I've seen this in practice too. Lots of unit tests for small things that actually make the overall codebase brittle.
I think it comes from a pretty decent practice of coding from the tests outward. However I'd argue the long term value of many of those tests is minimal.
Yes. I write lots of unit tests, but they are
mostly part of the coding process: I find they
are superior to the manual testing I would have
to do otherwise every time I add new code.
But as long term protections against regressions, such
tests are less valuable than more coarse-grained
integration tests.
I think it comes from a pretty decent practice of coding from the tests outward. However I'd argue the long term value of many of those tests is minimal.