* Removes `interop_deps` from the `ts_project` interop macro.
* Keeps `_rjs` suffix for now as we still need the interop targets for
e.g. `jasmine_node_test` and the `rules_nodejs` linker.
In follow-ups we can remove the suffix, and interop layer.
We should only force the type for files that we know are JavaScript.
Otherwise we risk breaking the magic type detection done by Karma.
The previous code broke for `jasmine.css`.
Fixes https://github.com/angular/angular-cli/issues/29190
This reverts commit 3af88fef8b3255971cc20034d31b01012b7cacb3.
No longer needed because the interop lays out the node modules directory
like without the interop & `ts_project` migration; so the chunk name
changed back to what it was before. Good news.
This commit updates `build_angular` to the `rules_js` ts_project rule.
Notably a few real type issues surfaced but previously didn't surface
due to some unknown resolution issues that resulted in `never` types;
where every possible value was assignable; so this change improves type
safety and a TODO was left for the "brittle code fragment".
Currently the interop resulting target of a `ts_project` ends up not
necessarily working at runtime. This may be the case because a consuming
Node program may end up with mixes of `node_modules` dependencies from
`rules_nodejs` (old) and `rules_js` (new). This sounds fine at first
glance, but in practice can break very subtly because:
* Rules NodeJS leverages the linker, creating `node_module` directories
outside of Bazel, at runtime. These don't depend on symlink resolving.
* Rules JS puts real node module folders via Bazel actions. These rely
on `pnpm` non-hoisting layout, and symlink resolving.
As we can see there is a hard conflict with symlinks. They need to be
enabled with the new toolchain, but the other one doesn't enable symlink
resolution, and enabling is not possible as we'd otherwise risk escaping
the sandbox and cause even more subtle errors.
A good compromise solution is to automatically drop the `rules_js` node
module files/folder in the interop-`rules_nodejs` target and instead
brining in the equivalent `@npm//` dependencies from `rules_nodejs`.
This kind of keeps the logic similar to when not using `rules_js` or the
interop, and enables the simplest & safest mental model; and it works
compared to other solutions I tried with symlinking. Notably, we can't
keep both node module variants as the linker doesn't override existing
node module files from e.g. rules_js then (and would break then).
I suspect there were some versioning changes with the e.g. `hoist =
false` setting in npmrc; so eslint now starts failing about an import
from `webpack-server.ts` resulting in unnecessary type cast lint errors.
The existing mapping didn't work due to the underscore conversion, so
this makes sense, and fixing the path mappings works.
I suspect there were some versioning changes with the e.g. `hoist =
false` setting in npmrc; so eslint now starts failing about this
unnessary type cast. Seems reasonable so this is committed without
deep investigation.
It seems that the chunk deterministic name has changed after
recent node module /lock file changes. It's unclear what specifically is
involved in Webpack's chunk name generation, but the output and all
other tests still look good; which makes this is a rather safe update to
the new chunk name. Consulting with CLI team members explained that this
can happen quite often.
Previously, the `baseHref` option under each locale allowed for generating a unique base href for specific locales. However, users were still required to handle file organization manually, and `baseHref` appeared to be primarily designed for this purpose.
This commit introduces a new `subPath` option, which simplifies the i18n process, particularly in static site generation (SSG) and server-side rendering (SSR). When the `subPath` option is used, the `baseHref` is ignored. Instead, the `subPath` serves as both the base href and the name of the directory containing the localized version of the app.
Below is an example configuration showcasing the use of `subPath`:
```json
"i18n": {
"sourceLocale": {
"code": "en-US",
"subPath": ""
},
"locales": {
"fr-BE": {
"subPath": "fr",
"translation": "src/i18n/messages.fr-BE.xlf"
},
"de-BE": {
"subPath": "de",
"translation": "src/i18n/messages.de-BE.xlf"
}
}
}
```
The following tree structure demonstrates how the `subPath` organizes localized build output:
```
dist/
├── app/
│ └── browser/ # Default locale, accessible at `/`
│ ├── fr/ # Locale for `fr-BE`, accessible at `/fr`
│ └── de/ # Locale for `de-BE`, accessible at `/de`
```
DEPRECATED: The `baseHref` option under `i18n.locales` and `i18n.sourceLocale` in `angular.json` is deprecated in favor of `subPath`.
The `subPath` defines the URL segment for the locale, serving as both the HTML base HREF and the directory name for output. By default, if not specified, `subPath` will use the locale code.
Closes#16997 and closes#28967
Running the SSR dev server when `browser-sync` is not installed would print the error, but then build the browser and server targets, then hang and never return control to the user until they manually Ctrl+C. This change skips building at all if `browser-sync` is not installed, immediately returning control to the user.
This is a simple workaround, but there are two deeper bugs which would benefit from investigation:
1. Figure out why NPM sometimes doesn't install `browser-sync`. It was happening inconsistently for me when running `ng add @angular/ssr`.
2. Figure out why Architect does not cancel/await targets still executing when a builder completes.
This was missed in the original addition of `ngServerMode` and incorrectly caused `withEventReplay` to run browser-specific code on the server and crash.
This commit removes the `RenderMode.AppShell` option. Instead, a new configuration parameter, `{ appShellRoute: 'shell' }`, is introduced to the `provideServerRoutesConfig` method.
```ts
provideServerRoutesConfig(serverRoutes, { appShellRoute: 'shell' })
```
The Karma file watching was racing with the file writes done by the
application builder. Since we already tell Karma when to reun via
`.refeshFiles()`, disabling Karma's own file watcher should make
things more reliable.
This allows removing a weird special-case in the test case and
removes the noisy "File chaned" logs generated by Karma.
Fixes https://github.com/angular/angular-cli/issues/28755
The development server's `hmr` option will now disable both global and component
stylesheet hot replacement if explicitly disabled. These features are enabled by
default for all projects.
Outside of single-run mode, the karma test provider was expected to
watch for changes. In the application builder branch, we only handled
the case of an explicit `--watch` though. This meant that the karma
runner was kept running but didn't see any file changes.
Fixes https://github.com/angular/angular-cli/issues/28730
