initial tests for clojure.core/merge-with

[perrygeo]

Testing clojure.core/merge-with

Manually tested with

• JVM
• Clojurescript
• Clojure CLR
• Babashka
• Basalisp (fails, see comment on nil handling)
• PHEL (I don't have access to PHP tooling at the moment)

Dialect Differences

• Clojurescript returns a nil in the zero-arg case. Clojure etc throw an exception.
• Clojurescript: numeric keys have different equality semantics than the JVM.
• Basalisp returns nil on (merge-with second-arg nil {}), rest return {}

---

Anything else I can do to improve this? Let me know. This is a really fun way to get to know the clojure standard library!

Resolves #361

[CLAassistant]

All committers have signed the CLA.

[E-A-Griffin]

basilisp and phel status is unknown, are the core 4 langs sufficient?

They are yeah, @jeaye disabled them for the time being, that being said noting how they fail is always nice for their respective maintainers to be aware of (and potentially fixed)

[E-A-Griffin]

Could you use the following reader conditional order: https://github.com/E-A-Griffin/clojure-test-suite/blob/e00fe2b8ff87fcf02020664cc57f11bbf4448904/doc/writing-tests.md#reader-conditional-order? This PR isn't merged yet but @jeaye and I agreed on it prior so going to try to maintain it going forward

[E-A-Griffin]

What is this TODO for?

[perrygeo]

When I run bb test-cljs, everything passes but the output contains this verbose warning:

...
Ran 223 tests containing 4957 assertions.
0 failures, 0 errors.
===============================================
[:test] Build completed. (324 files, 323 compiled, 1 warnings, 4.22s)

------ WARNING #1 - :fn-arity --------------------------------------------------
 File: /Users/matthew/projects/clojure-test-suite/test/clojure/core_test/merge_with.cljc:15:30
--------------------------------------------------------------------------------
  12 |      (testing "zero arg behavior"
  13 |        ;; NOTE: clojurescript differs from the rest here, bug?
  14 |        #?(:clj     (is (p/thrown? (merge-with)))
  15 |           :cljs    (is (nil? (merge-with))) ; TODO how to suppress warning
------------------------------------^-------------------------------------------
 Wrong number of args (0) passed to cljs.core/merge-with
--------------------------------------------------------------------------------
  16 |           :default (is (p/thrown? (merge-with)))))
  17 | 
  18 |      (testing "`nil`ish behavior"
  19 |        (is (nil? (merge-with second-arg nil)))
--------------------------------------------------------------------------------

Since we're explicitly testing the zero-args case, is there a way to silence this?

[E-A-Griffin]

I think sorted-map tests would also be nice

[perrygeo]

Thanks for the review @E-A-Griffin ! The reader conditional ordering actually matters, TIL. I added tests for sorted-map and cleaned up a bit.

The cljs zero-arg case is the only real question mark. I've noted the difference in the top PR description. The warning still bothers me though, hoping to find a way to suppress that.

[Copilot]

Pull request overview

Adds a new clojure.core compliance/characterization test namespace covering merge-with, with some dialect-conditional expectations (notably ClojureScript’s zero-arg behavior).

Changes:

• Introduces a new merge-with test file with coverage for nil handling, collision semantics, fold direction, duplicate-key invocation, sorted-map preservation, and nested-map merging.
• Adds negative tests around non-map inputs and exception behavior (with dialect conditionals).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[jeaye]

Thanks for the PR!

[jeaye]

We can delete this fn and just use second.

[perrygeo]

second takes one arg and returns the 2nd element. I guess we could write #(second [%1 %2]) but not sure if that's an improvement.