Can you remind me again what the problem with globbing is? This looks verbose and easy to miss tests by accident.

Can you remind me again what the problem with globbing is? This looks verbose and easy to miss tests by accident.

@Shillaker see the Note in the CMake docs. In my experience, the Note translates to having to remember to re-run CMake for the globbed file list to be re-generated, otherwise CMake has no way of knowing what files have changed. This is counter-intuitive, as it defeats a bit the purpose of globbing. On the other hand, adding the test names is more verbose, but once you get used to it you don't forget to add new files, and don't need to explicitly run CMake again.

Hmm, I'm still worried that somebody will accidentally delete a test file or forget to add one. Is it possible to have a catch-all glob at the end as well or is that really gross?

Why is this a problem? AFAICT we list sources by name everywhere else in the the repo, why should tests be any different?

If we leave sources out in the rest of the codebase, something will fail to compile (unless the code in that file isn't used anywhere, in which case it should be deleted). Tests are different because they're at the end of the include chain, i.e. they are not included by any other files implicitly or explicitly, so it's impossible to tell whether all of them are being run.

Even looking at this review, I have to go through the file tree to make sure that all of them are listed in the CMake file, as there's no way to check whether we've missed one (AFAICT). To me, this problem of accidentally missing tests (and thus thinking code is tested when it's actually not) outweighs the inconvenience of having to run cmake when you add a new one, however, if you feel strongly about it, I'm happy to merge as it's not a big deal and you're actively working on it atm.