* The `TransitionAnimationEngine` had a fallback where it would store classes directly on a node if it doesn't have a `classList`. Presumably this is to support old browsers or if an animation is set on something like `ng-container`. This information was never used for anything since `containsClass` was never called. These changes simplify the logic to just a null check.
* Deprecates the `AnimationDriver.matchesElement` method, because it was only used in one place which can be replaced with `classList.contains`. We can't remove the method completely, because `AnimationDriver` is a public API. We also can't turn it into a method on the base class in order to remove it from the sub-classes, because it can break apps using `noImplicitOverride` while extending `AnimationDriver`.
PR Close#44378
improve the configureTestingModule reset test by also spying on
console.error and making sure that an error is logged to the user
when it should be
PR Close#44377
This commit finishes the removal of View Engine from the codebase, deleting
those pieces of @angular/compiler which were only used for VE.
Co-Authored-By: JoostK <joost.koehoorn@gmail.com>
PR Close#44387
This commit makes several changes to the implementation of `NgForOf` to
reduce its code size in production builds:
1. The tailor-made message for an unsupported differ is fully
tree-shaken in production builds, in favor of the exception from the
differ factory itself.
2. The private `_perViewChange` method was changed into a free-standing
function, to allow its name to be minimized.
3. The need for an intermediate `RecordViewTuple` was avoided by
applying the operation in-place, instead of collecting all insertions
into a buffer first. This is safe as the `_perViewChange` operation
that used to be done on each `RecordViewTuple` is entirely local to
the tuple itself. Hence, it is invariant to execution ordering which
means that the `_perViewChange` can be executed directly during the
`forEachOperation` loop.
PR Close#44315
Fixes an issue where destroy hooks are not set up for `useClass` providers
that rely on `forwardRef` for passing the actual class reference.
Currently the destroy hooks are not captured because forward refs are
not resolved for class providers. In ES2015+ this issue becomes even
more critical if arrow-functions are used for
e.g. `useClass: () => forwardRef(..)`.
This is because arrow functions do not have a prototype, unlike in ES5
where the function assigned to `useClass` would always have a prototype.
In ES2015+ a runtime exception as followed is returned (and doesn't give
any useful indication of the forwardRef issue):
```
TypeError: Cannot read properties of undefined (reading 'ngOnDestroy')
at registerDestroyHooksIfSupported (dist/packages/core/src/render3/di_setup.ts:196:35 <- dist/legacy-test-bundle.spec.js:41508:37)
at resolveProvider (dist/packages/core/src/render3/di_setup.ts:153:9 <- dist/legacy-test-bundle.spec.js:41485:11)
at resolveProvider (dist/packages/core/src/render3/di_setup.ts:70:7 <- dist/legacy-test-bundle.spec.js:41445:9)
at providersResolver (dist/packages/core/src/render3/di_setup.ts:54:5 <- dist/legacy-test-bundle.spec.js:41438:7)
at Object.definition.providersResolver (dist/packages/core/src/render3/features/providers_feature.ts:48:18 <- dist/legacy-test-bundle.spec.js:41600:16)
at resolveDirectives (dist/packages/core/src/render3/instructions/shared.ts:1179:40 <- dist/legacy-test-bundle.spec.js:35611:17)
at elementStartFirstCreatePass (dist/packages/core/src/render3/instructions/element.ts:40:7 <- dist/legacy-test-bundle.spec.js:38432:27)
at ɵɵelementStart (dist/packages/core/src/render3/instructions/element.ts:87:7 <- dist/legacy-test-bundle.spec.js:38453:43)
at ɵɵelement (dist/packages/core/src/render3/instructions/element.ts:180:3 <- dist/legacy-test-bundle.spec.js:38510:5)
at MultipleFormControls_Template (ng:///MultipleFormControls.js:9:9)
```
PR Close#44281
Bundle spec files similar to how it is done within the Angular
Components repo. This should simplify the setup and also speed
up the Saucelab job as only a single spec bundle would need to be
downloaded, compared to having to load hundreds of files through the
Saucelabs tunnel.
Also makes a couple of tests more robust with the emulators/and accounts
for ES2015 test runner changes. The tests should be less reluctant to
such build process changes.
Note for reviewers: Some imports have been simplified here. This work
came from Joey's original WIP for this. It's unclear to me whether this
is still needed, but it sounded like this was necessary for the ESBuild
bundling to work. I have robusted the module resolution plugin though,
so I doubt it's still needed. At the same time though: Not worth
reverting/trying as these changes are nice to have anyway!
Co-Authored-By: Joey Perrott <josephperrott@gmail.com>
Co-Authored-By: Paul Gschwendtner <paulgschwendtner@gmail.com>
PR Close#44281
Cleans up all the tests that had to declare a separate module in order to pass in `entryComponents` when creating components dynamically. This covers almost all of the `entryComponents` usages, except for a few which I've left for now, because they're testing some of the existing leftover `entryComponents` logic in core.
PR Close#44276
This commit removes special functions that were used to run tests in ViewEngine or Ivy only.
Since ViewEngine is deprecated and we no longer run ViewEngine tests on CI, we can cleanup
those special helpers and ViewEngine-only tests.
PR Close#44120
PR Close#44262
Add error handling for pipe with component class name in which it has a problem. Add it in ngDevMode, so that it will be a tree-shake away from the production bundle.
PR Close#44081
This commit removes most tests that were designated as only covering View
Engine code. It also removes tag filters from CI and local commands to run
tests.
In a few cases (such as with the packages/compiler tests), this tag was
improperly applied, and certain test cases have been added back running in
Ivy mode.
This commit also empties `@angular/compiler/testing` as it is no longer
necessary (this is safe since compiler packages are not public API). It can
be deleted in the future.
PR Close#44218
The TestBed APIs to override metadata would crash when the metadata
contained objects with cyclic references. Metadata overrides use the
JSON serialized representation of a value to compare objects, which
throws an error if the value has cyclic references. This commit avoids
the error by replacing multiple occurrences of the same object using
a unique string representation for the object.
Fixes#43948
PR Close#44215
Fix various typos and also improve sentences (by making them more clear
or grammatically correct) present in comments inside the animations package
PR Close#44203
When a partially compiled component or directive is "linked" in JIT mode, the body
of its declaration is evaluated by the JavaScript runtime. If a class is referenced
in a query (e.g. `ViewQuery` or `ContentQuery`) but its definition is later in the
file, then the reference must be wrapped in a `forwardRef()` call.
Previously, query predicates were not wrapped correctly in partial declarations
causing the code to crash at runtime. In AOT mode, this code is never evaluated
but instead transformed as part of the build, so this bug did not become apparent
until Angular Material started running JIT mode tests on its distributable output.
This change fixes this problem by noting when queries are wrapped in `forwardRef()`
calls and ensuring that this gets passed through to partial compilation declarations
and then suitably stripped during linking.
See https://github.com/angular/components/pull/23882 and https://github.com/angular/components/issues/23907
PR Close#44113
This commit removes the dynamic creation of named arrays for internal
runtime storage arrays as they may cause CSP violations in development
mode, when an application's CSP configuration does not include
`unsafe-eval`.
Named arrays for view data can still be enabled in development mode
using the `ngDevMode=namedConstructors` query parameter when loading the
application.
The usage of native class syntax for named arrays does not have the
desired effect when the code is downleveled to ES5. Since ES5 targets
are becoming increasingly more rare this is considered less of a problem
than the CSP violation.
Fixes#43494
PR Close#43587
When a safe method call such as `person?.getName()` is used, the
compiler would generate invalid code if the argument list also contained
a safe method call. For example, the following code:
```
person?.getName(config?.get('title').enabled)
```
would generate
```
let tmp;
ctx.person == null ? null : ctx.person.getName((tmp = tmp) == null ?
null : tmp.enabled)
```
Notice how the call to `config.get('title')` has completely disappeared,
with `(tmp = tmp)` having taken its place.
The issue occurred due to how the argument list would be converted
from expression AST to output AST twice. First, the outer safe method
call would first convert its arguments list. This resulted in a
temporary being allocated for `config.get('title')`, which was stored in
the internal `_resultMap`. Only after the argument list has been
converted would the outer safe method call realize that it should be
guarded by a safe access of `person`, entering the `convertSafeAccess`
procedure to convert itself. This would convert the argument list once
again, but this time the `_resultMap` would already contain the
temporary `tmp` for `config?.get('title')`. Consequently, the safe
method in the argument list would be emitted as `tmp`.
This commit fixes the issue by ensuring that nodes are only converted
once.
Closes#44069
PR Close#44088
The `InjectFlags` argument was defined for both `getOrCreateInjectable()` and `Injector`, but was ignored in `NodeInjector`.
This PR adds support for getting parent token via `NodeInjector.get()`.
close#31776
PR Close#41592
Other parts of the intialization process are responsible for passing the locale id to ivy, so it does not need to happen in the provider factory. E.g. `PlatformRef.bootstrapModule()` and `R3TestBedCompiler.finalize()`.
PR Close#43913
Currently, it's impossible to cache (detach/attach) parent route without caching child routes.
This produces a bug, when navigating from a/b to c, then to a/d, where a route is cached.
On the last navigation, we incorrectly restore a/b route instead of a/d.
This change introduces new behavior: if the route should be detached/attached,
we do so, but we check also child routes recursively.
Fixes#17333
PR Close#43791
Remove the ivy-aot bazel tag from usage as it served a duplicate purpose as ivy-only which is now removed as both are
unneeded with Ivy as the default compiler used.
PR Close#43862
Using the tag "view-engine-only" better describes the expected usage of bazel targets with the test. They can
only be run with view engine.
PR Close#43862
Setting the angular_ivy_enabled environment variable to True will default Bazel builds to use the Ivy
compiler rather than defaulting to ViewEngine.
PR Close#43862
This reverts commit bba0a87055.
The reason for rollback: this change is breaking some targets in Google's codebase when there is no attribute value is displayed (attr.aria-label) when translated.
PR Close#43882
While fully dynamic bound properties (and attributes) cannot be marked for localization, properties that only contain interpolation can.
This commit ensure that attribute bindings that only contain interpolation can also be marked for localization.
Closes#43260
PR Close#43815
* beforePreactivation hook is unused
* The only place that uses afterPreactivation does not use the arguments
Not to say we won't want to provide hooks similar to this in the future,
but the current state is over-engineered for what it's being used for.
PR Close#43804
Fixes a unit test that now fails, because rethrowing errors is enabled by default. The problem is that we're checking the behavior of a component that throws during `ngOnDestroy`, but since the `resetTestingModule` call itself throws, the test is considered as failing.
PR Close#43714
NgModule factories are obsolete with Ivy and most of the operations can already be performed with NgModule classes without retreieving corresponding factories. NgModule factories were useful to generate an `NgModuleRef` (via `NgModuleFactory.create` method). This commit adds a new function to create an instance of the `NgModuleRef` class without having the need to create a factory first.
PR Close#43580
This commit adds a new function called `getNgModuleById` to the public API surface of the framework. The `getNgModuleById` function allows to retrieve loaded NgModule class by id (specified via `@NgModule.id`). The function is a replacement for the `getModuleFactory` function.
DEPRECATION:
The `getModuleFactory` function is deprecated in favor of the `getNgModuleById` one. With Ivy it's possible to work with NgModule classes directly, without retrieving corresponding factories, so the `getNgModuleById` should be used instead.
PR Close#43580
In order to support ESM for the `platform-server` package, we need to
remove two usages of dynamic imports and replace them with their
corresponding/equivalent import statement. This will also allow ESBuild
to recognize this import. Note that we want to keep these imports external, so
we explicitly specify the `externals` option for the `ng_package` rule.
PR Close#43431
For APF v13, terser has been updated to v5, and the prodmode output
has changed from ES2015 to ES2020. This results in some changes in
the symbol extractor test. Here are the two causes of changes:
* As said in the previous commit, Terser drops the initializer for
variables which are set to `undefined`. We have updated the symbol
extractor to always capture such declarations w/o initializer now.
This means that a couple of declarations that a couple of new
symbols are captured now. These previously didn't have an initializer
even without Terser, and just didn't show up before.
* Terser changed its inline mechanism for functions that recursively
call themselves. Such functions are no longer inlined and therefore
show up in goldens now.
PR Close#43431
In 3.6.0 jasmine added a required `pp` parameter to the `jasmineToString()` function
signature. This commit updates our own use of this signature to ensure that compilation
succeeds with the latest jasmine typings.
PR Close#43411
This commit removes the `WrappedValue` public API from `@angular/core`
which was deprecated in Angular 10 without replacement.
BREAKING CHANGE:
The `WrappedValue` class can no longer be imported from `@angular/core`,
which may result in compile errors or failures at runtime if outdated
libraries are used that are still using `WrappedValue`. The usage of
`WrappedValue` should be removed as no replacement is available.
PR Close#43507
This commit fixes an oversight in the JIT compilation of partial factory
declarations, where the literal `'invalid'` was not accounted for
(unlike the AOT linker).
Fixes#43609
PR Close#43619
This commit removes the ability to configure lazy routes using a string
for `loadChildren`, together with the supporting classes to load an
`NgModuleFactory` at runtime.
BREAKING CHANGE:
It is no longer possible to use `Route.loadChildren` using a string
value. The following supporting classes were removed from
`@angular/core`:
- `NgModuleFactoryLoader`
- `SystemJsNgModuleFactoryLoader`
The `@angular/router` package no longer exports these symbols:
- `SpyNgModuleFactoryLoader`
- `DeprecatedLoadChildren`
The signature of the `setupTestingRouter` function from
`@angular/core/testing` has been changed to drop its `NgModuleFactoryLoader`
parameter, as an argument for that parameter can no longer be created.
PR Close#43591
With Ivy it's possible to create a `ComponentFactory` instance based on the generated Component def, thus the `ViewContainerRef.createComponent` can accept a Component instance and create a factory inside, rather than requiring additional complexity of creating a ComponentFactory outside (by invoking `ComponentFactoryResolver`).
This should simplify the API that is used for creating components dynamically.
DEPRECATION:
Angular no longer requires component factories to dynamically create components. The factory-based signature of the `ViewContainerRef.createComponent` function is deprecated in favor of a different signature that allows passing component classes instead.
PR Close#43022
Currently the compiler has three different classes to represent a "call to something":
1. `MethodCall` - `foo.bar()`
2. `SafeMethodCall` - `foo?.bar()`.
3. `FunctionCall` - Any calls that don't fit into the first two classes. E.g. `foo.bar()()`.
There are a few problems with this approach:
1. It is inconistent with the TypeScript AST which only has one node: `CallExpression`.
2. It means that we have to maintain more code, because the various parts of the compiler need to know about three node types.
3. It doesn't allow us to easily implement some new JS features like safe calls (e.g. `foo.bar?.())`).
These changes rework the compiler so that it produces only one node: `Call`. The new node behaves similarly to the TypeScript `CallExpression` whose `receiver` can be any expression.
There was a similar situation in the output AST where we had an `InvokeMethodExpression` and `InvokeFunctionExpression`. I've combined both of them into `InvokeFunctionExpression`.
PR Close#42882
The current behavior of `routerLink` for `null` and `undefined` inputs is to treat
the input the same as `[]`. This creates several unresolvable issues with
correctly disabling the links because `commands = []` does _not_ behave the same
as disabling a link. Instead, it navigates to the current page, but will also
clear any fragment and/or query params.
The new behavior of the `routerLink` input will be to completely disable navigation
for `null` and `undefined` inputs. For HTML Anchor elements, this will also mean
removing the `href` attribute.
Fixes#21457Fixes#13980Fixes#31154
BREAKING CHANGE:
Previously `null` and `undefined` inputs for `routerLink` were
equaivalent to empty string and there was no way to disable the link's
navigation.
In addition, the `href` is changed from a property `HostBinding()` to an
attribute binding (`HostBinding('attr.href')`). The effect of this
change is that `DebugElement.properties['href']` will now return the
`href` value returned by the native element which will be the full URL
rather than the internal value of the `RouterLink` `href` property.
PR Close#43087
The management of `browserUrlTree` currently has several problems with
correctly tracking the actual state of the browser.
This change makes the Router eagerly update the `browserUrlTree` when
handling navigations triggered by browser events (i.e., not 'imperative'). This
is because with those types of navigations, the browser URL bar is
_already_ updated. If we do not update the internal tracking of the
`browserUrlTree`, we will be out of sync with the real URL if the
navigation is rejected.
It would be best if we could remove `browserUrlTree` completely, but doing that
would require a lot more investigation and is blocked by #27059 because
the SpyLocation used in tests does not emulate real browser behavior.
fixes#43101
PR Close#43102
Currently if a view throws an error during creation mode, we mark it as `incompleteFirstPass` so that we can try to recover later. The recovery is only possible inside component views.
The problem is that when this was introduced, I forgot to flip the `firstCreatePass` when an error is thrown which meant that calling `renderView` on the same component again is allowed. It will eventually hit an assertion which can be confusing for the end user. This issue only manifests itself when rendering views "manually" through `ViewContainerRef` (e.g. using `NgIf`).
These changes flip the `firstCreatePass` back to false on errors so that trying to re-render the same view will throw an error which is consistent to the one that broke the view during creation.
Fixes#41383.
PR Close#43005
We can't update the framework to rxjs7 until version 13, because it contains breaking changes, but we can allow users to opt into it since all of our code should be compatible.
These changes expand the allowed version range of rxjs and add an integration test to verify that we don't get compilation errors. Note that we also have a test that runs the AIO examples against rxjs 7 already (#42660).
Fixes#41897.
PR Close#42991
Several new functionalities are possible with this change: the most requested is that callers can now check whether a control has a required validator. Other uses include incrementally changing the validators set without doing an expensive operation to reset all validators.
Closes#13461.
PR Close#42838
Kristiyan Kostadinov