Currently, the `ng_module` rule incorrectly uses manifest paths for
generated imports from the Angular compiler.
This breaks packaging as prodmode output (i.e. `esnext`) is copied in
various targets (`es5` and `es2015`) to the npm package output.
e.g. imports are generated like:
_node_modules/my-pkg/es2015/imports/public-api.js_
```ts
import * as i1 from "angular/packages/bazel/test/ng_package/example/imports/second";
```
while it should be actually:
```ts
import * as i1 from "./second";
```
The imports can, and should be relative so that the files are
self-contained and do not rely on custom module resolution.
PR Close#35841
The options for `flatModuleId` and `flatModuleOutFile` had been removed in the CLI
from generated libraries with 718ee15b9a.
This has been done because `ng-packagr` (which is used to build the
libraries) automatically set these options in-memory when it compiles the library.
No migration has been created for this because there was no actual need to get rid of
this. Keeping the options in the library `tsconfig` does not cause any problems unless
the `tsconfig` is used outside of `ng-packagr`. This was not anticipated, but is now
commonly done in `ng update` migrations.
The `ng update` migrations try to create an instance of the `AngularCompilerProgram` by
simply parsing the `tsconfig`. The migrations make the valid assumption that `tsconfig` files
are not incomplete/invalid. They _definitely_ are in the file system though. It just works for
libraries because `ng-packagr` in-memory completes the invalid `tsconfig` files, so that they
can be passed to the `@angular/compiler-cli`.
We can't have this logic in the `ng update` migrations because it's
out-of-scope for individual migrations to distinguish between libraries
and applications. Also it would be out-of-scope to parse the
`ng-packagr` configuration and handle the tsconfig in-memory completion.
As a workaround though, we can remove the flat-module bundle options
in-memory when creating the compiler program. This is acceptable since
we don't emit the program and the flat module bundles are not needed.
Fixes#34985.
PR Close#35824
This version of `LockFile` creates an "unlocker" child-process that monitors
the main ngcc process and deletes the lock file if it exits unexpectedly.
This resolves the issue where the main process could not be killed by pressing
Ctrl-C at the terminal.
Fixes#35761
PR Close#35861
The previous implementation mixed up the management
of locking a piece of code (both sync and async) with the
management of writing and removing the lockFile that is
used as the flag for which process has locked the code.
This change splits these two concepts up. Apart from
avoiding the awkward base class it allows the `LockFile`
implementation to be replaced cleanly.
PR Close#35861
This reduces the time that `findEntryPoints` takes from 9701.143ms to 4177.278ms, by reducing the file operations done.
Reference: #35717
PR Close#35756
This reverts commit 00f3c58bb9.
Rolling back because it could be breaking e2e tests that assert that
there are no errors in the console after the assertions have run. We can
re-add this in v10.
PR Close#35845
Changes the Ivy unknown element/property messages from being logged with `console.warn` to `console.error`. This should make them a bit more visible without breaking existing apps. Furthermore, a lot of folks filter out warning messages in the dev tools' console, whereas errors are usually still shown.
Fixes#35699.
PR Close#35798
Using the --silent flag prevents the spammy logging messages in the
bazel execution logs.
```
INFO: From Bundling JavaScript packages/zone.js/dist/zone-rollup.umd.js [rollup]:
bazel-out/k8-fastbuild/bin/packages/zone.js/lib/browser/rollup-legacy-main.mjs → bazel-out/k8-fastbuild/bin/packages/zone.js/dist/zone-rollup.umd.js...
created bazel-out/k8-fastbuild/bin/packages/zone.js/dist/zone-rollup.umd.js in 736ms
```
PR Close#35835
It's an error to declare a variable twice on a specific template:
```html
<div *ngFor="let i of items; let i = index">
</div>
```
This commit introduces a template type-checking error which helps to detect
and diagnose this problem.
Fixes#35186
PR Close#35674
This is a follow up to #35637 which resolved a similar issue for `ComponentFactoryResolver`, but not the root cause. When a `NgModuleRef` is created, it instantiates an `Injector` internally which in turn resolves all of injector types. This can result in a circular call that results in an error, because the module is one of the injector types being resolved.
These changes work around the issue by allowing the constructor to run before resolving the injector types.
Fixes#35677.
Fixes#35639.
PR Close#35731
This patch is a follow-up patch to 35c9f0dc2f.
It changes the `computeStyle` function to handle situations where
non string based values are returned from `window.getComputedStyle`.
This situation usually ocurrs in Node-based test environments where
the element or `window.getComputedStyle` is mocked out.
PR Close#35810
Switches our tslint setup to the standard `tslint.json` linter excludes.
The set of files that need to be linted is specified through a Yarn script.
For IDEs, open files are linted with the closest tslint configuration, if the
tslint IDE extension is set up, and the source file is not excluded.
We cannot use the language service plugin for tslint as we have multiple nested
tsconfig files, and we don't want to add the plugin to each tsconfig. We
could reduce that bloat by just extending from a top-level tsconfig that
defines the language service plugin, but unfortunately the tslint plugin does
not allow the use of tslint configs which are not part of the tsconfig project.
This is problematic since the tslint configuration is at the project root, and we
don't want to copy tslint configurations next to each tsconfig file.
Additionally, linting of `d.ts` files has been re-enabled. This has been
disabled in the past and a TODO has been left. This commit fixes the
lint issues and re-enables linting.
PR Close#35800
The test libs should only be included in one jasmine_node_test
otherwise `bazel build //packages/language-service/...` would
end up running `feature_test` and `infra_test` twice.
PR Close#35816
`ɵɵgetInheritedFactory()` is called from generated code for a component which extends another class. This function is detected by Closure to have a side effect and is not able to tree shake the component as a result. Marking it with `noSideEffects()` tells Closure it can remove this function under the relevant tree shaking conditions.
PR Close#35769
`ɵɵNgOnChangesFeature()` would set `ngInherit`, which is a side effect and also not necessary. This was pulled out to module scope so the function itself can be pure. Since it only curries another function, the call is entirely unnecessary. Updated the compiler to only generate a reference to this function, rather than a call to it, and removed the extra curry indirection.
PR Close#35769
This marks the function are "pure" and eligible to be tree shaken by Closure. Without this, initializing `ngDevMode` is considered a side effect which prevents this function from being tree shaken and also any component which calls it.
PR Close#35769
This is useful for propagating return values without them being converted to a string. It still provides the same guarantees to Closure, which will assume that the function invoked is pure and can be tree-shaken accordingly.
PR Close#35769
This commit performs a modularization of the Language Service's existing
diagnostic messages. Such a modularization has two primary advantages:
- Centralization and decoupling of error messages from the code that
generates them makes it easy to add/delete/edit diagnostic messages,
and allows for independent iteration of diagnostic messages and
diagnostic generation.
- Prepares for additional features like annotating the locations where a
diagnostic is generated and enabling the configuration of which
diagnostics should be reported by the language service.
Although it would be preferable to place the diagnostics registry in an
independent JSON file, for ease of typing diagnostic types as an enum
variant of 'ts.DiagnosticCategory', the registry is stored as an object.
Part of #32663.
PR Close#35678
Before this change ngIvy implementation of queries would throw upon
encountering null / undefined query result collected from an embedded
view. It turns out that we might have a provider that explicitly provides
a null / undefined value in a place of a token queried for.
This commit removes a check from the ngIvy query implementation that was
asserting on a query result to be defined.
Fixes#35673
PR Close#35796
Before this change `[class]` and `[className]` were both converted into `ɵɵclassMap`. The implication of this is that at runtime we could not differentiate between the two and as a result we treated `@Input('class')` and `@Input('className)` as equivalent.
This change makes `[class]` and `[className]` distinct. The implication of this is that `[class]` becomes `ɵɵclassMap` instruction but `[className]` becomes `ɵɵproperty' instruction. This means that `[className]` will no longer participate in styling and will overwrite the DOM `class` value.
Fix#35577
PR Close#35668
Prior to this commit, i18n attributes defined on `<ng-template>` tags were not processed by the compiler. This commit adds the necessary logic to handle i18n attributes in the same way how these attrs are processed for regular elements.
PR Close#35681
Prior to this patch, the `margin` and `padding` properties were not
detected properly by Firefox due to them being shorthand properties.
This patch ensures that both `margin` and `padding` are converted
read as `top right bottom left` in the event that the shorthand
property detection fails for auto-styling in Angular animations.
Fix#35463 (FW-1886)
PR Close#35701
With this change we spawn workers lazily based on the amount of work that needs to be done.
Before this change we spawned the maximum of workers possible. However, in some cases there are less tasks than the max number of workers which resulted in created unnecessary workers
Reference: #35717
PR Close#35719
The `ng_package` rule currently creates incorrect UMD module exports
if an entry-point has a module name with numbers included.
For example, consider an entry-point called `@angular/cdk/a11y`. The UMD
module name should be `ng.cdk.a11y`. Instead, `ng_package` currently generates
an UMD module export called `ng.cdk.a11Y`.
This is because the logic for converting dash-case to camel case is
invalid as it uses Starlark's `title()` method. The title method
converts text to title case while we actually just want to capitalize
the first letter of a dash-case segment.
Fixes angular/components#18652.
PR Close#35792
Adds a new entry-point to the `@angular/bazel` `ng_package` test that
contains numbers in the name. e.g. `example/a11y`. This test is added
to replicate a bug where the UMD module export for such entry-points
is incorrectly generated. i.e. `example.a11Y` is generated instead of
`example.a11y`.
PR Close#35792
This commit updates the host bindings micro benchmark to run tests with mutliple directives (where each directive contains host bindings). The number of directives is configurable as a constant in the micro benchmark file. This change is needed to have an ability to measure/compare perf in different scenarios.
PR Close#35736
This commit extends the range of tNode types that may have local refs to include `TNodeType.Container` to account for `<ng-template>`s. Original changes in https://github.com/angular/angular/pull/33415 didn't include that type and as a result, an error is thrown at runtime in case an i18n block contains an `<ng-template>` with local refs.
PR Close#35758
When the `NgIf` directive is used in a template, its context variables
can be used to capture the bound value. This is typically used together
with a pipe or function call, where the resulting value is captured in a
context variable. There's two syntax forms available:
1. Binding to `NgIfContext.ngIf` using the `as` syntax:
```html
<span *ngIf="(user$ | async) as user">{{user.name}}</span>
```
2. Binding to `NgIfContext.$implicit` using the `let` syntax:
```html
<span *ngIf="user$ | async; let user">{{user.name}}</span>
```
Because of the semantics of `ngIf`, it is known that the captured
context variable is non-nullable, however the template type checker
would not consider them as such and still report errors when
`strictNullTypes` is enabled.
This commit updates `NgIf`'s context guard to make the types of the
context variables non-nullable, avoiding the issue.
Fixes#34572
PR Close#35125
This commit removes the `NullAstVisitor` and `visitAstChildren` exported
from `packages/compiler/src/expression_parser/ast.ts` because they
contain duplicate and buggy implementation, and their use cases could be
sufficiently covered by `RecursiveAstVisitor` if the latter implements the
`visit` method. This use case is only needed in the language service.
With this change, any visitor that extends `RecursiveAstVisitor` could
just define their own `visit` function and the parent class will behave
correctly.
A bit of historical context:
In language service, we need a way to tranverse the expression AST in a
selective manner based on where the user's cursor is. This means we need a
"filtering" function to decide which node to visit and which node to not
visit. Instead of refactoring `RecursiveAstVisitor` to support this,
`visitAstChildren` was created. `visitAstChildren` duplicates the
implementation of `RecursiveAstVisitor`, but introduced some bugs along
the way. For example, in `visitKeyedWrite`, it visits
```
obj -> key -> obj
```
instead of
```
obj -> key -> value
```
Moreover, because of the following line
```
visitor.visit && visitor.visit(ast, context) || ast.visit(visitor, context);
```
`visitAstChildren` visits every node *twice*.
PR Close#35619
This commit differentiates language service feature and language service
infrastructure tests. This is primarily to make testing of different
components at the development level easier. This commit continues a
small effort to expand our test coverage and normalize testing
structure.
Also adds test coverage to language service tests. We have quite a bit
to go on that front 🙂.
PR Close#35688
The `packages/localize/src/tools` folder was excluded
from the top level `tsconfig.json` which meant that in IDEs
these source files were not being given the correct configuration.
It was originally excluded because it required the native `node` typings
but this is no longer a requirement.
Removing this folder from the exclusion list exposed a new issue
where there was a typings mismatch between `@babel/...` sources
and the associated `@types/babel__...` typings packages.
A clean up of the package.json and yarn.lock appears to fix this.
PR Close#35711
For view and content queries, the Ivy compiler attempts to statically
evaluate the predicate token so that string predicates containing
comma-separated reference names can be split into an array of strings
during compilation. When the predicate is a dynamic value that cannot be
statically interpreted at compile time, the compiler would previously
produce an error. This behavior breaks a use-case where an `InjectionToken`
is being used as query predicate, as the usage of the `new` keyword
prevents such predicates from being statically evaluated.
This commit changes the behavior to no longer produce an error for
dynamic values. Instead, the expression is emitted as is into the
generated code, postponing the evaluation to happen at runtime.
Fixes#34267
Resolves FW-1828
PR Close#35307
Prior to this change, the logic that compiles Injectable in JIT mode used incorrect configuration that triggers a problem when `ChangeDetectorRef` is used as a dependency. This commit updates the logic to generate correct inject instruction to add the `ChangeDetectorRef` dependency in case it's requested in @Injectable class.
PR Close#35706
This commit adds micro benchmark for host bindings, so that we can assess the impact of changes related to host bindings (for example PR #35568).
PR Close#35705
Source-maps in the wild could be badly formatted,
causing the source-map flattening processing to fail
unexpectedly. Rather than causing the whole of ngcc
to crash, we gracefully fallback to just returning the
generated source-map instead.
PR Close#35718
Previously when rendering flattened source-maps, it was assumed that no
mapping would come from a line that is outside the lines of the actual
source content. It turns out this is not a valid assumption.
Now the code that renders flattened source-maps will handle such
mappings, with the additional benefit that the rendered source-map
will only contain mapping lines up to the last mapping, rather than a
mapping line for every content line.
Fixes#35709
PR Close#35718
If a package has a source-map but it does not provide
the actual content of the sources, then the source-map
flattening was crashing.
Now we ignore such mappings that have no source
since we are not able to compute the merged
mapping if there is no source file.
Fixes#35709
PR Close#35718
This commit adds a new ngcc configuration, `ignorableDeepImportMatchers`
for packages. This is a list of regular expressions matching deep imports
that can be safely ignored from that package. Deep imports that are not
ignored cause a warning to be logged.
// FW-1892
Fixes#35615
PR Close#35683
If an injectable has a `useClass`, Ivy injects the token in `useClass`, rather than the original injectable, if the injectable is re-provided under a different token. The correct behavior is that it should inject the re-provided token, no matter whether it has `useClass`.
Fixes#34110.
PR Close#34574
ɵAnimationDriver can be safely removed from private exports as AnimationDriver
is already a public export. Since its already available, we can safely remove
its declaration and migrate its only usage in our repo to rely on the public
AnimationDriver symbol.
PR Close#35690
Template type-checking within the Ivy compiler has been disabled internally
in g3 until compatibility with the whole codebase could be verified. As that
verification is now complete and template type-checking is known to be
compatible with g3, this commit enables it.
PR Close#35672
This commit normalizes hover and util tests in the language service.
This is part of a small effort to simplify and normalize the language
service testing structure, which currently contains specs that are
largely created and left without relation to other tests.
PR Close#35656
It's possible to pass a directive as an input to itself. Consider:
```html
<some-cmp #ref [value]="ref">
```
Since the template type-checker attempts to infer a type for `<some-cmp>`
using the values of its inputs, this creates a circular reference where the
type of the `value` input is used in its own inference:
```typescript
var _t0 = SomeCmp.ngTypeCtor({value: _t0});
```
Obviously, this doesn't work. To resolve this, the template type-checker
used to generate a `null!` expression when a reference would otherwise be
circular:
```typescript
var _t0 = SomeCmp.ngTypeCtor({value: null!});
```
This effectively asks TypeScript to infer a value for this context, and
works well to resolve this simple cycle. However, if the template
instead tries to use the circular value in a larger expression:
```html
<some-cmp #ref [value]="ref.prop">
```
The checker would generate:
```typescript
var _t0 = SomeCmp.ngTypeCtor({value: (null!).prop});
```
In this case, TypeScript can't figure out any way `null!` could have a
`prop` key, and so it infers `never` as the type. `(never).prop` is thus a
type error.
This commit implements a better fallback pattern for circular references to
directive types like this. Instead of generating a `null!` in place for the
reference, a type is inferred by calling the type constructor again with
`null!` as its input. This infers the widest possible type for the directive
which is then used to break the cycle:
```typescript
var _t0 = SomeCmp.ngTypeCtor(null!);
var _t1 = SomeCmp.ngTypeCtor({value: _t0.prop});
```
This has the desired effect of validating that `.prop` is legal for the
directive type (the type of `#ref`) while also avoiding a cycle.
Fixes#35372Fixes#35603Fixes#35522
PR Close#35622
NG6002/NG6003 are errors produced when an NgModule being compiled has an
imported or exported type which does not have the proper metadata (that is,
it doesn't appear to be an @NgModule, or @Directive, etc. depending on
context).
Previously this error message was a bit sparse. However, Github issues show
that this is the most common error users receive when for whatever reason
ngcc wasn't able to handle one of their libraries, or they just didn't run
it. So this commit changes the error message to offer a bit more useful
context, instructing users differently depending on whether the class in
question is from their own project, from NPM, or from a monorepo-style local
dependency.
PR Close#35620
When binding to `[style]` we correctly sanitized/unwrapped properties but we did not do it for the object itself.
```
@HostBinding("style")
style: SafeStyle = this.sanitizer.bypassSecurityTrustStyle(
"background: red; color: white; display: block;"
);
```
Above code would fail since the `[style]` would not unwrap the `SafeValue` and would treat it as object resulting in incorrect behavior.
Fix#35476 (FW-1875)
PR Close#35564
The library used by ngcc to update the source files (MagicString) is able
to generate a source-map but it is not able to account for any previous
source-map that the input text is already associated with.
There have been various attempts to fix this but none have been very
successful, since it is not a trivial problem to solve.
This commit contains a novel approach that is able to load up a tree of
source-files connected by source-maps and flatten them down into a single
source-map that maps directly from the final generated file to the original
sources referenced by the intermediate source-maps.
PR Close#35132
Currently we resolve the `NgModuleRef.componentFactoryResolver` by going through the injector, but the problem is that `ComponentFactoryResolver` has a dependency on `NgModuleRef`, which means that if the module that's attached to the ref tries to inject `ComponentFactoryResolver` in its constructor, we'll create a circular dependency which throws at runtime.
These changes resolve the issue by creating the `ComponentFactoryResolver` manually ahead of time without going through the injector. We can do this safely, because the only dependency for the resolver is the current module ref which is providing it.
Aside from fixing the issue, another advantage to this approach is that it should reduce the amount of generated JS, because it removes a getter and a provider definitio.
Fixes#35580.
PR Close#35637
Technically, function definitions can live anywhere because they are
hoisted. However, in this case Closure optimizations break when exported
function definitions are referred in another static object that is
exported.
The bad pattern is:
```
exports const obj = {f};
export function f() {...}
```
which turns to the following in Closure's module system:
```
goog.module('m');
exports.obj = {f};
function f() {...}
exports.f = f;
```
which badly optimizes to (note module objects are collapsed)
```
var b = a; var a = function() {...}; // now b is undefined.
```
This is an optimizer bug and should be fixed in Closure, but in the
meantime this change is a noop and will unblock other changes we want to
make.
PR Close#32230
For example, '<div><p string-model~{cursor}></p></div>', when provide the hover info for 'string-model', the 'path.head' is root tag 'div'. Use the parent of 'path.tail' instead.
PR Close#35317
Currently if TestBed detects that TestBed.overrideModule was used for module X, transitive scopes are recalculated recursively for all modules that X imports and previously calculated data (stored in cache) is ignored. This behavior was introduced in https://github.com/angular/angular/pull/33787 to fix stale transitive scopes issue (cache was not updated if module overrides are present).
The perf issue comes from a "diamond" problem, where module X is overridden which imports modules A and B, which both import module C. Under previous logic, module C gets its transitive deps recomputed multiple times, during the recompute for both A and B. For deep graphs and big common/shared modules this can be super costly.
This commit updates the logic to recalculate ransitive scopes for the overridden module, while keeping previously calculated scopes of other modules untouched.
PR Close#35454
Prior to this commit the service worker only treated 504 errors as "effectively offline".
This commit changes the behaviour to treat both 503 (Service Unavailable) and 504 as "offline".
Fixes#35571
PR Close#35595
The only test case for `ngFor` exercises an incorrect usage which causes
two bound attributes to be generated . This commit adds a canonical and
correct usage to show the difference between the two.
PR Close#35671
Now Angular doesn't support add event listeners as passive very easily.
User needs to use `elem.addEventListener('scroll', listener, {passive: true});`
or implements their own EventManagerPlugin to do that.
Angular may finally support new template syntax to support passive event, for now,
this commit introduces a temp solution to allow user to define the passive event names
in zone.js configurations.
User can define a global varibale like this.
```
(window as any)['__zone_symbol__PASSIVE_EVENTS'] = ['scroll'];
```
to let all `scroll` event listeners passive.
PR Close#34503
As a part of the process of setting up Router providers, we use `ApplicationRef` as a dependency while providing `Router` token. The thing is that `ApplicationRef` is actually unused (all referenced were removed in 5a849829c4 (diff-c0baae5e1df628e1a217e8dc38557fcb)), but it's still listed as dependency. This is causing problems in case `Router` is used as a dependency for factory functions provided as `APP_INITIALIZERS` multi-token (causing cyclic dependency). This commit removes unused `ApplicationRef` dependency in `Router`, so it can be used without causing cyclic dependency issue.
PR Close#35642
Previously if there were two path-mapped libraries that are in
different directories but the path of one started with same string
as the path of the other, we would incorrectly return the shorter
path - e.g. `dist/my-lib` and `dist/my-lib-second`. This was because
the list of `basePaths` was searched in ascending alphabetic order and
we were using `startsWith()` to match the path.
Now the `basePaths` are searched in reverse alphabetic order so the
longer path will be matched correctly.
// FW-1873
Fixes#35536
PR Close#35592
* it's tricky to get out of the runfiles tree with `bazel test` as `BUILD_WORKSPACE_DIRECTORY` is not set but I employed a trick to read the `DO_NOT_BUILD_HERE` file that is one level up from `execroot` and that contains the workspace directory. This is experimental and if `bazel test //:test.debug` fails than `bazel run` is still guaranteed to work as `BUILD_WORKSPACE_DIRECTORY` will be set in that context
* test //integration:bazel_test and //integration:bazel-schematics_test exclusively
* run "exclusive" and "manual" bazel-in-bazel integration tests in their own CI job as they take 8m+ to execute
```
//integration:bazel-schematics_test PASSED in 317.2s
//integration:bazel_test PASSED in 167.8s
```
* Skip all integration tests that are now handled by angular_integration_test except the tests that are tracked for payload size; these are:
- cli-hello-world*
- hello_world__closure
* add & pin @babel deps as newer versions of babel break //packages/localize/src/tools/test:test
@babel/core dep had to be pinned to 7.6.4 or else //packages/localize/src/tools/test:test failed. Also //packages/localize uses @babel/generator, @babel/template, @babel/traverse & @babel/types so these deps were added to package.json as they were not being hoisted anymore from @babel/core transitive.
NB: integration/hello_world__systemjs_umd test must run with systemjs 0.20.0
NB: systemjs must be at 0.18.10 for legacy saucelabs job to pass
NB: With Bazel 2.0, the glob for the files to test `"integration/bazel/**"` is empty if integation/bazel is in .bazelignore. This glob worked under these conditions with 1.1.0. I did not bother testing with 1.2.x as not having integration/bazel in .bazelignore is correct.
PR Close#33927
When a pipe inherits its constructor, and as a result its factory, from an injectable in AOT mode, it can end up throwing an error, because the inject implementation hasn't been set yet. These changes ensure that the implementation is set before the pipe's factory is invoked.
Note that this isn't a problem in JIT mode, because the factory inheritance works slightly differently, hence why this test isn't going through `TestBed`.
Fixes#35277.
PR Close#35468
Under View Engine's default (non-fullTemplateTypeCheck) checking, object and
array literals which appear in templates are treated as having type `any`.
This allows a number of patterns which would not otherwise compile, such as
indexing an object literal by a string:
```html
{{ {'a': 1, 'b': 2}[value] }}
```
(where `value` is `string`)
Ivy, meanwhile, has always inferred strong types for object literals, even
in its compatibility mode. This commit fixes the bug, and adds the
`strictLiteralTypes` flag to specifically control this inference. When the
flag is `false` (in compatibility mode), object and array literals receive
the `any` type.
PR Close#35462
In its default compatibility mode, the Ivy template type-checker attempts to
emulate the View Engine default mode as accurately as is possible. This
commit addresses a gap in this compatibility that stems from a View Engine
type-checking bug.
Consider two template expressions:
```html
{{ obj?.field }}
{{ fn()?.field }}
```
and suppose that the type of `obj` and `fn()` are the same - both return
either `null` or an object with a `field` property.
Under View Engine, these type-check differently. The `obj` case will catch
if the object type (when not null) does not have a `field` property, while
the `fn()` case will not. This is due to how View Engine represents safe
navigations:
```typescript
// for the 'obj' case
(obj == null ? null as any : obj.field)
// for the 'fn()' case
let tmp: any;
((tmp = fn()) == null ? null as any : tmp.field)
```
Because View Engine uses the same code generation backend as it does to
produce the runtime code for this expression, it uses a ternary for safe
navigation, with a temporary variable to avoid invoking 'fn()' twice. The
type of this temporary variable is 'any', however, which causes the
`tmp.field` check to be meaningless.
Previously, the Ivy template type-checker in compatibility mode assumed that
`fn()?.field` would always check for the presence of 'field' on the non-null
result of `fn()`. This commit emulates the View Engine bug in Ivy's
compatibility mode, so an 'any' type will be inferred under the same
conditions.
As part of this fix, a new format for safe navigation operations in template
type-checking code is introduced. This is based on the realization that
ternary based narrowing is unnecessary.
For the `fn()` case in strict mode, Ivy now generates:
```typescript
(null as any ? fn()!.field : undefined)
```
This effectively uses the ternary operator as a type "or" operation. The
resulting type will be a union of the type of `fn()!.field` with
`undefined`.
For the `fn()` case in compatibility mode, Ivy now emulates the bug with:
```typescript
(fn() as any).field
```
The cast expression includes the call to `fn()` and allows it to be checked
while still returning a type of `any` from the expression.
For the `obj` case in compatibility mode, Ivy now generates:
```typescript
(obj!.field as any)
```
This cast expression still returns `any` for its type, but will check for
the existence of `field` on the type of `obj!`.
PR Close#35462
I think the bug is introduced in my PR#34847. For example, 'model="{{title}}"', the attribute value cannot be parsed by 'parseTemplateBindings'.
PR Close#35494
In ES5 code, TypeScript requires certain helpers (such as
`__spreadArrays()`) to be able to support ES2015+ features. These
helpers can be either imported from `tslib` (by setting the
`importHelpers` TS compiler option to `true`) or emitted inline (by
setting the `importHelpers` and `noEmitHelpers` TS compiler options to
`false`, which is the default value for both).
Ngtsc's `StaticInterpreter` (which is also used during ngcc processing)
is able to statically evaluate some of these helpers (currently
`__assign()`, `__spread()` and `__spreadArrays()`), as long as
`ReflectionHost#getDefinitionOfFunction()` correctly detects the
declaration of the helper. For this to happen, the left-hand side of the
corresponding call expression (i.e. `__spread(...)` or
`tslib.__spread(...)`) must be evaluated as a function declaration for
`getDefinitionOfFunction()` to be called with.
In the case of imported helpers, the `tslib.__someHelper` expression was
resolved to a function declaration of the form
`export declare function __someHelper(...args: any[][]): any[];`, which
allows `getDefinitionOfFunction()` to correctly map it to a TS helper.
In contrast, in the case of emitted helpers (and regardless of the
module format: `CommonJS`, `ESNext`, `UMD`, etc.)), the `__someHelper`
identifier was resolved to a variable declaration of the form
`var __someHelper = (this && this.__someHelper) || function () { ... }`,
which upon further evaluation was categorized as a `DynamicValue`
(prohibiting further evaluation by the `getDefinitionOfFunction()`).
As a result of the above, emitted TypeScript helpers were not evaluated
in ES5 code.
---
This commit changes the detection of TS helpers to leverage the existing
`KnownFn` feature (previously only used for built-in functions).
`Esm5ReflectionHost` is changed to always return `KnownDeclaration`s for
TS helpers, both imported (`getExportsOfModule()`) as well as emitted
(`getDeclarationOfIdentifier()`).
Similar changes are made to `CommonJsReflectionHost` and
`UmdReflectionHost`.
The `KnownDeclaration`s are then mapped to `KnownFn`s in
`StaticInterpreter`, allowing it to statically evaluate call expressions
involving any kind of TS helpers.
Jira issue: https://angular-team.atlassian.net/browse/FW-1689
PR Close#35191
This is in preparation of using the `KnownFn` type for known TypeScript
helpers (in addition to built-in functions/methods). This will in turn
allow simplifying the detection of both imported and emitted TypeScript
helpers.
PR Close#35191
When statically evaluating CommonJS code it is possible to find that we
are looking for the declaration of an identifier that actually came from
a typings file (rather than a CommonJS file).
Previously, the CommonJS reflection host would always try to use a
CommonJS specific algorithm for finding identifier declarations, but
when the id is actually in a typings file this resulted in the returned
declaration being the containing file of the declaration rather than the
declaration itself.
Now the CommonJS reflection host will check to see if the file
containing the identifier is a typings file and use the appropriate
stategy.
(Note: This is the equivalent of #34356 but for CommonJS.)
PR Close#35191
Paul Gschwendtner
JiaLiPassion
Keen Yee Liau
ivanwonder
Joey Perrott
Pete Bacon Darwin
Matias Niemelä
Alan Agius
Santosh Yadav
Andrew Scott
Yiting Wang
crisbeto
Alex Rickabaugh
Keen Yee Liau
Greg Magolan
Doug Parker
ayazhafiz
Pawel Kozlowski
Miško Hevery
Andrew Kushnir
JoostK
Ayaz Hafiz
KostyaTretyak
Lars Gyrup Brink Nielsen
Zac
Felix Becker
Rado Kirov
Alex Wiese
Alex Eagle
Pusztai Tibor
George Kalpakas