I can't take this seriously without proper tests. In the Real World, none of this code would have been approved and deemed ready for production without proper test coverage, even for simple CRUD, imo.
But seriously, these apps are large enough where adding in testing initially would only complicate the core mission: to show the common software patterns that are required for day to day development.
I think tests are a good idea, and I’m a huge fan of TDD in general. Building Conduit implementations without tests are a large time investment as is, so we originally didn’t include testing in the spec because we figured that if people wanted it, then it would be a great “extra credit” objective for the repo. For example, the Angular 2 repo had a request for unit tests and folks are now starting to work on a PR for it.
Another reason we didn’t include them in the spec is actually from this line in the current spec:
> The quality & architecture of Conduit implementations should reflect something similar to an early stage startup's MVP: functionally complete & stable, but not unnecessarily over-engineered.
Most startups I know that work in consumer facing apps (like Conduit) don’t apply TDD until they have solid product-market fit, which is smart because they then spend most of their time iterating on product & UI and thus are far more likely to find PMF.
This doesn’t mean that TDD === over-engineering, but in certain circumstances that statement does evaluate true (ex: consumer product finding PMF, sideprojects, robust prototypes, etc).
So TL;DR — we chose to not have a strict requirement for TDD in these repos, but we'd definitely prefer all of them to include excellent tests if the maintainers are willing to add it (or if someone in the community is kind enough to make a pull request :)
If you want to help, the repo's here: https://github.com/mchandleraz/realworld-vue
Or, I'm sure the other repos would greatly appreciate some help with tests too :)
Have you read any research into test coverage and bug frequency ? From what I have seen it does not have significant impact.
I like tests for lots of things they are useful, but code can be exemplary without them. And enforcing x% coverage before release is ridiculous.
Are there any studies showing that mutation testing significantly reduces the amount of bugs released ?