Clojurians Log v2

That's interesting. What's the docstring?

This is the entire test:

(defdescribe keyword-url-lazy-test "keyword-url"
  (expect-it "should return the correct URL"
             (= "/greek-match-making"
                (keyword-url {:nationality "Greek" :defaultURL ""}
                             {:keyword "Match Making"}))))

oooh the checkmark, yes, my apologies

Whatever Kondo does for deftest suppresses references and this warning.

ugh

If you lint as deftest you get no warnings 🙂

But I assume Kondo is doing something specific for deftest regarding references?

that relies on a clj-kondo bug: https://github.com/clj-kondo/clj-kondo/issues/2428

If I switch to deftest and this bug is fixed, then the docstrings will be marked as "unused value"

Hmm I have an idea

Heh, fun.

okay, i think i figured it out, due to #lsp’s tireless work

:exclude-when-defined-by #{my-ns/defflow}

so i've switched from the hook to excluding vars defined with defdescribe from :clojure-lsp/unused-public-vars checks

i pushed the change, let me know if that works for you

i'm offline now, but i can help out more once my kids are in bed later lol

Yup, that addresses it. Nice!

No. It's a Polylith test runner. It wraps other test runners. Like there's a Polylith test runner that wraps Kaocha.

The issue is that every test runner has its own machinery for discovering tests, figuring out which tests to skip or to run, executing the tests and collecting all the results from them, and then reporting those results. In my p.e.t.r case, Polylith figures out what namespaces need to be tested. My test runner only has to detect whether a given namespace has any tests and, if so, run them and get a hash map summary. Then Polylith reports the success/failures at a namespace level (and whatever default reporter each test library uses prints details).

Cognitect's test-runner has a whole bunch of machinery for figuring out what tests it will run -- and a whole bunch of options to control that. LazyTest has a whole bunch of machinery for figuring out what tests it will run -- and a whole bunch of options to control that. Kaocha is similar. So, for the most part, you might as well just run the two command-lines. In the Polylith context, my test runner does very little -- it exists primarily to provide process isolation between the Polylith program and the project tests (the default test runner executes tests in-process, so you can run into library conflicts, memory leaks, and all sorts of other stuff in large projects). Because my test runner does so little, adding a call to lazytest.repl/run-tests was easy -- and really none of the filtering options are directly exposed (there's a Polylith-specific way of passing configuration all the way through, so you don't need to deal with command-line options etc), and none of that filtering is even wired up for LazyTest in p.e.t.r.

A quick fork of Cognitect's test-runner that will run both clojure.test and LazyTest stuff https://github.com/seancorfield/test-runner -- you'll need the head SHA of the master branch of that and you'll need to depend on the head SHA of the main branch of lazytest (and run -X:deps prep) in order to try it out. It's "dumb" integration in that no options are passed to LazyTest, so all the include/exclude stuff only applies to clojure.test tests. I'll wire up options if anyone cares enough to use this. And maybe I'll even publish releases to Clojars 🙂 Assuming Cognitect don't want a PR for it.