How to run ember-cli test in jenkins - testing

Question

How can i run Ember-cli test in jenkins?
Currently to run the tests i added a build step "Execute shell" with the following in it:
ember test --silent --config-file ${WORKSPACE}/testem.json > ${WORKSPACE}/xunit-ember-dev.xml;
But it doesn't work, this is the output
<testsuite name="Testem Tests" tests="0" failures="0" timestamp="Thu Feb 12 2015 14:20:24 GMT+0100 (CET)" time="0">
</testsuite>
If I do the same manually in the workspace as jenkins user I got the expected results.
<testsuite name="Testem Tests" tests="70" failures="0" timestamp="Thu Feb 12 2015 15:06:40 GMT+0100 (CET)" time="15">
<testcase name="PhantomJS 1.9 Integration - Homepage: Should display the homepage"/>
<testcase name="PhantomJS 1.9 Integration - Profile: Should display the profile sections"/>
...
Every time I make Jenkins run the tests, he find no tests.
Thanks

Answer 1

Just add an Execute Shell step that runs:
npm run test > results.tap
This command tells npm to run a script called test and redirect the output to a file called results.tap. (The purpose of redirecting to results.tap is so you can take this file and hand it to Publish TAP Results post-build step and get pretty charts of test runs.)
In your package.json, you should have a block that looks like this:
"scripts": {
"start": "ember server",
"build": "ember build",
"test": "ember test"
},
(This is the default from ember-cli 1.13.8 and likely a few versions earlier than that.)
Using this approach, you don't need to have ember-cli globally installed on your build system: it will simply pick it up from the project's node_modules folder.
Caveat emptor: you'll likely find some blogs/forum posts that tell you to run npm run test | tee results.tap. This will run the tests, however, it will eat the return code. (If the test run fails, the npm process exits with a return value of 1, but since you're piping output to a second command (tee), the return code of that command is what jenkins sees. The net result is that jenkins will interpret test failures as success and keep on running build steps.)