This commit fixes an inconsistency where a type check location for an inline
type check block would be interpreted to occur in a type-checking shim instead.
This resulted in a missing template mapping, causing a crash due to an unsafe
non-null assertion operator.
In the prior commit the `TcbLocation` has been extended with an `isShimFile`
field that is now being used to look for the template mapping in the correct
location. Additionally, the non-null assertion operator is refactored such
that a missing template mapping will now ignore the warning instead of crashing
the compiler.
Fixes#45413
PR Close#45478
Extends `TcbPosition` with a field that indicates whether the `tcbPath` is a
type-checking shim file, or an original source file with an inline type check
block.
This field is used in an upcoming commit that fixes an inconsistency with how
inline type check blocks are incorrectly interpreted as a type-checking shim
file instead.
PR Close#45478
Inline type check blocks (TCBs) are emitted into the original source file, but
node positions would still be represented as a `ShimLocation` with a `shimPath`
corresponding with the type-checking shim file. This results in inconsistencies,
as the `positionInShimFile` field of `ShimLocation` would not correspond with
the `shimPath` of that `ShimLocation`.
This commit is a precursor to letting `ShimLocation` also represent the correct
location for inline type check blocks, by renaming the interface to
`TcbLocation`. A followup commit addresses the actual inconsistency.
PR Close#45478
Updates the AIO payload size goldens to reflect the Angular update
affecting both the `aio` and `aio_local` build (due to CLI devkit
changes affecting both goldens, and the Angular update also affecting
the non-local `aio` build / job).
Patch-port of 6e1fce529a
Aside from AIO, the forms integration test size golden also
needs to be updated. A zonejs change caused the 500 bytes limit
to be exceeded. The change landed in `master` without exceeding,
but exceeded in the `13.3.x` branch (unrelated to the Angular
update). Manually confirmed by diffing the bundles that this change
is the culprit.
https://github.com/angular/angular/commit/c08705899f538bd2d26475c94c2832631c16977b
PR Close#45469
The API golden Bazel rule has changed in the shared dev-infra code. Instead
of putting golden API reports into the golden folder as per their original
location in the NPM package, golden reports are now stored based on the
package exports. This makes it more obvious how entry-points are consumable
As part of this change, since the API golden rule now consutls the NPM package
`exports` field, the `localize/tools` entry-point golden file is now generated.
Previously it wasn't generated due to it not having a secondary entry-point
`package.json` file. Such package json files should not be needed anymore
and will be gone in future APF versions.
PR Close#45469
assign different aria labels to the primary nav and the one used for
guides and docs, so that impaired users can more easily distinguish the
two
PR Close#45209
remove redundant main role as pages should always have a
single element with a main role (also remove the role assigne
to the main tag as that is implied)
PR Close#45209
currently the navigation ul used in aio-top-menu has a role of
navigation, but listitems should be owned by list parents
(see more: https://www.w3.org/TR/wai-aria-1.1/#listitem)
so wrap the ul in a nav and remove the role="navigation" from the
ul element to fix such issue
resolves#44562resolves#16938 (the second point)
PR Close#45209
Close#44913
The following case is not handled correctly by `zone.js`.
```
const delayedPromise = new Promise((resolve) => {
setTimeout(resolve, 1, 'timeout');
});
new Promise((resolve) => {
resolve(delayedPromise);
resolve('second call');
}).then(console.log);
```
It should output `timeout`, since the promise is resolved by the
1st resolve, the `second call` should be ignored.
So this is a bug that the original implementation not ensure the
`resolve` is only called once.
PR Close#45283
Uploading payload size is unreliable from within Bazel. This is
because tests might not run for every commit, tests might have
been cached from a pull request RBE-build (causing payload uploading
to be skipped most of the time as every change comes from a PR)
We should disable the uploading as this is a fundamental problem
(good thing to note now) that we can solve with better payload
size tracking that we want to establish as part of dev-infra.
PR Close#45446
The `patchCallbacks` is used for patching the `document.registerElement` and
`customElements.define`. We explicitly wrap the patching code into try-catch since
callbacks may be already patched by other web components frameworks (e.g. LWC), and they
make those properties non-writable. This means that patching callback will throw an error
`cannot assign to read-only property`. See this code as an example:
https://github.com/salesforce/lwc/blob/master/packages/@lwc/engine-core/src/framework/base-bridge-element.ts#L180-L186
We don't want to stop the application rendering if we couldn't patch some
callback, e.g. `attributeChangedCallback`.
PR Close#42546
PR Close#45400
For quite some time now, since we started to use Bazel for integration tests, we
relied on some size tracking logic that did not actually fully work under Bazel.
It was thought that all the necessary CI push/PR information is available to the
Bazel test, but that was not the case. This was now fixed with the recent Rules NodeJS
v5 update where I made sure the `env.sh` variables are actually available before we
write them to the temporary file for the Bazel-access.
This now will unveil an issue because payload size goldens would start being based
on their branch name. e.g. the golden key in `13.3.x` should not be `master` but
`13.3.x`. This makes more sense than `master` as key, but makes things more
cumbersome and ideally we would not store the branch name at all (this is a larger
change though -- not worth now since we might refactor this anyway). For now we will
update the size tracking logic to always use `master` as golden key (like it worked
in the past year(s))
With the environment fix we now (again) start uploading payload size results to Firebase.
This did not work by accident either. The uploading logic is reliant on the CircleCI
commit range which is not working/reliable in upstream branches. This commit
removes this reliance on `COMMIT_RANGE` since it's not strictly necessary and
currently breaking renovate PRs. We can re-enable this when we have a solution with
CircleCI, or a workaround/resolution logic provided in e.g. `ng-dev ci determine-commit-range`.
PR Close#45444
Improves stability of the Windows Bazel CI job by
installing Bazelisk globally.
Also makes the environment helpers more convenient by
evaluating the variable assignments directly, simplifying
some Bash logic.
PR Close#45443
Dedupes the Yarn run steps, avoiding the need to manually keep
this step in sync (e.g. with the timeout -- which is currently missing
for the windows job)
PR Close#45443
Recent changes in `rules_nodejs` caused the test case copy file actions
to be transitioned into the `exec` configuration, resulting in much larger
file paths. These paths break on Windows with the shell argument limit, and
with the path limit, causing errors like:
```
ERROR: C:/users/circleci/ng/packages/compiler-cli/test/compliance/test_cases/BUILD.bazel:9:12: Copying file packages/compiler-cli/test/compliance/test_cases/r3_compiler_compliance/components_and_directives/value_composition/structural_directives_if_directive_def.js failed: (Exit 1): cmd.exe failed: error executing command
cd /d C:/users/circleci/_bazel_circleci/u4uoan2j/execroot/angular
SET PATH=C:\Program Files\Git\usr\bin;C:\Program Files\Git\bin;C:\Windows;C:\Windows\System32;C:\Windows\System32\WindowsPowerShell\v1.0
SET RUNFILES_MANIFEST_ONLY=1
cmd.exe /C bazel-out\x64_windows-opt-exec-2B5CBBC6\bin\packages\compiler-cli\test\compliance\test_cases\test_cases--1973427149-cmd.bat
The system cannot find the path specified
```
https://app.circleci.com/pipelines/github/angular/angular/44038/workflows/4b530cb2-f232-4e1d-b35a-e6e085151d08/jobs/1140017
PR Close#45443
It is totally fine, and expected to use Git Bash for running Bazel
on Windows. In fact this is the most common setup for Bazel on Windows
and it's unrealistic to run without it.
This allows us to remove the old/legacy Powershell setup from CI
which is also quite flaky sometimes and does not reproduce how
Bazel is used on windows-users dev machines.
PR Close#45443
The Angular components repository can only start using Bazel Rules NodeJS v5
when `@angular/bazel` is published with support for it. To work around this
cycle we temporarily disable the unit tests job until we migrated the COMP
repo as well.
PR Close#45443
We updated buildifier and a few warnings became errors now. This commit
cleans up the failing unused variable instances, making the linter happy.
Additionally for the API extractor BUILD file, the package defaults
need to move to satisfy buildifier.
PR Close#45443
As mentioned in previous commits (check them for more details), `@bazel/typescript`
no longer contains `ts_library`-specific code, so we no longer need that dependency.
PR Close#45443
`@bazel/typescript` code moved to `@bazel/concatjs` for the tsc-wrapped code.
Note that this code is likely going to be removed anyway soon when we
move dts bundling from `ng_module` to `ng_package`.
PR Close#45443
Previously `tsc-wrapped` which is the foundation for `ngc-wrapped`, resided
in `@bazel/typescript`. It has been moved to `@bazel/concatjs` in rules_nodejs
so we need to account for that as part of our rules_nodejs v5 update.
PR Close#45443
We must read `Symbol.species` safely because `this` may be anything. For instance, `this`
may be an object without a prototype (created through `Object.create(null)`); thus
`this.constructor` will be undefined. One of the use cases is SystemJS creating
prototype-less objects (modules) via `Object.create(null)`. The SystemJS creates an empty
object and copies promise properties into that object (within the `getOrCreateLoad`
function). The zone.js then checks if the resolved value has the `then` method and invokes
it with the `value` context. Otherwise, this will throw an error: `TypeError: Cannot read
properties of undefined (reading 'Symbol(Symbol.species)')`.
PR Close#45369
Update the `Dockerfile` used to create the preview server to use the
latest stable version of Debian (`bullseye`) and also update package
versions to latest versions.
Also, unpin the versions of installed packages (except for Node.js
related ones) as pinning proved problematic due to many packages
removing old versions from the official repositories.
NOTE:
This change will allow the preview server to be updated on the VM and
take advantage of recent fixes, such as #45349. Currently, the update
fails with the error:
```
E: Version '7.64.0-4+deb10u1' for 'curl' was not found
The command '/bin/sh -c apt-get update -y && apt-get install -y curl=7.64.0-4+deb10u1' returned a non-zero code: 100
```
PR Close#45390
Jasmine checks internally if `process` and `process.on` is defined. Otherwise,
it installs the browser rejection handler through the `global.addEventListener`.
This code may be run in the browser environment where `process` is not defined, and
this will lead to a runtime exception since Webpack 5 removed automatic Node.js polyfills.
PR Close#42260
PR Close#45392
The Bazel NodeJS rules provide two ways of accessing node modules:
* A linker which creates a `node_modules` directory in the execroot/or in the runfiles.
* A patched module resolution where no node modules directory necessarily needs to exist.
The first is the default in `rules_nodejs` and the second is technically the most idiomatic
resolution mechanism in Bazel (as it matches with a runfile resolution library).
The linker is prone to race conditions in persistent workers, or non-sandbox environments (like
windows). This is because the linker for all workers will operate on a shared `execroot` directory
and the same `node_modules` directory is modified all the time / potentially conflicting with other
linker processes from other concurrently-running workers.
We rely on the patched module resolution anyway, but just need to disable the unused linker to avoid
issues like the following:
```
---8<---8<--- Start of log, file at /private/var/tmp/_bazel_splaktar/280f06d55552a0d01f89f0955b5acd78/bazel-workers/worker-8-TypeScriptCompile.log ---8<---8<---
[link_node_modules.js] An error has been reported: [Error: ENOENT: no such file or directory, unlink 'node_modules'] {
errno: -2,
code: 'ENOENT',
syscall: 'unlink',
path: 'node_modules'
} Error: ENOENT: no such file or directory, unlink 'node_modules'
---8<---8<--- End of log ---8<---8<---
INFO: Elapsed time: 12.796s, Critical Path: 5.39s
INFO: 645 processes: 477 internal, 12 darwin-sandbox, 156 worker.
```
PR Close#45393
Move to the CircleCI v2 api as the authentication fails for downloading artifacts using the v1 methods.
CircleCI v2 api now requires authentication to occur view the headers instead of being done in a
query parameter, all of the CircleCI interactions are now performed through one fetchFromCircleCi method
which ensures the token is provided in the headers as expected.
PR Close#45349
Dylan Hunn