not use Validation ProblemDetails? 🤔
Yes, i could've used that
OK cool, I wondered for a moment whether its avoidance was from deliberate reason that I was unaware of.
Thanks 👍
This is pretty nice! How would you take the approach of functionally testing a Blazor application?
I guess something similar to Selenium? I don't do Blazor, but I expect something like that to be available
Can I access this codebase somewhere?
Check out this: github.com/m-jovanovic/testcontainers-sample/
Is this code on your github account? I'd like to look at your request and response/result base classes. Thanks.
The Result type? gist.github.com/m-jovanovic/aa25b1ae424c985ff8ae696a79b6fe6e
Authentication and Authorization part is missing.
Grab a Cookie/JWT and add it to the request, shouldn't be too hard
I use the exact same setup but call it an integration test. You can make this easier without TestContainers and just use a DbContext options factory.
It goes by many names, it's all integration testing as long as you use a real service under the hood
@drhdev are you running your integration tests with live external services? The dB context options just let's you configure it for existing data bases
CustomResult and result.Match ? I don't know
It's a ProblemDetails, just a helper to let me turn the Result object into ProblemDetails
@@MilanJovanovicTech I can't follow in your footsteps like this, there are many updates in many places, we don't know