adjust assertion to allow for either undefined or null values #16485
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
(cherry picked from commit a3ba91f)
Previously, any errors thrown during `didInsertElement` would leave the running environment in an invalid state (`env.inTransaction` would be `true` but `this._transaction` would have been nullified). This commit ensures that we _always_ reset `inTransaction` if `Environment.prototype.commit` is called. Thus avoiding an error RE: "calling commit on null"... (cherry picked from commit ab4a2e3)
(cherry picked from commit e99dd89)
`Ember.trySet` is designed to be an error-tolerant form of `Ember.set`
(per its public API documentation). However, in some circumstances an
error is still thrown when setting on a destroyed object.
Prior to this change, the following would properly prevent an error:
```js
let obj = { some: { deep : 'path', isDestroyed: true }};
Ember.trySet(obj, 'some.deep', 'stuff');
```
But the following would throw an error incorrectly:
```js
let obj = { some: 'path', isDestroyed: true };
Ember.trySet(obj, 'some', 'stuff');
```
This fixes the latter case...
(cherry picked from commit 1f4a3bc)
[ci skip]
Fixes emberjs#16180 (cherry picked from commit 37d6bea)
(cherry picked from commit 94f2b2f)
(cherry picked from commit f9e51c9)
Prior to this change, if `Ember.get` (or `this.get`) is used to access a property that is an Ember.Object with an `unknownProperty` that always returns a non-`undefined` value, the internals of `Ember.get` would check if that property is a "descriptor" and (due to the `unknownProperty` never returning `undefined`) the dev time assertion for accessing a proxy's content without using `Ember.get` would be triggered. (cherry picked from commit 4e950a9)
(cherry picked from commit b659214)
The current sanitizer regexp works well when there is a single invalid character, but does not handle the case where multiple invalid chars are present (e.g. with a branch name of `__DESCRIPTOR__-something__something/special`). (cherry picked from commit 2e99d1f)
Fixes an issue with static numbers > 4000 being used in a template. (cherry picked from commit 5bb4b66)
Main changes: * ES6ify a few classes. * Fix issue when calling `.replaceWith` when the current transition is also a `.replaceWith` (previously this would surprisingly result in a `.transitionTo` _not_ a replace). [Full changelog](tildeio/router.js@v2.0.0-beta.1...v2.0.0-beta.2) (cherry picked from commit f470727)
The "symbols" we generate and use internally in Ember are not really `Symbol`s (yet) so they do not get automatically "whitelisted" by the pre-existing `typeof property === 'symbol'` check. This change ensures that any of our "internal symbols" are allowed to be looked for without `.get`. (cherry picked from commit ea2590b)
[ci skip]
Before this PR, having a component with an `{{input}}` with bound type
would make that input only to work the first time it was rendered. This
was because the first param would be shifter from params, and that would
modify the params of the rest of the instances.
Fixes emberjs#16256
(cherry picked from commit 51f09b5)
(cherry picked from commit 9ba002d)
In older Ember versions the default components were generally labelled as `Ember.Component` or `Ember.TextField` via the `NAME_KEY` property on the class itself. During the glimmer-vm upgrade which landed during the 3.1 canary cycle the custom `NAME_KEY` was lost for `Ember.TextField`. This commit removes needless usage of `NAME_KEY` (not needed since our normal `Ember.Object` instance's `.toString` methods check for the factories own `toString` properly), and updates the actual paths to match the new JS modules API locations (thus reducing more "globals" shenanigans). (cherry picked from commit 3462932)
[ci skip]
(cherry picked from commit 5c6de1f)
…ctor code (cherry picked from commit 04b04f3)
…lude source of error (cherry picked from commit 2add803)
Fixes emberjs#16344 (cherry picked from commit dc63a24)
Fixes emberjs#16257 (cherry picked from commit 8862e91)
…ption (cherry picked from commit 6bd74a4)
(cherry picked from commit b008da7)
[ci skip]
In general,Ember's testing system leverages `ajaxSend` and
`ajaxComplete` events to know when `jQuery.ajax` requests are pending
(and it should therefore "pause settledness" of the testing helpers like
`click` / `andThen` / etc).
In the Ember 3.1 cycle Ember changed from using `jQuery.on` to
`document.addEventListener` to subscribe to these events. Unfortunately,
`jQuery.ajax` **does not** trigger listeners added in this way (we would
have had to do `document.onajaxSend = function() { /* stuff here */ }`).
The result of this change is that "normal" acceptance testing helpers
will _no longer_ be able to detect and wait for pending `jQuery.ajax`.
So, for example, you might expect that given the following
```js
await click('#some-selector');
assert.equal(
$('#something-that-should-exist-after-an-ajax-call').text(),
'hi'
);
```
the click helper would wait until `jQuery.ajax` calls were settled
before the assertion, but prior to the changes in this commit
the promise returned by `click` would **not** wait until the
`jQuery.ajax` request was completed and therefore the assertion would
fail.
---
This commit reverts the changes to use `document.addEventListener` back
to using `jQuery.on` directly when `jQuery` is available.
Many bothans died to bring us this information. 😭 😢 :crying_cat:
Paired with @rwjblue to resolve this issue.
(cherry picked from commit 8609228)
Resolve emberjs#14741. (cherry picked from commit dfaa867)
(cherry picked from commit 0b81f52)
* helper-addon * initializer-addon * instance-initializer-addon (cherry picked from commit 5c6eef3)
* helper * initializer * instance-initializer (cherry picked from commit 14d5bb0)
Also fix missing apostrophe in router examples. Fixes emberjs#16394 (cherry picked from commit 428246f)
The fact that this private property was named router was preventing user
from injecting the router service into routes like this:
```js
export default Route.extend({
router: service()
});
```
I consider that the above code is pretty reasonable, and it should be
the private property the one that should be renamed to a clearer `_router`
name that is less likely to collide with user-defined properties or methods.
If you deem this an _intimate_ API we could consider adding a deprecated
alias, but I don't have evidence to consider it such, as `router` is too
common of a term to search for in emberobserver.
(cherry picked from commit 4690f63)
[ci skip]
|
Seems good to me, |
|
@rwjblue i also targeted the release branch because master wouldn't let my app boot for reasons I didn't look into, is that a valid target as well? |
|
Sadly due to somewhat arcane release processes (that are completely my fault) we need to target this PR at master, then cherry-pick (or land separate PRs) into beta and release. |
|
I'm going to re-open this since I kinda screwed up my branches over here |
|
Moved to #16494 |
|
Sorry for the run around @Dhaulagiri, thank you! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Based on conversation with @rwjblue in #16148 he suggested adjusting this assertion to allow
valueto be null. I confirmed locally that this fixes the instances of #16259 we are seeing which we are unable to correct with the upstream Ember Data fix just yet.If this looks good I can adjust the commit message as needed, but ideally this would be a [BUGFIX Release].