Implement infinite loop detection of child process

[make-github-pseudonymous-again]

It is not very clear as if there is a way to limit the execution time of a test run. It would be useful for example if the tested code contains an infinite loop.

[kof]

You mean per test or per file?

[kof]

There is QUnit.config.testTimeout but I didn't expose this setting to be configurable via runner api. Currently its accessible within tests.

What is your use case?

[make-github-pseudonymous-again]

I would like to try the code mutation methods provided by https://github.com/shybyte/grunt-mutation-testing on some JavaScript project. I would then test the mutated code with node-qunit. Mutated code could produce non terminating runs, but in this case a non terminating run would mean that the mutated code is buggy and thus that the test was successful since mutated code is not supposed to pass all the tests.

[kof]

I am not sure I understand you. If its an async code and callback never gets called, qunit will trigger a timeout error anyways.

[kof]

Maybe just give me an example of your code and test you would have ... and a concrete problem with the current version.

[make-github-pseudonymous-again]

Concrete problem is that I don't know how to set a timeout on this testrun.js script (see code below).

If its an async code and callback never gets called, qunit will trigger a timeout error anyways.

How much time will qunit wait?

lib.js

exports.version = "1.0.0"

infinite-loop.js

test("infinite-loop", function(){ for(;;); });

runtest.js

var path = require('path');
var testrunner = require('qunit');

testrunner.options.coverage = true;
testrunner.options.log.tests = false;
testrunner.options.log.testing = false;
testrunner.options.log.assertions = false;
testrunner.options.log.summary = false;

var n = 1;
var failed = 0;
var done = 0;

var cb = function(err, report) {
    if (err) {
        console.log('error :  ' + err.message);
        process.exit(-1);
    }

    failed += report.failed;
    ++done;
    if (done === n) process.exit(failed);
};

testrunner.run(
    {
        code : {
            path : path.normalize("lib.js"),
            namespace: "lib"
        },
        tests :  path.normalize("infinite-loop.js")
    }, cb
);

[kof]

Concrete problem is that I don't know how to set a timeout on this testrun.js script (see code below).
For me this is not the problem, this is what you want to be able to do, I am still trying to understand your motivation.

http://api.qunitjs.com/QUnit.config/

Just found out that the default number is undefined, so it will not, I was wrong. And we have currently no api to pass it. Do you want to submit a pull request for this?

[kof]

Basically we can allow setting any of this options.

[kof]

But also you can set it as I said within the test file itself, thats why I am trying to understand what you do.

[kof]

You have a direct access to QUnit within your test file via QUnit namespace.

[make-github-pseudonymous-again]

You mean putting QUnit.config.testTimeout = 12345 at the beginning of the test file? What if there are multiple test files?

[kof]

yes, every test file is runnning in own spawned process, so you need to do that in every test file then.

[make-github-pseudonymous-again]

Well, putting QUnit.config.testTimeout = 1 doesn't seem to work so...

[kof]

you also used an asyncTest or stopped the test?

[make-github-pseudonymous-again]

QUnit.config.testTimeout = 1;
asyncTest("infinite-loop", function(){ for(;;); });