This commit refactors the aliasing system to support multiple different
AliasingHost implementations, which control specific aliasing behavior
in ngtsc (see the README.md).
A new host is introduced, the `PrivateExportAliasingHost`. This solves a
longstanding problem in ngtsc regarding support for "monorepo" style private
libraries. These are libraries which are compiled separately from the main
application, and depended upon through TypeScript path mappings. Such
libraries are frequently not in the Angular Package Format and do not have
entrypoints, but rather make use of deep import style module specifiers.
This can cause issues with ngtsc's ability to import a directive given the
module specifier of its NgModule.
For example, if the application uses a directive `Foo` from such a library
`foo`, the user might write:
```typescript
import {FooModule} from 'foo/module';
```
In this case, foo/module.d.ts is path-mapped into the program. Ordinarily
the compiler would see this as an absolute module specifier, and assume that
the `Foo` directive can be imported from the same specifier. For such non-
APF libraries, this assumption fails. Really `Foo` should be imported from
the file which declares it, but there are two problems with this:
1. The compiler would have to reverse the path mapping in order to determine
a path-mapped path to the file (maybe foo/dir.d.ts).
2. There is no guarantee that the file containing the directive is path-
mapped in the program at all.
The compiler would effectively have to "guess" 'foo/dir' as a module
specifier, which may or may not be accurate depending on how the library and
path mapping are set up.
It's strongly desirable that the compiler not break its current invariant
that the module specifier given by the user for the NgModule is always the
module specifier from which directives/pipes are imported. Thus, for any
given NgModule from a particular module specifier, it must always be
possible to import any directives/pipes from the same specifier, no matter
how it's packaged.
To make this possible, when compiling a file containing an NgModule, ngtsc
will automatically add re-exports for any directives/pipes not yet exported
by the user, with a name of the form: ɵngExportɵModuleNameɵDirectiveName
This has several effects:
1. It guarantees anyone depending on the NgModule will be able to import its
directives/pipes from the same specifier.
2. It maintains a stable name for the exported symbol that is safe to depend
on from code on NPM. Effectively, this private exported name will be a
part of the package's .d.ts API, and cannot be changed in a non-breaking
fashion.
Fixes#29361
FW-1610 #resolve
PR Close#33177
Changed `setValue` documentation for throwing an error as it contained a grammar
mistake and also may have caused ambiguity around when exactly the
method would throw.
PR Close#33126
Static methods that return a type of ModuleWithProviders currently
do not have to specify a type because the generic falls back to any.
This is problematic because the type of the actual module being
returned is not present in the type information.
Since Ivy uses d.ts files exclusively for downstream packages
(rather than metadata.json files, for example), we no longer have
the type of the actual module being created.
For this reason, a generic type should be added for
ModuleWithProviders that specifies the module type. This will be
required for all users in v10, but will only be necessary for
users of Ivy in v9.
PR Close#33217
Previously, the `FileSystem` abstraction featured a `mkdir()` method. In
`NodeJSFileSystem` (the default `FileSystem` implementation used in
actual code), the method behaved similar to Node.js' `fs.mkdirSync()`
(i.e. failing if any parent directory is missing or the directory exists
already). In contrast, `MockFileSystem` (which is the basis or mock
`FileSystem` implementations used in tests) implemented `mkdir()` as an
alias to `ensureDir()`, which behaved more like Node.js'
`fs.mkdirSync()` with the `recursive` option set to `true` (i.e.
creating any missing parent directories and succeeding if the directory
exists already).
This commit fixes this inconsistency by removing the `mkdir()` method,
which was not used anyway and only keeping `ensureDir()` (which is
consistent across our different `FileSystem` implementations).
PR Close#33237
When `ngcc` is running in parallel mode (usually when run from the
command line) and the `createNewEntryPointFormats` option is set to true
(e.g. via the `--create-ivy-entry-points` command line option), it can
happen that two workers end up trying to create the same directory at
the same time. This can lead to a race condition, where both check for
the directory existence, see that the directory does not exist and both
try to create it, with the second failing due the directory's having
already been created by the first one. Note that this only affects
directories and not files, because `ngcc` tasks operate on different
sets of files.
This commit avoids this race condition by allowing `FileSystem`'s
`ensureDir()` method to not fail if one of the directories it is trying
to create already exists (and is indeed a directory). This is fine for
the `ensureDir()` method, since it's purpose is to ensure that the
specified directory exists. So, even if the `mkdir()` call failed
(because the directory exists), `ensureDir()` has still completed its
mission.
Related discussion: https://github.com/angular/angular/pull/33049#issuecomment-540485703
FW-1635 #resolve
PR Close#33237
Often the types of an `@Input`'s field don't fully reflect the types of
assignable values. This can happen when an input has a getter/setter pair
where the getter always returns a narrow type, and the setter coerces a
wider value down to the narrow type.
For example, you could imagine an input of the form:
```typescript
@Input() get value(): string {
return this._value;
}
set value(v: {toString(): string}) {
this._value = v.toString();
}
```
Here, the getter always returns a `string`, but the setter accepts any value
that can be `toString()`'d, and coerces it to a string.
Unfortunately TypeScript does not actually support this syntax, and so
Angular users are forced to type their setters as narrowly as the getters,
even though at runtime the coercion works just fine.
To support these kinds of patterns (e.g. as used by Material), this commit
adds a compiler feature called "input coercion". When a binding is made to
the 'value' input of a directive like MatInput, the compiler will look for a
static function with the name ngCoerceInput_value. If such a function is
found, the type-checking expression for the input will be wrapped in a call
to the function, allowing for the expression of a type conversion between
the binding expression and the value being written to the input's field.
To solve the case above, for example, MatInput might write:
```typescript
class MatInput {
// rest of the directive...
static ngCoerceInput_value(value: {toString(): string}): string {
return null!;
}
}
```
FW-1475 #resolve
PR Close#33243
As a hack to get the Ivy compiler ngtsc off the ground, the existing
'allowEmptyCodegenFiles' option was used to control generation of ngfactory
and ngsummary shims during compilation. This option was selected since it's
enabled in google3 but never enabled in external projects.
As ngtsc is now mature and the role shims play in compilation is now better
understood across the ecosystem, this commit introduces two new compiler
options to control shim generation:
* generateNgFactoryShims controls the generation of .ngfactory shims.
* generateNgSummaryShims controls the generation of .ngsummary shims.
The 'allowEmptyCodegenFiles' option is still honored if either of the above
flags are not set explicitly.
PR Close#33256
Angular v9 schematics should print out a link to the migration
guide associated with each schematic. This way, users have an
easy way to find more information about the automatic code
transformations they will see with `ng update`.
PR Close#33258
With Ivy the `entryComponents` array isn't necessary anymore. These changes mark it as deprecated so that it can be removed in a future version.
PR Close#33205
Currently if a `ModuleWithProviders` is missng its generic type, we throw a cryptic error like:
```
error TS-991010: Value at position 3 in the NgModule.imports of TodosModule is not a reference: [object Object]
```
These changes add a better error to make it easier to debug.
PR Close#33187
Resubmit #31168 now that google3 tests can pass. This requires http://cl/272696717 to be patched.
Original description from jasonaden:
Without this change when using UrlTree redirects in urlUpdateStrategy="eager", the URL would get
updated to the target location, then redirected. This resulted in having an additional entry in the
history and thus the back button would be broken (going back would land on the URL causing a new
redirect).
Additionally, there was a bug where the redirect, even without urlUpdateStrategy="eager", could
create a history with too many entries. This was due to kicking off a new navigation within the
navigation cancelling logic. With this PR the new navigation is pushed to the next tick with a
setTimeout, allowing the page being redirected from to be cancelled before starting a new
navigation.
Related to #27148
fix(router): adjust UrlTree redirect to replace URL if in eager update
Fix lint errors
PR Close#32988
Until now, the template type checker has not checked any of the event
bindings that could be present on an element, for example
```
<my-cmp
(changed)="handleChange($event)"
(click)="handleClick($event)"></my-cmp>
```
has two event bindings: the `change` event corresponding with an
`@Output()` on the `my-cmp` component and the `click` DOM event.
This commit adds functionality to the template type checker in order to
type check both kind of event bindings. This means that the correctness
of the bindings expressions, as well as the type of the `$event`
variable will now be taken into account during template type checking.
Resolves FW-1598
PR Close#33125
In ES5 modules, the class declarations consist of an IIFE with inner
and outer declarations that represent the class. The `EsmReflectionHost`
has logic to ensure that `getDeclarationOfIdentifier()` always returns the
outer declaration.
Before this commit, if an identifier referred to an alias of the inner
declaration, then `getDeclarationOfIdentifier()` was failing to find
the outer declaration - instead returning the inner declaration.
Now the identifier is correctly resolved up to the outer declaration
as expected.
This should fix some of the failing 3rd party packages discussed in
https://github.com/angular/ngcc-validation/issues/57.
PR Close#33252
This commit removes `@angular-devkit/build-angular` from package.json
for a project that opts into Bazel. This is because the package adds a
dependency on node-sass, which is rejected by Bazel due to its absense.
This commit also appends to `scripts.postinstall` if it already exists.
This is needed because `ng new` in CLI v9 now automatically adds a
postinstall step for `ngcc`.
PR Close#32946
BREAKING CHANGE:
In v5, we deprecated support for the intl API in order to improve the browser support. We are now removing these deprecated APIs for v9. See the original change here for more info on why: #18284.
PR Close#29250
This commit fixes ngtsc's import generator to use the ReflectionHost when
looking through the exports of an ES module to find the export of a
particular declaration that's being imported. This is necessary because
some module formats like CommonJS have unusual export mechanics, and the
normal TypeScript ts.TypeChecker does not understand them.
This fixes an issue with ngcc + CommonJS where exports were not being
enumerated correctly.
FW-1630 #resolve
PR Close#33192
Normally, when ngcc encounters a package with missing dependencies while
attempting to determine a compilation ordering, it will ignore that package.
This commit adds a configuration for a flag to tell ngcc to compile the
package anyway, regardless of any missing dependencies.
FW-1931 #resolve
PR Close#33192
In the ReflectionHost API, a 'viaModule' indicates that a particular value
originated in another absolute module. It should always be 'null' for values
originating in relatively-imported modules.
This commit fixes a bug in the CommonJsReflectionHost where viaModule would
be reported even for relatively-imported values, which causes invalid import
statements to be generated during compilation.
A test is added to verify the correct behavior.
FW-1628 #resolve
PR Close#33192
This allows disabling parallelism in ngcc if desired, which is mainly useful
for debugging. The implementation creates the flag and passes its value to
mainNgcc.
No tests are added since the feature mainly exists already - ngcc supports
both parallel and serial execution. This commit only allows switching the
flag via the commandline.
PR Close#33192
Prior to this commit, the absolute spans (relative to template source
file rather than the start of an expression) of expressions in a
template attribute like `*ngIf` were generated incorrectly, equating to
the relative spans.
This fixes the bug by passing an `absoluteOffset` parameter when parsing
template bindings.
Through some levels of indirection, this is required for the Language
Service to support text replacement in
https://github.com/angular/angular/pull/33091.
PR Close#33189
These were getting included in the @angular/localize package.
Instead, patch the upstream files to work with TS typeRoots option
See bazelbuild/rules_nodejs#1033
PR Close#33226
Prior to this fix if a map-based class or style binding wrote
its values onto an elemenent, the internal styling context would
not register the binding if the initial value as a `NO_CHANGE`
value. This situation occurs if a directive takes control of the
`class` or `style` input values and then returns a `NO_CHANGE` value
if the initial value is empty.
This patch ensures that all bindings are always registered with the
`TStylingContext` data-structure even if their initial value is
an instance of `NO_CHANGE`.
PR Close#33236
Prior to this fix, all style/class bindings (e.g. `[style]` and
`[class.foo]`) would quietly update a binding value if and when the
current binding value changes during checkNoChanges.
With this patch, all styling instructions will properly check to see
if the value has changed during the second pass of detectChanges()
if checkNoChanges is active.
PR Close#33103
Prior to this commit, metadata defined on ICU container element was not inherited by the ICU if the whole message is a single ICU (for example: `<ng-container i18n="meaning|description@@id">{count, select, ...}</ng-container>). This commit updates the logic to use parent container i18n meta information for the cases when a message consists of a single ICU.
Fixes#33171
PR Close#33191
LocaleID defs are not considered public API, so the property
that contains them should be prefixed with Angular's marker
for "private" ('ɵ') to discourage apps from relying on def
APIs directly.
This commit adds the prefix and shortens the name from
ngLocaleIdDef to loc. This is because property names
cannot be minified by Uglify without turning on property
mangling (which most apps have turned off) and are thus
size-sensitive.
PR Close#33212
Previously, when `ngcc` was reflecting on class members it did not
account for the fact that a member could be of the kind
`IndexSignature`. This can happen, for example, on abstract classes (as
is the case for [JsonCallbackContext][1]).
Trying to reflect on such members (and failing to recognize their kind),
resulted in warnings, such as:
```
Warning: Unknown member type: "[key: string]: (data: any) => void;
```
While these warnings are harmless, they can be confusing and worrisome
for users.
This commit avoids such warnings by detecting class members of the
`IndexSignature` kind and ignoring them.
[1]: https://github.com/angular/angular/blob/4659cc26e/packages/common/http/src/jsonp.ts#L39
PR Close#33198
Turns out that writing to global state is more expensive than writing to
a property on an object.
Slower:
````
let count = 0;
function increment() {
count++;
}
```
Faster:
````
const state = {
count: 0
};
function increment() {
state.count++;
}
```
This change moves all of the instruction state into a single state object.
`noop_change_detection` benchmark
Pre refactoring: 16.7 us
Post refactoring: 14.523 us (-13.3%)
PR Close#33093
Prior to this commit, all `className` inputs were not set because the runtime code assumed that the `classMap` instruction is only generated for `[class]` bindings. However the `[className]` binding also produces the same `classMap`, thus the code needs to distinguish between `class` and `className`. This commit adds extra logic to select the right input name and also throws an error in case `[class]` and `[className]` bindings are used on the same element simultaneously.
PR Close#33188
Prior to this change, the template type checker would incorrectly bind
non-property bindings such as `[class.strong]`, `[style.color]` and
`[attr.enabled]` to directive inputs of the same name. This is
undesirable, as those bindings are never actually bound to the inputs at
runtime.
Fixes#32099Fixes#32496
Resolves FW-1596
PR Close#33130
This patch introduces the `printTable()` and `printSources()`
methods to `DebugStylingContext` which can be used via the
`window.ng.getDebugNode` helpers when debugging an application.
PR Close#33179
This commit speeds up the tests by calling `MockHost.reset()` in
`beforeEach()` instead of destroying the entire language service and
creating a new one. The creation of a new language service instance is
expensive due to the need to initialize many core Symbols when creating
a new program.
This speeds ups the test (on my local machine) from 35 secs to 15 secs.
PR Close#33200
Alex Rickabaugh
Colum Ferry
Adam Plumer
Matias Niemelä
George Kalpakas
Kara Erickson
crisbeto
Suguru Inatomi
Alison Gale
JoostK
Pete Bacon Darwin
Keen Yee Liau
Igor Minar
Sahan Serasinghe
ayazhafiz
Olivier Combe
vikerman
Alex Eagle
Andrew Kushnir
Miško Hevery