-
-
Notifications
You must be signed in to change notification settings - Fork 408
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Rethink Acceptance Testing #268
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Really looking forward to see the testing unification (oups, I said it 😬) move forward! But could you maybe explain the relationship between this and #119? Does it supercede the older RFC, at least with regard to acceptance testing?
}); | ||
``` | ||
|
||
As you can see, this proposal is unifies on Qunit's nested module syntax following |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
typo
* setup `this.visit` method to visit the given url | ||
* setup getter for `this.currentRouteName` which returns the current route name | ||
* setup getter for `this.currentURL` which returns the current URL | ||
* add DOM interaction helpers (heavily influenced by @cibernox's lovely addon [ember-native-dom-helpers](https://github.com/cibernox/ember-native-dom-helpers)) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I assume they will be using native DOM APIs as well (probably even reuse most of the implementation?), and not rely on jQuery anymore!? This could be stated explicitly maybe?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, exactly. My plan (and discussed with @cibernox and @Turbo87 a couple of times) is to essentially copy the native DOM implementations from ember-native-dom-helpers
(the addons own README infers that it is specifically aiming for this).
* setup getter for `this.currentURL` which returns the current URL | ||
* add DOM interaction helpers (heavily influenced by @cibernox's lovely addon [ember-native-dom-helpers](https://github.com/cibernox/ember-native-dom-helpers)) | ||
* setup a getter for `this.element` which returns the DOM element representing | ||
the root element |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What would the root element be in the case of an acceptance test? The app's root element?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@simonihmig - Yes. I will update the RFC though.
* setup a getter for `this.element` which returns the DOM element representing | ||
the root element | ||
* if `jQuery` is present in the application sets up `this.$` method to run | ||
jQuery selectors rooted to `this.element` |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Actually, I would like to not do this! Although this is not part of this RFC, I feel the general direction for Ember as regarding to its relation to jQuery should be to get rid of it, IMHO. At least we should try to reduce/eliminate the coupling as much as possible, so it is easy for users to opt out. Which is technically possible today with things like @cibernox's ember-native-dom-helpers
and your ember-native-dom-event-dispatcher
. While this does not technically increase the coupling maybe, I feel it still opens a trap for users, who might want to use jQuery then because of convenience and because it seems to be actively supported by this API, but later suffer from the pain having to refactor those tests when they chose to move away from it.
When dropping this, they could still wrap this.element
in a jQuery
/$
call if they really want to use jQuery, but we should not endorse to do so, IMHO!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I had written a very long explanation about how you were wrong, at the end of which I realized you were right...
Suggesting that folks use $(selector, this.element)
as needed (and from the codemod) seems absolutely fine to avoid creating a cliff.
helpers are being introduced to help avoid extra noise (e.g. "rightward shift") | ||
in tests. | ||
|
||
This means that users will be able to use either of the following interchangably: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do we really want/need two equivalent ways to use those helpers, or should'n there be one "right" way to do this? In the sense of strong Ember conventions, so teams don't end up bikeshedding about the right way. And maybe also useful for easier tooling (linting rules, codemods)? Just a thought...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@simonihmig - Yes, the blueprints will use the imported helpers. The only reason I describe both is that the importable helpers defer to the local versions on the test context (essentially as a desugaring). Do you think I need to clarify that a bit?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I updated some verbiage in this section to hopefully clarify things a bit better.
Here is a brief list of the more important but possibly understated changes | ||
being proposed here: | ||
|
||
* The global test helpers that exist now, will no longer be present (e.g. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Really not present, or rather deprecated? So they can co-exist, see the following section.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
When running tests with the old moduleForAcceptance
API they will be present (as mentioned in the section below about both API's co-existing), this section is referring to tests using setupAcceptanceTest(...)
API and is correct in saying that the global methods will not be present.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That makes sense! Maybe this should be stated to prevent misunderstandings like I had?
Yes, basically #119 is too large and too sprawling to actually implement. I started chopping it up into smaller more bite sized RFCs (like #232 and this one) so that we can make incremental progress forward. The main thing left in #119 that still is not addressed by either #232 or this RFC is the auto-discovery of test helpers and hooks (this section). |
* invoke `ember-test-helper`s `setupContext` with the tests context (which does the following): | ||
* create an owner object and set it on the test context (e.g. `this.owner`) | ||
* setup `this.set`, `this.setProperties`, `this.get`, and `this.getProperties` to | ||
the test context |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
why do we need those methods?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The main reason set
/ setProperties
methods exist today, is that sets during testing trigger an assertion if not manually wrapped in a runloop. This is something that we are actively working to remove, but that effort will likely take time (as it essentially requires Ember 3.0 where we can rely on all supported platforms having a microtask queue).
Once that deficiency is addressed, and we no longer need to use manual runloop wrapping during set operations in testing mode, we should consider deprecating (in favor of simply setting them as appropriate) them...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
thanks for the explanation, but I still don't see in what situations one would use e.g. this.set()
in an acceptance test 🤔
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ahh, I see what you are saying now, sorry about that. I agree, I don't think we strictly need them (and I suspect 99.9% of tests will not use them anyways). I included them here so that the differences between the various types of tests are as small as possible...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think I'd prefer to not include them, so that people don't get the wrong idea about the things that they could use
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Seems ok to me. We don't expect folks to need them, seems fine to avoid adding them. Will require some additional changes to the existing system in ember-test-helpers
, but that's quite simple anyways.
|
||
# Alternatives | ||
|
||
* Do nothing? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
An alternative (or maybe just a suggestion?) would be to make ember-native-dom-helpers
into a semi-official addon. Instead of copying the source code into ember-test-helpers
, why not start shifting the helper code out of the core and into an addon?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Great question, I will add this to the alternatives section.
ember-test-helpers
is an addon 😸 (completely separate from Ember's own source), supports many Ember versions simultaneously, is focused on providing consistent testing experiences for Ember users, is already a dependency of every application that uses ember-qunit or ember-mocha (roughly all projects), and is a better name 😈.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It is indeed an alternative, although I think it makes more sense to make ember-test-helpers
to be THE PLACE where all "official" testing stuff (that is not qunit-specific) lives.
(and implemented in | ||
[ember-qunit](https://github.com/emberjs/ember-qunit)@3.0.0) will be modified to add the same DOM interaction helpers mentioned above: | ||
|
||
* setup `this.click` helper method |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I still find weird that we have two supported ways of doing the same thing. Also, specially in integration tests, it wouldn't be extraordinary that someone did
this.click = function(e) {
assert.ok(e, 'The event is fired')
}
await render(hbs`{{my-component action=click}}`)
messing with the helper. It could be fixed by making this.click
a readonly property, but it still feels weird.
Do you think that supposed improvement in debuggeablility justifies this duality? Couldn't just the docs serve as ...well... documentation of what helpers exists and how to use them?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
FYI There is one potential advantage of using await this.click()
over await click()
in that you could potentially run tests concurrently. The click()
variant has no reliable way of figuring out which test just called that helper, while this.click()
is bound to the right context.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There is one potential advantage of using await this.click() over await click() in that you could potentially run tests concurrently.
Confirm, however I tried to avoid talking about this in #119 (and here) because there are many other issues with running tests in parallel (there is only one DOM, focus and other events don't fire the same, etc) and I didn't want to open a can of worms.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Even if I slightly favour the imported helpers, my point is about having both. I think it should be one or the other.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do you think that supposed improvement in debuggeablility justifies this duality? Couldn't just the docs serve as ...well... documentation of what helpers exists and how to use them?
Sure, I agree that this is definitely a reasonable way to go also. I'll add a fleshed out version of it to alternatives.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Even if I slightly favour the imported helpers, my point is about having both. I think it should be one or the other.
Right, totally understood. I guess I'm saying that the fact that there are both is an implementation detail. The primary public API and what blueprints will use will be the importable versions. The fact that they desugar to invoking methods on the test context is a minor detail (and might enable features like what @Turbo87 alludes to in #268 (comment)).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Agree. Made the same point here. @rwjblue just added some clarification here, which seems reasonable. I would not see it as a problem to have both, as long as there is one ideomatic way to write your tests, that is consistently used (guides, blueprints, codemods etc.) Was too late here. My comment was refering to Miguel's comment, but Rob already clarified this 😉
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh, y'all did it now. Just removed the this.foo
versions from the RFC. I agree having two different public API's for the same thing is confusing, and reserving a bunch of names on the test context is annoying and error prone for users.
It is still likely that during implementation I will have some mechanism to discover and access these helpers on the test context but it does not need to be public API at all...
No longer publicly expose two systems for the same thing, all in on imports...
@rwjblue I like the idea of moving the functions from I would mention that the summary does a good job of describing motivations, but doesn't really give a top level explanation of what changes are being proposed and assumes that contributors are deeply in the know on testing unification and prior RFCs. |
// **** after **** | ||
import { module, test } from 'qunit'; | ||
import { setupAcceptanceTest } from 'ember-qunit'; | ||
import { visit, fillIn, click } from 'ember-test-helpers'; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
as mentioned in emberjs/ember-test-helpers#240 we might want to move to @ember/test-helpers
for the import paths
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I was trying to avoid this by cross-link to any prior work that I referred to, did I miss some references?
Sounds good, I've tried to do a better job and pushed those changes in b05ed2c. Mind reviewing? |
@rwjblue seems like the main big changes are:
|
No one is suggesting this. Certainly not me. |
I was kidding. But the |
This is awesome. 🎉 |
This was discussed at the Ember core team meeting today, and given that no objections / suggestions / tweaks have come up during the final comment period, and it seems that most folks massively in favor of this improvement we are going to land.... FWIW, I have begun implementation and hope to be able to use this in apps soon... |
Implement `startApplicationTest` from emberjs/rfcs#268.
As of [email protected] (and [email protected]) the testing blueprints are automatically emitting emberjs/rfcs#232 or emberjs/rfcs#268 compatible output. With those, these helpers are no longer used for new apps. Existing apps should only delete these files once they have migrated to the new testing system...
As of [email protected] (and [email protected]) the testing blueprints are automatically emitting emberjs/rfcs#232 or emberjs/rfcs#268 compatible output. With those, these helpers are no longer used for new apps. Existing apps should only delete these files once they have migrated to the new testing system...
As of [email protected] (and [email protected]) the testing blueprints are automatically emitting emberjs/rfcs#232 or emberjs/rfcs#268 compatible output. With those, these helpers are no longer used for new apps. Existing apps should only delete these files once they have migrated to the new testing system...
As of [email protected] (and [email protected]) the testing blueprints are automatically emitting emberjs/rfcs#232 or emberjs/rfcs#268 compatible output. With those, these helpers are no longer used for new apps. Existing apps should only delete these files once they have migrated to the new testing system...
As of [email protected] (and [email protected]) the testing blueprints are automatically emitting emberjs/rfcs#232 or emberjs/rfcs#268 compatible output. With those, these helpers are no longer used for new apps. Existing apps should only delete these files once they have migrated to the new testing system...
As of Ember 3.0 the new testing APIs introduced in emberjs/rfcs#232 and emberjs/rfcs#268 are enabled and used by default. In these APIs no globals are used, therefore this extra `tests/` override is removed. This specifically removes the following globals from being allowed: * `andThen` * `click` * `currentPath` * `currentRouteName` * `currentURL` * `fillIn` * `find` * `findWithAssert` * `keyEvent` * `pauseTest` * `resumeTest` * `triggerEvent` * `visit` * `wait`
As of Ember 3.0 the new testing APIs introduced in emberjs/rfcs#232 and emberjs/rfcs#268 are enabled and used by default. In these APIs no globals are used, therefore this extra `tests/` override is removed. This specifically removes the following globals from being allowed: * `andThen` * `click` * `currentPath` * `currentRouteName` * `currentURL` * `fillIn` * `find` * `findWithAssert` * `keyEvent` * `pauseTest` * `resumeTest` * `triggerEvent` * `visit` * `wait`
As of [email protected] (and [email protected]) the testing blueprints are automatically emitting emberjs/rfcs#232 or emberjs/rfcs#268 compatible output. With those, these helpers are no longer used for new apps. Existing apps should only delete these files once they have migrated to the new testing system...
As of Ember 3.0 the new testing APIs introduced in emberjs/rfcs#232 and emberjs/rfcs#268 are enabled and used by default. In these APIs no globals are used, therefore this extra `tests/` override is removed. This specifically removes the following globals from being allowed: * `andThen` * `click` * `currentPath` * `currentRouteName` * `currentURL` * `fillIn` * `find` * `findWithAssert` * `keyEvent` * `pauseTest` * `resumeTest` * `triggerEvent` * `visit` * `wait`
Continues the testing related RFC's (slowly chipping away at #119) by proposing migrating the acceptance testing API "unify" (though I'm not allowed to use that word in the title any more) with the newly landed and implemented ember-qunit API's (proposed in #232).
rendered