Test does not run via grunt/phantom, but OK in web browser

[jrburke]

OS 10.8.2, using phantomjs 1.8.1 and grunt v0.3.17:

• git clone --recursive git@github.com:fabien-d/require-qunit-test.git
• grunt qunit (success: "1 assertions passed")
• Edit test/index.html and use the ../vendor/require-2.1.4.js instead of the 2.0.6 script.
• grunt qunit (failure: 0 assertions passed)

This was originally filed as requirejs/requirejs#618. I am filing a ticket here because the require-2.1.4.js version of the test runs fine in QUnit in the browser, just opening test/index.html in a web browser.

While I am open to this being an issue with requirejs, my first thought is it may be due to how the qunit grunt task binds to the running qunit tests and how it detects with QUnit is "done".

One guess at the problem may be that requirejs 2.1.4 waits a bit longer to start the actual qunit test, after whatever signal the grunt qunit driver uses to detect "done".

So, any insight into what I might look at in the grunt qunit code, or pointers is appreciated.

[kevinparkerson]

Any progress with this issue? I've come across it for my build processes, though downgrading to 2.0.6 does work.

[Zammy]

Bah.... I was pulling my hair why the f### it was not working.

[vernak2539]

Is there any progress on this? It's been a while since anything. I have downgraded and it works, but seems like a larger workaround.

[jrburke]

@frederickf in this comment may have found something about the test setup that may be important, you might try that to see if it fixes the issue for you:
requirejs/requirejs#618 (comment)

If it does fix the issue, please leave a comment back here and then maybe we can close this issue.

[shama]

Is this related to this issue #35 where qunit needs to not start until everything is loaded?

[vernak2539]

@jrburke @shama just tried to fix my problem using #35 as an example, and it works correctly. Thanks for the help.

All I had to do was put <script>QUnit.config.autostart = false;</script> after my qunit script tag and before my require script tag. Then in my first test file, inside the require statement I put QUnit.start();

Works perfect, and now I don't have to have multiple versions of require lying around

[JeremyCarlsten]

I believe I am seeing something a little different. I've got a plugin that is converting decimals to the equivalent fraction. While trying to convert it to a grunt project It almost looks to me like the tests via grunt round differently then in the browser?

are a few of the errors I'm seeing via grunt; however, they are all green when the tests are run in the browser(s). Any suggestions?

[maboiteaspam]

sounds to be related to phantomjs. Did you report there ? With some reproducible examples.

[JeremyCarlsten]

I did not but I'll give that a whirl. Thanks.

Update(3/20/2014):
I went ahead and got a post out there on phantomjs(#12078)
Also, during some downtime at work I've discovered the issue occurs only when no GCF can be found, still pinpointing the exact issue. DemoRepo

Update:(3/21)
It appears to be an issue with how far down parseint converts numbers to either 1 or 0, grunt will make them 0 while the browser will make very small numbers 1.

[markusfalk]

Found this related to the issue. http://stackoverflow.com/questions/18392463/grunt-not-running-qunit-tests-on-phantom

Is there a solution yet?

[davidbau]

I've run into this issue for every requirejs-based grunt-contrib-qunit test I have, in several different projects. Doesn't affect tests that don't use require.js.

Suspect it is realted this change: db075a2

[jzaefferer]

@davidbau that change is a week old, this issue and most of the comments are much older.

For an example of using QUnit, grunt-contrib-qunit and AMD via requirejs, check out these two files.

[Arkni]

Going to close this out, as @jzaefferer already pointed (#19 (comment)) to an example of using QUnit, grunt-contrib-qunit and requirejs, which we can consider as a solution.

for a similar solution, see http://stackoverflow.com/a/34928952/3904791