lambdaisland / chui Goto Github PK

When cljs.core.async.interop/<p! is used inside the test results are shown immediately as "Pass" while the test is still in progress. And when an error occurs nothing is updated on UI: "Running" and "Pass" are shown forever

Code example

(deftest some-async-test
  (testing "Some async test should fail"
    (async done
      (cljs.core.async/go
          (cljs.core.async.interop/<p! (returns-promise)) ; error occurs here
          (is (= false true)) ; never executed
          (done))))))

Steps to reproduce

Open the page with tests

Actual result

"Run x tests" is disabled

Expected result

"Run x tests" is enabled on the first render

Workaround

Change input value or click any of search checkboxes

It's common to have the editor and the browser (with the test tab open) positioned side-by-side on the screen. But the current chui-ui is not responsive, e.g. what about when the browser viewport width is small, it just hides the list of test runs, since most of the time people only care about the last run. Or it could provide a button to manually hide a column.

Hi,

in the test output the funnel-client/whoami is included as part of a test's id.

it would be useful to be able to set the whoami to a fixed name, so that the test id's are constant over time. this allows tools that show what commit caused a test to fail. stats about flaky tests etc.

For whatever reason, my cljs tests that use (is (match? ,,,)) from nubank/matcher-combinators are blowing up in Safari 15.6.1 with the error TypeError: re-find must match against a string.

It looks like the issue is an unexpected nil at https://github.com/lambdaisland/chui/blob/main/modules/chui-core/src/lambdaisland/chui/runner.cljs#L93-L94. My quick logging check on the erroring tests was giving me a value for line-col of "@".

It seems to me this is imperfect and likely to be revisited when conditions change in cljs.test.

Can I propose the following patch? If so, I'll make up a PR. Test line/col reporting remains broken (as it is now). But least it will get us to actionable test failures on Safari:

diff --git a/modules/chui-core/src/lambdaisland/chui/runner.cljs b/modules/chui-core/src/lambdaisland/chui/runner.cljs
index c3d9575..af1c89f 100644
--- a/modules/chui-core/src/lambdaisland/chui/runner.cljs
+++ b/modules/chui-core/src/lambdaisland/chui/runner.cljs
@@ -90,8 +90,8 @@
         (let [line-col (drop (- (count frame) 2) frame)
               file     (str/join ":" (take (- (count frame) 2) frame))]
           {:file   file
-           :line   (js/parseInt (re-find #"\d+" (first line-col)) 10)
-           :column (js/parseInt (re-find #"\d+" (second line-col)) 10)})))))
+           :line   (js/parseInt (re-find #"\d+" (str (first line-col))) 10)
+           :column (js/parseInt (re-find #"\d+" (str (second line-col))) 10)})))))
 
 (defmulti report :type)
 (defmethod report :default [_])

After upgrading to Clojurescript version 1.10.844, we started seeing this error when loading the browser runner in chui-shadow:

Failed to load lambdaisland/chui/shadow/browser_runner.cljs TypeError: G__52095.addHandler is not a function
    cljs$core$IFn$_invoke$arity$2 glogi.cljs:134
    cljs$core$IFn$_invoke$arity$2 glogi.cljs:156
    cljs$core$IFn$_invoke$arity$1 glogi.cljs:151
    lambdaisland$glogi$console$install_BANG_ console.cljs:81

Bumping glogi to to 1.0.100 as suggested here lambdaisland/glogi#10 solved the issue for us.

$ clojure -Abrowser-test -Stree

lambdaisland/chui-shadow 1.0.173
  . lambdaisland/chui-ui 1.0.173
    . lambdaisland/chui-core 1.0.173
      X lambdaisland/glogi 1.0.70 :use-top
lambdaisland/glogi 1.0.100

Happy to make a PR to bump the glogi dependency in chui if you think it's a good idea?