One rule of thumb I've heard and follow is that every time you encounter a bug, you write a unit test that would catch it. I find that does a pretty good job of getting high code coverage, though maybe that's cause my code is naturally buggy 😅.
Not testing is crazy. Once you realize you can actually refactor without ever having the fear you've broken something, there's actually opportunity to make rapid improvments in structure and performance. Taking 2 minutes to write the test can save your hours of debugging. Unless you're building a throwaway prototype, not unit testing is always the wrong choice.
All you folks are crazy not to unit test personal projects. Unit tests don't need to be fancy and exhaustive. A sanity check and having a simple way to execute isolated code is well worth the 15 minutes of setting it up. Heck, just use them as scratch files to try out libraries and APIs. I can't imagine having the kind of time to raw-dog that f12 button and sifting through print() nonsense all night.
I think it also very much depends on your tooling & how easy it feels to start writing unit tests.
When I work in a Java project for example I always write unit tests even for personal stuff, because the IDE integration is great and it's really quick to create a test class, run it and see granular results. I don't feel the same way about testing JavaScript because the tooling at least for me hasn't worked quite as well (though that could very well be my own fault, it's been a while since I looked into it). The more cumbersome setting up and running the tests is, the more tempting it becomes to just use the console or manually test parts instead.
If I ever want to take a break from my personal project and come back to it. Unit tests are necessary. If I try to continue working on a project that doesn't have unit tests. I feel like every bit of code I touch is introducing countless hours of debugging. It really demoralizes me.
Here's my take. In order to be able to write meaningful unit tests the code should be structured in a certain way, with very modular, decoupled units, dependency injection, favoring composition and polymorphism over inheritance and so on.
If you manage to write your code this way it will be an objective advantage that will benefit the project even if you don't write a single unit test. But it does make unit tests much easier to write, so presumably you'll end up with more tests than otherwise.
IMO teams should prioritize this way of writing code over high test coverage of non-modular code. Unit tests for deeply-coupled code are a nightmare to write and maintain and are usually mostly meaningless too.
I really hate the projects I work on where they've overtested to the point of meaninglessness. Like every single class has a full suite of tests mocking every single dependency and it's impossible to look at it without breaking 50 test cases.
Similarly I hate the projects with no tests because I can't change anything and be sure I've not broken some use-case I didn't think about.
Much prefer the middle ground with modular, loosely coupled code and just enough tests to cover all the use cases without needing to mock every single little thing. It should be possible to refactor without breaking tests.
TDD is pretty difficult when you don't have a very clear idea of what's the best way to achieve what you want to do – which is most projects most of the time.
It's usually also a hard sell for management, especially if the project is on its first steps and it has to show something functional in order to keep its budget.
Last but not least, the problem domain for unit tests is pretty focused. The parts of the code that are suited for unit testing may not be all that much. I know that TDD is not just unit tests but I mentioned it in the context of this discussion.
1st problem - you should be coding such that you can make at least two iterations. It should be perfectly okay to code it once so that you have an idea and then throw it out and do it correctly. If you work in sufficiently small steps, there's barely any time waste plus you save time that you would spend maintaining
2nd problem - Kent Beck's (the guy who "invented" TDD) idea behind TDD was that it should help you be less stressed. If your job is to make a prototype or get something really out the door really fast and you won't lose sleep over it, then there is no need to write tests. TDD is there to enable you.
I am not exactly sure, what you mean with the unit tests, so apologies if I am misunderstanding. I use integration tests in my red-green-refactor loop as well
I hate manually verifying stuff. There's no way I'm going to walk through all the cases I care about after making a change. So I just run my stupid tests.
this especially makes you feel idiotic when you spend a long ass time finding & fixing a simple bug resulting from a typo then realize that it would've been caught immediately if you just wrote tests like you said you would 3 days ago
you'll change your mind when it's more effort to set up a test situation for a manual test. like when you change something that communicates with an expensive propietary API and you can't just push it to production.
I prefer acceptance tests or end-to-end tests mostly for personal projects. Libraries I write will have unit tests though.
However, for the most part. Personal projects are a lot of prototyping and throwing away. Unit testing and testing in general can get in the way of that.
A good unit test is an expression of the intent of the programmer that wrote the unit of code. How would Copilot know what your intent was?
What Copilot does is to test exactly what the code is currently doing... which kinda defeats the purpose.
But it does raise a good point, because that's usually exactly what teams are asked to do (especially when there's a high coverage requirement). It's completely pointless busywork, but that's what makes it perfect for Copilot.
Copilot happily writes failing testa too, TDD style. It just generates code tests that are very common and probably won't catch those hard to miss cases anyway.
So so many unit tests I see don't meaningfully test anything. It would be faster to just read the unit under test because the test itself presents nothing that you wouldn't instantly recognize. Or the test is so tightly coupled to some arbitrary property that of course the test fails whenever you change something. UI tests at my current place are terrible for this, as they're just comparing DOM structures so any change breaks it.
No. If your goal is to move towards automated testing, see what language is used by the most popular automated frameworks and pick up a beginner's tutorial for that language.
At least for frontend development, with new faster e2e testing frameworks that can even record the test while you see your work in a browser, lately I've been feeling I want to write more e2e tests and less unit tests.