Can you share screenshot of some trivial repro case, like (is (= 3 (+ 1 2))?

kaocha output:

indeed, im seeing the exact same results. also, for the record, loading this file:

(ns cursive.test-runner-test
  (:require [clojure.test :refer :all]))

(deftest expected-actual-order-test
  (is (= 2 (+ 3 3))))

(run-test expected-actual-order-test)

Loading test/cursive/test_runner_test.clj... 

Testing cursive.test-runner-test

FAIL in (expected-actual-order-test) (test_runner_test.clj:5)
expected: (= 2 (+ 3 3))
  actual: (not (= 2 6))

Ran 1 tests containing 1 assertions.
1 failures, 0 errors.
Loaded

tried it from the command line too, without any kaocha or matcher combinators being present on the classpath and the result is the same:

❯ time clojure -Srepro -Sdeps '{:paths ["test"]}' -M -e "(require 'cursive.test-runner-test)"

Testing cursive.test-runner-test

FAIL in (expected-actual-order-test) (test_runner_test.clj:5)
expected: (= 2 (+ 3 3))
  actual: (not (= 2 6))

Ran 1 tests containing 1 assertions.
1 failures, 0 errors.
clojure -Srepro -Sdeps '{:paths ["test"]}' -M -e   0.98s user 0.08s system 80% cpu 1.326 total

The thing is that there’s no defined order - kaocha might have a convention, but (is (= x y)) just checks that two things are equal, with no concept of which is expected and which has been calculated. However, that message in the test runner doesn’t come from Cursive so it must be coming from either the IntelliJ test runner or clojure.test - I’ll check. It’s possible I’m swapping the first and second values somewhere.

I was also thinking about where did the convention of the ordering come from. I think the most authorative suggestion is the docstring of the is macro:

(defmacro is
  "Generic assertion macro.  'form' is any predicate test.
  'msg' is an optional message to attach to the assertion.
  
  Example: (is (= 4 (+ 2 2)) "Two plus two should be 4")
...

Thank you @U0567Q30W and @U086D6TBN for looking into this . That said, it's not a critical issue so don't feel pressed.

BTW, there is a package for kaocha, which flips the order of actual and expected: https://github.com/magnars/kaocha-noyoda it's mentioned here: https://github.com/lambdaisland/kaocha/blob/main/README.md?plain=1#L261-L262

Right, I definitely think the order of expected/actual is pretty loosely defined here. I’d have to check, but I’m pretty sure the Cursive diff view uses “Left” and “Right” instead of “Expected” and “Actual”. But I’ll have to figure out where that message is actually coming from first.