Code passes locally, fails after submitting.
https://exercism.org/mentoring/external_requests/e25773a53b664a71837bb76b1b066377
Hi! Is there a way to trigger CI run on test runner repo even if PR is triggered on Pharo track? This is because whenever new exercise is created on Pharo track, we need to update test runner docker image as well (test runner Pharo image that runs in docker takes Pharo track repo and run tests against known/implemented exercises). Right now the Pharo track repo are out-of-sync with test runner repo, because docker image (with test runner Pharo image inside) has older specification of exercises.
Any thoughts? I can push some dummy changes to test runner, so PR is triggered and docker image is re-built. Some sort of dependency/automation would be nice.
Some evidence: This is change related to test runner logic: Exit code should be always 0, when running tests. - commit from 10th July.
But CI build on test runner ran last time on 31st May. I will trigger CI on test runner manually now.
@glennj Please try now! Docker image should be updated with latest changes. It should work now (passed there vanilla solution).
There are still some trouble with producing/interpreting output json file with test results, if problem occurs in particular test case (e.g. typo in method name). All other test cases should pass and error for particular case should be displayed. Now it shows just generic error, which is empty. I need to find out, how ruby code interprets json data, or I put wrong data in error output (which is weird since vanilla/golden tests pass).
I’m pretty sure that output.json is produced and exit code is 0 (tested locally). But for some reason I see in UI only message: “Error occured: We received the following error when we ran your code:” But there isn’t nothing more. I still don’t know, how the output.json is being interpreted.
Yes, all passing now.
Right, but if you make some typo (intentionally) and run tests again, failed test cases aren’t shown unfortunately. I’m not sure, under which circumstances is results.json consumed by UI. Here is snippet:
{
"version" : "2",
"status" : "error",
"tests" : [
{
"name" : "Reading empty buffer should fail",
"status" : "pass",
"test_code" : "test01_ReadingEmptyBufferShouldFail\n\t\"Tip: Remember to review the class [Comment] tab\"\n\n\t<exeTestName: 'eading empty buffer should fail'>\n\t<exeTestUUID: '28268ed4-4ff3-45f3-820e-895b44d53dfa'>\n\tcircularBufferCalculator capacity: 1.\n\tself should: [circularBufferCalculator read] raise: Error."
},
{
"name" : "Can read an item just written",
"status" : "error",
"message" : "Message not understood: CircularBuffer >> #capacityu\n---Stack-trace---
....