We recently allowed users to have a dynamic input `type` with signal forms, but the logic that infers the value type falls back to `string` even though in theory it can be any of the other types.
These changes expand the inferred type to `string | number | boolean | Date | null` if we detect a dynamic `type` binding.
Currently when the signal debug name transform sees something like `const foo = signal(0);`, it transforms the signal into `signal(0, {...(ngDevMode ? { debugName: 'foo' } : {})})`. After minification this becomes `signal(0, {})` which will allocate memory for the empty object literal.
These changes rework the logic to produce `signal(0, ...(ngDevMode ? [{ debugName: 'foo' }] : []))` which will be fully tree shaken away to `signal(0)`.
Improves the symbol linking logic to handle Angular component selectors (e.g., ngCombobox). It attempts to convert Angular selector patterns to their corresponding class names, improving navigation to Angular API documentation.
Escape @-prefixed template control flow constructs during doc extraction so JSDoc parsing keeps description text intact. Add regression coverage for @for snippets.
Currently when we detect a `field` binding on a native element, we treat it as a built-in native control. This might not be the case if it's a pre-existing `ControlValueAccessor` relying on the CVA interop.
These changes try to detect any CVA-like directive on the element and disable the additional type checking if there are any.
Fixes#65468.
Ignore files that fail fs.exists() rather than throw ENOENT. Some applications deliberately create "broken" symbolic links that are not relevant to compilation (e.g. emacs lock files that link to owner "user@host").
The `type_check_block.ts` file was getting quite large and difficult to navigate. These changes split up the different pieces of functionality into separate files.
Reworks the way we approach type checking of signal forms to be closer to the behavior at runtime. There are a couple of scenarios that we handle:
1. For native controls, we now produce simplified type checking code that looks as follows:
```
var t1 = null! as number | string; // Type depends on the input `type`.
t1 = someField().value();
```
2. For custom controls we generate bindings to the individual inputs, rather than checking conformance against `FormValueControl`/`FormCheckboxControl`. This is closer to the behavior at runtime and it allows us to handle generic directives properly.
The `fullInheritane` flag from the metadata and the `CopyDefinitionFeature` that it controls appear to no longer be used since `fullInheritance` is always false. The feature appears to have been there to support ngcc which was removed some time ago.
Currently the logic for detecting `Field` directives only works if it's imported from `@angular/forms/signals` which doesn't cover our own tests.
These changes make the check more robust.
Adds some logic that won't report the `value` or `checked` inputs as missing when the `Field` directive is present since it will bind to the inputs implicitly.
Switches to checking against `FormValueControl<any>` instead of `FormValueControl<unknown>` when checking whether custom controls conform to the interface.
Fixes#64946.
Currently when a required input is missing, we produce a diagnostic on the entire start tag. This can be really noisy if there are already some attributes on the element.
This change switch to only highlighting the tag name instead.
This adds a new compilation error if someone attempts to put legacy animations and `animate.enter` or `animate.leave` in the same component.
PR Close#64569
Fixes a bug in the missingStructuralDirective diagnostic where structural directives with missing imports were not reported when the element using the structural directive contained other directives
Fixes#64467
co-authored-by: Matt Lewis <npm@mattlewis.me>
PR Close#64579
Fixes a bug in the missingStructuralDirective diagnostic where structural directives with missing imports were not reported when the element using the structural directive contained other directives
Fixes#64467
PR Close#64470
This fixes a performance regression from #63754, which is almost a revert of the
prior performance fix in #57291; the latter was provided as quick fix to address
the severe performance overhead this extended diagnostic used to have, with #57337
as follow-up change to address the false negatives that were introduced in #57291.
That follow-up never landed, though, so this commit is re-applying the changes
from #57337 to fix the performance regression.
Fixes#64403
PR Close#64410
Warns when @defer blocks define unreachable or redundant triggers, such as multiple main triggers, ineffective prefetches, or timer delays not scheduled before rendering.
PR Close#64069
