To divide jobs across multiple browsers and speed up tests.

https://github.com/NullVoxPopuli/emberclear/pull/728/checks?check_run_id=580846382

Running in to the above ^ where I think the element isn't found, but I don't know why.

Locally, I can't anything to happen -- potentially because execa doesn't work on windows. (execa is what I'm using to launch a fake server from built assets)

It's not showing up anymore.

It is currently leaving them open

Sometimes it's useful to see the test run at a human speed, instead of as fast as possible.

This appears to be supported in webdriver https://stackoverflow.com/a/23787258/1703845, but I don't think webdriver.io has implemented it. If I wanted to do this, I would probably have to dig into webdriver.io source.

When you hit a breakpoint, you may expect the browser to pause, but it continues running. This means you might put a breakpoint to try and debug a loading state, but the browser finishes it's loading state while you're paused. https://webdriver.io/docs/api/chromium.html#freeze is supposed to pause the JS execution for you to debug properly. If we could use this somehow, that would be great.

The reason --env is separate from --tag is because you only ever run one env at a time. The same is true for roles. Perhaps this is a indicator that roles should be separate too?

For example, on emberclear, I get the following output:

link: https://github.com/NullVoxPopuli/emberclear/pull/279/checks?check_run_id=641002972

 1) chat
       New users can add each other and then communicate
         fresh users can begin communicating with each other:
     waitForInsert: waitForInsert([data-clipboard-text]): waitUntil condition timed out after 30000ms
      at Browser.waitForExist [as waitForInsert] (node_modules/@faltest/browser/index.js:321:15)

This really doesn't tell me anything. It'd be great to know the current state of the browser / dom / screenshots or whatever so that the C.I. log is useful.

because sometimes you need to escape the ! with --tag '!tag'. It may be useful in this case to alias the sigil with another, like –tag ^tag.

I'm wanting to use all the bells and whistles, and I'm having a bunch of trouble.

• how do I properly setup the chai plugin?
• how do I integrate this with C.I.?
• etc?

I'm learning these via lots of googling, but it'd be nice to have everything all in one place, maybe on a docs site or something.

If a test fails, then passes (using --retries=1), we should log a debug message so you can see flakey tests in the logs.

Right now we bundle a version of chromedriver and geckodriver. We should make it so the consumer can choose their own version.

Instead of adding a bunch of --envs to get different urls, we could also allow a base url option to target any number of urls.

I see :

const { setUpWebDriver } = require('@faltest/mocha');

but isn't that supposed to be from @faltest/lifecycle?

My builds over at https://github.com/NullVoxPopuli/emberclear/pulls started failing due to a chrome-driver mismatch.

Can I ignore the error? or.. what needs to happen to be a little more flexible with chrome versions?
(maybe I need to pin the chrome version, but I worry about that becoming stale (side project, minimal effort, etc))

We should check overrides like waitForTimeout, logLevel and create new sessions if they are different.

atm, I have this:

describe('smoke', function() {
  before(async function() {
    switch (process.env.WEBDRIVER_TARGET) {
      case 'pull-request': {
        console.info('---- DEPLOY PREVIEW ----');
        console.info(process.env.DEPLOY_URL);

        this.host = process.env.DEPLOY_URL;

        if (!this.host) {
          throw new Error(`host not set. Did you forget to set $DEPLOY_URL?`);
        }

        break;
      }
      case 'local': {
        let { server, port } = await startServer();

        this.server = server;

        this.host = `http://localhost:${port}`;

        break;
      }
      case 'ember': {
        this.host = `https://localhost:4201`;

        break;
      }
      default: {
        this.host = 'https://emberclear.io';

        break;
      }
    }
  });

and only pull-request, local, and default are allowed targets., so I actually need to comment/uncomment stuff to swap between local and ember.

I ran this command:

C:\Program Files\Volta\bin\node.exe --inspect-brk=49376 node_modules\@faltest\cli\bin\index.js --browsers 2 --timeouts-override 900000 --target ember 

Maybe percy.io would be a good fit.

example https://github.com/NullVoxPopuli/emberclear/blob/0982305844/packages/smoke-tests/page-objects/login.js#L19

cc @NullVoxPopuli

I see that there is a chai package, and that it exports chaiWebDriver, what's that do and where do I put it?

roles('#foo #bar #baz', function() {

into

let _roles = [
  '#foo',
  '#bar',
  '#baz'
].join(' ');

roles(_roles,  function() {

into

roles('#foo', '#bar', '#baz', function() {

or

roles(
  [
    '#foo',
    '#bar',
    '#baz'
  ],
  function() {

cc @MichalBryxi

Since Chrome is using a new session / config each run, it hasn't accepted my local SSL Cert that I'm using for local dev.

S far my experiments concluded that adding .only after it in the internal test suite (e.g: packages/mocha/test/acceptance/index-test.js) does not filter the tests being run via yarn test. So I just trial-and-errored on:

npx mocha packages/mocha/test/acceptance/index-test.js

• Is there a better way to run the internal test suite?
• Should that be documented somewhere?
• Shall that have an explicit term (I see "internal test suite" mentioned once in the source code)

mochajs/mocha#4142 (comment)

I have this code in a page object:

  get inviteUrl() {
    let dataAttr = 'data-clipboard-text';
    let element = this._create(`[${dataAttr}]`);

    return element.getAttribute(dataAttr);
  }

and I placed a breakpoint on the return, and in the VS Code console, I did:

element.getAttribute('[data-clipboard-text]').then(console.log)
Promise { pending }
element.getAttribute('data-clipboard-text').then(console.log)
Promise { pending }
// results:
undefined
undefined

so, the element exists, and is selectable by attribute, but I can't get the attribute value? maybe?

anything look obviously wrong?
I have a WIP PR: here: NullVoxPopuli/emberclear#279

something unrelated, (if anyone tries to debug this), is that I have interactions that don't seem to finish before moving on to the next step, but also somehow don't fail?
I seem to only be able to get the test moving along by setting these breakpoints:

any help is much appreciated <3

We should also try to log any ajax errors too.

cc @NullVoxPopuli

I watched someone provide --disable-timeout instead of --disable-timeouts. Yargs should say unrecognised option because we set strict mode, but it didn't.

In my app, https://github.com/NullVoxPopuli/emberclear , I have some websocket behavior that I want to ensure communication is happening between the two browsers.
This could be for testing:

• changes are immediately reflected on a page for all - chat
• etc

So, the browsers can't share the same session, as same session sync can happen without websockets

Thank you for all the hard work ⭐️. It would be nice if we could have the description be a bit longer. The current description reads:

A different take on WebDriver browser testing

As a complete novice I’m not entirely sure what this means. As a veteran WebDriver user I would like to know what the differences are. Could this perhaps be expanded upon?

I am mostly curious if this project uses webdriver

In the readme it doesn't explain too much about this project, I don't know why I would use this project over others.

If you kill webdriver (chromedriver) while tests are running, the test errors don't help you figure out what's wrong. The first test will show:

connect ECONNREFUSED 127.0.0.1:45244

and the following tests will show something like

TypeError: Cannot read property '0' of null
    at Context.setUpWebDriverBeforeEach (/foo/node_modules/@faltest/lifecycle/src/index.js:175:32)

When running headless locally, you often get the popup "Google Chrome wants to use your confidential information stored in ’Chrome Safe Storage’ in your keychain. Do you want to allow access to this item?".

You can ignore it and all tests will pass normally, but if you deny it, it fails the current and subsequent tests. I'm not sure clicking "Always allow" is the best solution since you can ignore and tests pass. We may be able to start chrome with a flag to disable?

I found more info at the bottom of https://docs.parasoft.com/display/SOA9103/Web+Functional+Testing+-+Overview.

because in the code it's also referenced as _extendUnscopedMany

The first goal is to share page objects between Ember.js tests and FalTest. Then other frameworks can come later.

Browserstack offers access to tooooooons of devices and browsers. Would be great if faltest worked in those environments :)

Took me a bit to get through purgeOld method to implement it properly. There are few things that are not obvious:

• When you add new record, you want to use name = Rows.encodeString('something')
• When you call purgeOld, you don't want to use encodeString: await rules.table.rows.purgeOld(['something]);
• Why is there timeout and what does it do? Why would I want to implement this method / What are the common use-cases?