Context:
As part of the `test_docs_examples_ivy` job, we run 5 concurrent builds
on each VM (each for a different example/project). Additionally, all
example projects share the same `node_modules/` (via a symlink to
`aio/tools/examples/shared/node_modules/`), so all concurrent builds
operate on the same files.
Previously, we pre-ran ngcc with `--properties module` to process the
fesm5 bundles. Since we have switched to es2015 in 661a57d9e, we now
need the esm2015 bundles. As a result, the initial ngcc run is
redundant and ngcc runs again during each build (to process the fesm2015
bundles). Since there are 5 concurrent builds, we often end up with
multiple ngcc instances processing the same package and trying to write
to the same directories at the same time, causing a
`file already exists` error
This commit fixes it by pre-processing the esm2015 bundles, so there is
no need to re-run ngcc during each concurrent build.
Fixes#30577
PR Close#30593
There is an encoding issue with using delta `Δ`, where the browser will attempt to detect the file encoding if the character set is not explicitly declared on a `<script/>` tag, and Chrome will find the `Δ` character and decide it is window-1252 encoding, which misinterprets the `Δ` character to be some other character that is not a valid JS identifier character
So back to the frog eyes we go.
```
__
/ɵɵ\
( -- ) - I am ineffable. I am forever.
_/ \_
/ \ / \
== == ==
```
PR Close#30546
Previously, the processor that excludes certain cli commands
(`filterHiddenCommand`) was being run after the `createSitemap`
processor, resulting in those commands to be present in `sitemap.xml`,
while the actual pages where missing. This also resulted in 404s, when
search engine crawlers tried to index the missing URLs.
This commit fixes it by ensuring that the `filterHiddenCommand`
processor is run before the `createSitemap` processor.
PR Close#30395
This commit also changes the config files and their layout to
(reasonably closely) match what the cli would generate for a new app.
Related Jira issue: [TOOL-815](https://angular-team.atlassian.net/browse/TOOL-815)
PR Close#29926
The `Δ` caused issue with other infrastructure, and we are temporarily
changing it to `ɵɵ`.
This commit also patches ts_api_guardian_test and AIO to understand `ɵɵ`.
PR Close#29850
The `announcements.json` file should not be included in the sitemap and
including it causes an error in Google Search Console (because the
generated URL does not exist).
(This is a follow-up to fbef94a8e.)
PR Close#29754
Now, one can add an `@alias` tag to API docs, which tells dgeni that this
API element (usually a `const`) is really just an alias for some API element
defined elsewhere.
Dgeni will then look up this API element and copy over the properties from
the alias to the current doc.
For example, we would like to privately export an Enum from `@angular/core`
but then publicly export this from `@angular/common`:
**packages/core/private_exports.ts**
```ts
/**
* Description of this document.
*/
export enum ɵSomeEnum { ... }
```
**packages/common/public_api.ts**
```ts
import {ɵSomeEnum} from '@angular/core';
/**
* @alias core/ɵSomeEnum
*/
export const SomeEnum = ɵSomeEnum;
```
In the generated docs there will be a page for `common/SomeEnum`, which
will be rendered as an enum, rather than a const, showing the description
extracted from the `core/ɵSomeEnum`.
---
The implementation of this feature required some refactoring of the other
processing:
1. Previously `ɵ` prefixed exports were not even considered.
2. Due to 1. some processors needed to have guards added to ignore such
private exports (`addMetadataAliases` and `checkContentRules`).
3. The processing of package pages had to be reworked (and split) so that
it picked up the aliased export docs after their alias proeprties had
been copied.
See FW-1207, FW-632, #29249
PR Close#29673
Currently our plan is to skip the publish, docgen, and update steps for this package.
During RC, we'll determine if the breaking change is too difficult for users, in which case we might restore the package for another major.
PR Close#29550
PhantomJS is [not being developed any more][1] and with modern
alternatives (such as Chrome headless) there is no reason to keep it as
a dependency.
[1]: https://github.com/ariya/phantomjs/issues/15344
PR Close#29611
The API changes are due to enabling strict checks in TypeScript (via `strict: true`).
The payload size changes in `polyfills.js` are due to more browser APIs being patched in recent versions (e.g. `fetch`, `customElement v1`).
PR Close#28219
Sometimes, in ESM5 code, aliases to exported variables are used internally
to refer to the exported value. This prevented some analysis from being
able to match up a reference to an export to the actual export itself.
For example in the following code:
```
var HttpClientXsrfModule = /** @class */ (function () {
function HttpClientXsrfModule() {
}
HttpClientXsrfModule_1 = HttpClientXsrfModule;
HttpClientXsrfModule.withOptions = function (options) {
if (options === void 0) { options = {}; }
return {
ngModule: HttpClientXsrfModule_1,
providers: [],
};
};
var HttpClientXsrfModule_1;
HttpClientXsrfModule = HttpClientXsrfModule_1 = tslib_1.__decorate([
NgModule({
providers: [],
})
], HttpClientXsrfModule);
return HttpClientXsrfModule;
}());
```
We were not able to tell that the `ngModule: HttpClientXsrfModule_1` property
assignment was actually meant to refer to the `function HttpClientXrsfModule()`
declaration. This caused the `ModuleWithProviders` processing to fail.
This commit ensures that we can compile typings files using the ESM5
format, so we can now update the examples boilerplate tool so that it
does not need to compile the ESM2015 format at all.
PR Close#29092
Now that we are using package.json properties to indicate which
entry-point format to compile, it turns out that we don't really
need to distinguish between flat and non-flat formats, unless we
are compiling `@angular/core`.
PR Close#29092
You can now specify a list of properties in the package.json that
should be considered (in order) to find the path to the format to compile.
The build marker system has been updated to store the markers in
the package.json rather than an additional external file.
Also instead of tracking the underlying bundle format that was compiled,
it now tracks the package.json property.
BREAKING CHANGE:
The `proertiesToConsider` option replaces the previous `formats` option,
which specified the final bundle format, rather than the property in the
package.json.
If you were using this option to compile only specific bundle formats,
you must now modify your usage to pass in the properties in the package.json
that map to the format that you wish to compile.
In the CLI, the `--formats` is no longer available. Instead use the
`--properties` option.
FW-1120
PR Close#29092
With 63fb6c08cf1d69f912a0a4e9a28846d6e6985d04, the bug that required
us to temporarily disable these two SystemJS JIT tests has been fixed.
Therefore we can re-enable these tests.
PR Close#29083
As a side effect of 09b34bae8655d4251516655c317b150c46cd3653,
we fixed that the docs systemjs examples currently do not run
with Ivy in JIT mode. This now uncovered new failures with the JIT
resource loading. e.g.
```
zone.js:665 Unhandled Promise rejection: Component 'PhoneListComponent' is not resolved:
- templateUrl: ./phone-list.template.html
Did you run and wait for 'resolveComponentResources()'? ; Zone: <root> ; Task: Promise.then ; Value: Error: Component 'PhoneListComponent' is not resolved:
- templateUrl: ./phone-list.template.html
Did you run and wait for 'resolveComponentResources()'?
at Function.get (directive.ts:54)
at getComponentDef (definition.ts:648)
at verifyDeclarationsHaveDefinitions (module.ts:185)
at Array.forEach (<anonymous>)
at verifySemanticsOfNgModuleDef (module.ts:159)
at Function.get (module.ts:132)
at getInjectorDef (defs.ts:181)
at R3Injector.processInjectorType (r3_injector.ts:230)
at eval (r3_injector.ts:114)
at eval (r3_injector.ts:451) Error: Component 'PhoneListComponent' is not resolved:
```
We temporarily disable these two failing SystemJS examples by adding them to the
`fixmeIvyExamples` list.
PR Close#28984
Currently when adding the example boilerplate to all
examples with Ivy enabled, we run Ngcc and transform
all found formats. This potentially slows down the build and
is not necessary as we only need the "fesm5" and "fesm2015" bundles.
PR Close#28984
Angular supports using <style> and <link> tags inline in component
templates, but previously such tags were not implemented within the ngtsc
compiler. This commit introduces that support.
FW-1069 #resolve
PR Close#28997
Previously, `ng-packages-installer` would replace the version ranges for
all dependencies that were peer dependencies of an Angular package with
the version range used in the Angular package. This effectively meant
that the pinned version (from `yarn.lock`) for that dependency was
ignored (even if the pinned version satisfied the new version range).
This commit reduces non-determinism in CI jobs using the locally built
Angular packages by always using pinned versions of dependencies for
Angular package peer dependencies if possible.
For example, assuming the following versions for the RxJS dependency:
- **aio/package.json**: `rxjs: ^6.3.0`
- **aio/yarn.lock**: `rxjs@^6.3.0: 6.3.3`
- **@angular/core#peerDependencies**: `rxjs: ^6.0.0`
...the following versions would be used with `ng-packages-installer`:
- Before this commit:
- **aio/package.json**: `rxjs: ^6.0.0`
- **node_modules/rxjs/**: `6.4.0` (latest version satisfying `^6.0.0`)
- After this commit:
- **aio/package.json**: `rxjs: ^6.3.0`
- **node_modules/rxjs/**: `6.3.3` (because it satisfies `^6.0.0`)
PR Close#28510
`ng-packages-installer` can be used to replace Angular packages with
locally built ones (from `dist/packages-dist/`) along with their peer
dependencies.
Previously, in order to achieve this, `yarn install` was called with the
`--no-lockfile` option, which resulted in installing the latest versions
of all dependencies (including transitive ones) permitted by the
corresponding version ranges in `package.json` files. As a result, newly
released versions would be picked, resulting in unexpected,
non-deterministic breakages in CI.
This commit calls `yarn install` with the `--pure-lockfile` option
instead. As a result, only the Angular packages (for which the locally
built ones are used) and their peer dependencies are unpinned; the
pinned versions from `yarn.lock` are used for all other (direct and
transitive) dependencies.
While this does not eliminate non-determinism across builds, it
significantly reduces it.
PR Close#28510
Since b43f8bc7d, RxJS does not need to be patched any more in the
top-level `node_modules/`, so we don't need to special-case RxJS in
`ng-package-installer` and use `node_modules/rxjs/`.
PR Close#28510
By default, `webdriver-manager update` will download the latest
ChromeDriver version, which might not be compatible with the Chrome
version included in the [docker image used on CI], causing CI failures.
Previously, we used to pin the ChromeDriver version on CI in
[ngcontainer's Dockerfile][2]. This was accidentally broken in #26691,
while moving from ngcontainer to default CircleCI docker images.
This commit fixes the issue by pinning ChromeDriver to a known
compatible version.
[1]: bfd48d156d/.circleci/config.yml (L16)
[2]: bfd48d156d/tools/ngcontainer/Dockerfile (L63)
PR Close#28494
Previously, cli-based docs examples were tested using `yarn e2e ...`. In
some cases, it might make sense to run different or additional checks
for a docs example (when running `yarn example-e2e` in `aio/`).
Currently, the only option is to define a custom project type and
overwrite the `e2e` yarn script in `package.json`. Doing so (in addition
to being cumbersome and verbose) would also end up in the `.zip` archive
that users can download to run the example locally. This would be
confusing, if these custom tests are specific to our CI needs.
This commit adds support for defining a custom list of commands per
example. These commands (if specified) would be run instead of the
default `yarn e2e ...`, when testing the docs examples on CI (via
`yarn example-e2e`).
(This feature will be used to verify that the
`service-worker-getting-started` example is set up correctly in a
subsequent commit, but can be useful in other cases as well.)
PR Close#28020
File overwrites:
- **angular.json**: Add `serviceWorker: true` to production config.
- **package.json**: Add `@angular/service-worker` to dependencies.
This will make any `service-worker` examples work out-of-the-box, when
downloading and running locally from the `.zip` archives.
PR Close#28020
Some engineers were already on Yarn 0.10.x which was permitted by the range in our package.json#engines
However this introduced 'integrity sha512' lines into the yarn.lock files.
Then when engineers use yarn 0.9 (in particular, Bazel did this) then the lock files get tons of meaningless edits.
We could force everyone back to yarn 0.9 but this commit chooses to instead advance everyone past 0.10
PR Close#27193
In some overloads, the parameter type can be a large anonymous
object type.
This change displays such types as `object`. It is then up to the
documentation author to put more information about the type in the
method usage notes.
PR Close#24976
* Make individual overloads collapsible
* Show only the first overload expanded, rest collapsed
* Text changes to 'collapse all' once 'show all' is clicked
* Fix chevron/carrot rotation animation when overloads / overload item is expanded or collapsed
PR Close#24976
Some of the text in CLI API docs were being auto-linked to API
pages. This was not correct, and in fact these blocks should not
have any auto links to Angular API at all.
Closes#26570
PR Close#26675
This commit includes the following changes:
* CLI version information is read from the CLI package from which
we read the help files.
* CLI API pages now contain GH links
* line numbers are not shown in GH links, if the doc does not
have a truthy `startingLine` value. This allows us to remove
hard coded checks for `guide` pages
* content pages and CLI api docs no longer have a `startingLine`
* the hard-coded `packages` path segment has been removed from
the templates; instead we now only use the `realProjectRelativePath`.
* the `realProjectRelativePath` has been updated accordingly for API
and CLI API docs.
PR Close#26515
Since #26396, the `title` property is ignored and `header` should be
used instead for specifying a code snippet's header.
This commit ensures that we don't accidentally set `title` have it be
silently ignored.
PR Close#26514
This commit also removes the extra jasminewd2 typings, since the changes
have been merged in the official typings with
DefinitelyTyped/DefinitelyTyped#28957.
PR Close#26139
If the documentation contains a `@selectors` tag then the content of that
is used to describe the selectors of a directive.
Otherwise the selector string is split and each selector is listed as
a list item in an unordered list.
PR Close#25768
The default dgeni config is to concatenate leading comments in front of API items.
In the case that you have an API item that starts a file with no import statements,
the license comment at the top of the file was being added to the front of the
API item's comment. SInce the license comment includes the `@license` tag
and the API item's comment did not start with `@description` the content of
the API item's comment was being put inside the `@license` tag, and no
description was being extracted from the API item's comment.
This commit updates to a version of dgeni-packages that has a switch to turn off
this concatenation, and then also configures this switch.
Closes#26045
PR Close#26050
All directives and pipes must now be tagged with one ore more
public NgModule, from which they are exported.
If an item is exported transitively via a re-exported internal NgModule
then it may be that the item appears to be exported from more than
one public NgModule. For example, there are shared directives that
are exported in this way from `FormsModule` and `ReactiveFormsModule`.
The doc-gen will error and fail if a directive or pipe is not tagged correctly.
NgModule pages now list all the directives and pipes that are exported from it.
Directive and Pipe pages now list any NgModule from which they are exported.
Packages also now list any NgModules that are contained - previously they were
missed.
PR Close#25734
In an overloaded method, the overload with the function body is the
actual method doc, and this doc is not included in the list of "additional"
overloads.
Moreover, the logic (all in dgeni-packages) is that if none of the items
has a body then we use the first overload as the actual method doc.
In the case of abstract methods, none of the methods have a body. So we
have a situation where the overloads collection does not contain the first
abstract method, even though it is not the "implementation" of the method.
Therefore we need to still render it.
Closes#25610
PR Close#25670
It seems that occasionally the sharding of docs examples e2e tests gets
messed up resulting in some tests not being run. This can cause CI to be
green on a PR, when they shouldn't (because the failing tests didn't run
at all).
It is unclear under what circumstances this happens, so printing the
paths of found e2e specs will help debug the issue when it comes up
again.
PR Close#25293
Brandon
George Kalpakas
Ben Lesh
jenniferfell
Jimmy Kasprzak
Santosh Yadav
JiaLiPassion
arjunyel
Dustin M. Eastway
Filipe Silva
Alex Rickabaugh
Pete Bacon Darwin
Alex Eagle
Judy Bogart
Greg Magolan
Paul Gschwendtner
Adam Plumer
Marc Laval
Kapunahele Wong
jithil-kore
Alan Agius
WilliamKoza
Alex Eagle
Stefanie Fluin
William Marques
Martin Sikora