https://github.com/nedap/speced.def does plenty of reader conditionals in test/. I find it a sound approach overall, don't recall problems from it.

thanks, gonna take a look

What is this file for: https://github.com/nedap/speced.def/blob/master/src/nedap/speced/def/test.cljc I found it confusing to find something outside the test folder which seems to be related to tests.

That's a minor ns, it provides better IDE integration for Cursive users when using (is spec-assertion-thrown?) clojure.test's API is a bit odd about how to extend it (namely with multimethods, which won't play always nicely with IDEs, linters, etc) See also https://github.com/nedap/speced.def#testing

Take a look at how HoneySQL does this: it's all cljc files and the run-tests.sh file runs both clj and cljs tests.

@USW6URK47 do you also will need to test async functions?

@U3Y18N0UC not at the moment anyway. I'll cross that bridge when (if) I get there

@U45T93RA6 I can get the tests running, and I see the feedback on STDOUT, but "return value" of node target/out/tests.js is the same regardless of success or failure. It doesn't play nice with my CI/CD pipeline.

@U04V70XH6 thanks for the pointer to HoneySQL. Sadly for reasons outside things I can change I have to use lein in the foreseeable future for the project I'm working on.

I don't think that's the case. We've paid close attention to the return value of the Node process If I intentionally break a :cljs reader branch, CircleCI will fail as expected: https://github.com/reducecombine/speced.def/commit/b1f20a3603268bd2e5c2f85e6f9d1ee2bb44647e

The magic sauce being https://github.com/nedap/utils.test/blob/1d4f5a66e89b4703bb8ab227021441946d295373/src/nedap/utils/test/api.cljc#L32-L42 IIRC many codebases implement their own flavor of this - you're right that the unix return code is a gotcha, but not an insurmountable one

I was looking at this: https://github.com/nedap/speced.def/blob/master/test/nedap/utils/spec/test_runner.cljs I must admit I expected less DIY and more "plug this and it works". I guess lein test set my expectations a bit to high when it came to ClojureScript 😄

I will take a look at the "magic sauce" 😄

lein test generally hasn't worked well for me for running two targets at once (jvm + node). it's seems 'complected' to me, which is why it's explicitly disabled in project.clj

lein test works in my opinion VERY well for Clojure (JVM) only. No hairy setup or knowledge about "test runners" is necessary - it just works as expected. A combined Clojure and ClojureScrip (JVM + node) is an entirely different story - sadly.

The JVM-only experience had set my expectations very high for a combined project. Expectations that obviously was not met 😅

@USW6URK47 Are you using CLJS with Shadow-CLJS? If so, there's a target called :node-tests that do what you want. If you don't, or can't, use it theres another option like re-defining the summary to exist on the right exit code 😄

An example on how I did in a project: https://github.com/mauricioszabo/duck-repled/blob/master/test/duck_repled/tests.cljs#L33-L41