This commit adds generation of .ngsummary.js shims alongside .ngfactory.js
shims when generated files are enabled.
Generated .ngsummary shims contain a single, null export for every exported
class with decorators that exists in the original source files. Ivy code
does not depend on summaries, so these exist only as a placeholder to allow
them to be imported and their values passed to old APIs. This preserves
backwards compatibility.
Testing strategy: this commit adds a compiler test to verify the correct
shape and contents of the generated .ngsummary.js files.
PR Close#26495
This commit refactors the shim host to be agnostic to the shims being
generated, and provides an API for generating additional shims besides
the .ngfactory.js. This will be used in a following commit to generate
.ngsummary.js shims.
Testing strategy: this refactor introduces no new behavior, so it's
sufficient that the existing tests for factory shim generation continue
to pass.
PR Close#26495
This simple refactor of the build rules renames the .ngfactory.js shim
generator to 'shims' instead of 'factories', in preparation for adding
.ngsummary.js shim generation.
Testing strategy: this commit does not introduce any new behavior and
merely moves files and symbols around. It's sufficient that the existing
ngtsc tests pass.
PR Close#26495
Originally, the ivy_switch mechanism used Bazel genrules to conditionally
compile one TS file or another depending on whether ngc or ngtsc was the
selected compiler. This was done because we wanted to avoid importing
certain modules (and thus pulling them into the build) if Ivy was on or
off. This mechanism had a major drawback: ivy_switch became a bottleneck
in the import graph, as it both imports from many places in the codebase
and is imported by many modules in the codebase. This frequently resulted
in cyclic imports which caused issues both with TS and Closure compilation.
It turns out ngcc needs both code paths in the bundle to perform the switch
during its operation anyway, so import switching was later abandoned. This
means that there's no real reason why the ivy_switch mechanism needed to
operate at the Bazel level, and for the ivy_switch file to be a bottleneck.
This commit removes the Bazel-level ivy_switch mechanism, and introduces
an additional TypeScript transform in ngtsc (and the pass-through tsc
compiler used for testing JIT) to perform the same operation that ngcc
does, and flip the switch during ngtsc compilation. This allows the
ivy_switch file to be removed, and the individual switches to be located
directly next to their consumers in the codebase, greatly mitigating the
circular import issues and making the mechanism much easier to use.
As part of this commit, the tag for marking switched variables was changed
from __PRE_NGCC__ to __PRE_R3__, since it's no longer just ngcc which
flips these tags. Most variables were renamed from R3_* to SWITCH_* as well,
since they're referenced mostly in render2 code.
Test strategy: existing test coverage is more than sufficient - if this
didn't work correctly it would break the hello world and todo apps.
PR Close#26550
* If all guards return `true`, operator returns `true`
* `false` and `UrlTree` are now both valid returns from a guard
* Both these values wait for higher priority guards to resolve
* Highest priority `false` or `UrlTree` value will be returned
PR Close#26478
Since the SW immediately takes over all clients, it is safe to delete
caches used by older (e.g. beta) `@angular/service-worker` versions to
avoid running into browser storage quota limitations.
PR Close#26319
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
Using Renderer’s setElementAttribute or setElementStyle with a null or undefined value removes the
corresponding attribute or style. The argument type should allow this when using strictNullChecks.
Closes#13686
PR Close#17065
The 'animations' field of @Component metadata should be copied directly
into the ngComponentDef for that component and should not pass through
static resolution.
Previously the animations array was statically resolved and then the
values were translated back when generating ngComponentDef.
PR Close#26322
Previously we only removed assignments to `Class.decorators = [];`
if the array was not empty.
Now we also remove calls to `__decorate([])`, similarly.
PR Close#26236
Previously, classes that were declared via variable declarations,
rather than class declarations, were being excluded from the
parsed classes.
PR Close#26236
The most recent Angular distributions have begun to use __decorate instead of Class.decorators.
This prevents `ngcc` from recognizing the classes and then fails to perform the transform to
ivy format.
Example:
```
var ApplicationModule = /** @class */ (function () {
// Inject ApplicationRef to make it eager...
function ApplicationModule(appRef) {
}
ApplicationModule = __decorate([
NgModule({ providers: APPLICATION_MODULE_PROVIDERS }),
__metadata("design:paramtypes", [ApplicationRef])
], ApplicationModule);
return ApplicationModule;
}());
```
Now `ngcc` recognizes `__decorate([...])` declarations and performs its transform.
See FW-379
PR Close#26236
In some formats variables are declared as `var` or `let` and only
assigned a value later in the code.
The ngtsc resolver still needs to be able to resolve this value,
so the host now provides a `host.getVariableValue(declaration)`
method that can do this resolution based on the format.
The hosts make some assumptions about the layout of the
code, so they may only work in the constrained scenarios that
ngcc expects.
PR Close#26236
While creating FESM files, rollup usually drops all unused symbols.
All *__POST_NGCC__ are unused unless ngcc rewires stuff. To prevent this DCE
we reexport them as private symbols. If ngcc is not used, these symbols will
be dropped when we optimize an application bundle.
PR Close#26071
This commit builds on the NgtscTestEnvironment helper work before and
introduces template_typecheck_spec.ts, which contains compiler tests
for template type-checking.
PR Close#26203
This commit gets ready for the introduction of ngtsc template
type-checking tests by refactoring test environment setup into a
custom helper. This helper will simplify the authoring of future
ngtsc tests.
Ngtsc tests previously returned a numeric error code (a la ngtsc's CLI
interface) if any TypeScript errors occurred. The helper has the
ability to run ngtsc and return the actual array of ts.Diagnostics, which
greatly increases the ability to write clean tests.
PR Close#26203
This commit enables generation and checking of a type checking ts.Program
whenever the fullTemplateTypeCheck flag is enabled in tsconfig.json. It
puts together all the pieces built previously and causes diagnostics to be
emitted whenever type errors are discovered in a template.
Todos:
* map errors back to template HTML
* expand set of type errors covered in generated type-check blocks
PR Close#26203
This commit adds an ngTemplateGuard_ngIf static method to the NgIf
directive and an ngTemplateContextGuard static method to NgFor. The
function of these two static methods is to enable type narrowing
within generated type checking code for consumers of the directives.
PR Close#26203
Before type checking can be turned on in ngtsc, appropriate metadata for
each component and directive must be determined. This commit adds tracking
of the extra metadata in *DefWithMeta types to the selector scope handling,
allowing for later extraction for type-checking purposes.
PR Close#26203
This commit introduces the template type-checking context API, which manages
inlining of type constructors and type-check blocks into ts.SourceFiles.
This API will be used by ngtsc to generate a type-checking ts.Program.
An TypeCheckProgramHost is provided which can wrap a normal ts.CompilerHost
and intercept getSourceFile() calls. This can be used to provide source
files with type check blocks to a ts.Program for type-checking.
PR Close#26203
This commit introduces the main functionality of the type-check compiler:
generation of type check blocks. Type check blocks are blocks of TypeScript
code which can be inlined into source files, and when processed by the
TypeChecker will give information about any typing errors in template
expressions.
PR Close#26203
Template type-checking will make use of expression and statement
translation as well as the ImportManager, so this code needs to
live in a separate build target which can be depended on by both
the main ngtsc transform as well as the template type-checking
mechanism. This refactor introduces a separate build target
for that code.
PR Close#26203
Previously in Ivy, metadata for directives/components/modules/etc was
carried in .d.ts files inside type information encoded on the
DirectiveDef, ComponentDef, NgModuleDef, etc types of Ivy definition
fields. This works well, but has the side effect of complicating Ivy's
runtime code as these extra generic type parameters had to be specified
as <any> throughout the codebase. *DefInternal types were introduced
previously to mitigate this issue, but that's the wrong way to solve
the problem.
This commit returns *Def types to their original form, with no metadata
attached. Instead, new *DefWithMeta types are introduced that alias the
plain definition types and add extra generic parameters. This way the
only code that needs to deal with the extra metadata parameters is the
compiler code that reads and writes them - the existence of this metadata
is transparent to the runtime, as it should be.
PR Close#26203
This commit introduces //packages/compiler-cli/src/ngtsc/typecheck as a
container for template type-checking code, and implements an initial API:
type constructor generation.
Type constructors are static methods on component/directive types with
no runtime implementation. The methods are used during compilation to
enable inference of a component or directive's generic type parameters
from the types of expressions bound to any of their @Inputs. A type
constructor looks like:
class Directive<T> {
someInput: T;
static ngTypeCtor<T>(init: Partial<Pick<Directive<T>, 'someInput'>>): Directive<T>;
}
It can be used to infer a type for T based on the input:
const _dir = Directive.ngTypeCtor({someInput: 'string'}); // Directive<T>
PR Close#26203
This commit introduces the "t2" API, which processes parsed template ASTs
and performs a number of functions such as binding (the process of
semantically interpreting cross-references within the template) and
directive matching. The API is modeled on TypeScript's TypeChecker API,
with oracle methods that give access to collected metadata.
This work is a prerequisite for the upcoming template type-checking
functionality, and will also become the basis for a refactored
TemplateDefinitionBuilder.
PR Close#26203
This commit adds a generic type parameter to the SelectorMatcher
class and its associated response types. This makes the API for
matching selectors and obtaining information about the matched
directives significantly more ergonomic and type-safe.
PR Close#26203
Upcoming implementation work for template type-checking will need to reuse the
code which matches directives inside a template, so this refactor commit moves
the code to a shared location in preparation.
This commit pulls the code needed to match directives against a template node
out of the TemplateDefinitionBuilder into a utility function, in preparation
for template type-checking and other TemplateDefinitionBuilder refactoring.
PR Close#26203
* Pull out `activateRoutes` into new operator
* Add `asyncTap` operator
* Use `asyncTap` operator for router hooks and remove corresponding abstracted operators
* Clean up formatting
* Minor performance improvements
PR Close#25740
This is a major refactor of how the router previously worked. There are a couple major advantages of this refactor, and future work will be built on top of it.
First, we will no longer have multiple navigations running at the same time. Previously, a new navigation wouldn't cause the old navigation to be cancelled and cleaned up. Instead, multiple navigations could be going at once, and we imperatively checked that we were operating on the most current `router.navigationId` as we progressed through the Observable streams. This had some major faults, the biggest of which was async races where an ongoing async action could result in a redirect once the async action completed, but there was no way to guarantee there weren't also other redirects that would be queued up by other async actions. After this refactor, there's a single Observable stream that will get cleaned up each time a new navigation is requested.
Additionally, the individual pieces of routing have been pulled out into their own operators. While this was needed in order to create one continuous stream, it also will allow future improvements to the testing APIs as things such as Guards or Resolvers should now be able to be tested in much more isolation.
* Add the new `router.transitions` observable of the new `NavigationTransition` type to contain the transition information
* Update `router.navigations` to pipe off of `router.transitions`
* Re-write navigation Observable flow to a single configured stream
* Refactor `switchMap` instead of the previous `mergeMap` to ensure new navigations cause a cancellation and cleanup of already running navigations
* Wire in existing error and cancellation logic so cancellation matches previous behavior
PR Close#25740
In some cases, example when the user clears the caches in DevTools but
the SW remains active on another tab and keeps references to the deleted
caches, trying to write to the cache throws errors (e.g.
`Entry was not found`).
When this happens, the SW can no longer work correctly and should enter
a degraded mode allowing requests to be served from the network.
Possibly related:
- https://github.com/GoogleChrome/workbox/issues/792
- https://bugs.chromium.org/p/chromium/issues/detail?id=639034
This commits remedies this situation, by ensuring the SW can enter the
degraded `EXISTING_CLIENTS_ONLY` mode and forward requests to the
network.
PR Close#26042
Properties are not allowed usage notes, and in this case the example
is so simple it didn't warrant moving it to the overall class documentation.
PR Close#26039
* Pull out `activateRoutes` into new operator
* Add `asyncTap` operator
* Use `asyncTap` operator for router hooks and remove corresponding abstracted operators
* Clean up formatting
* Minor performance improvements
PR Close#25740
This is a major refactor of how the router previously worked. There are a couple major advantages of this refactor, and future work will be built on top of it.
First, we will no longer have multiple navigations running at the same time. Previously, a new navigation wouldn't cause the old navigation to be cancelled and cleaned up. Instead, multiple navigations could be going at once, and we imperatively checked that we were operating on the most current `router.navigationId` as we progressed through the Observable streams. This had some major faults, the biggest of which was async races where an ongoing async action could result in a redirect once the async action completed, but there was no way to guarantee there weren't also other redirects that would be queued up by other async actions. After this refactor, there's a single Observable stream that will get cleaned up each time a new navigation is requested.
Additionally, the individual pieces of routing have been pulled out into their own operators. While this was needed in order to create one continuous stream, it also will allow future improvements to the testing APIs as things such as Guards or Resolvers should now be able to be tested in much more isolation.
* Add the new `router.transitions` observable of the new `NavigationTransition` type to contain the transition information
* Update `router.navigations` to pipe off of `router.transitions`
* Re-write navigation Observable flow to a single configured stream
* Refactor `switchMap` instead of the previous `mergeMap` to ensure new navigations cause a cancellation and cleanup of already running navigations
* Wire in existing error and cancellation logic so cancellation matches previous behavior
PR Close#25740
`TypeScript` only supports merging and extending of `compilerOptions`. This is an implementation to support extending and inheriting of `angularCompilerOptions` from multiple files.
Closes: #22684
PR Close#22717
Previously, when you attempted to bootstrap a component that had a
router-outlet using ngsummaries, it would complain that the component
was not provided by any module even if it was. This commit fixes a
mistake (AFAICT) which caused the lookup of the component in the AOT
summaries to fail.
I believe this change is safe. I've run the affected tests within Google
and there were no breakages caused by this change.
PR Close#24892
Create getter methods `getXXXDef` for each definition which
uses `hasOwnProperty` to verify that we don't accidently read form the
parent class.
Fixes: #24011Fixes: #25026
PR Close#25736
Various user code uses 'instanceof' to check whether a particular instance
is a TemplateRef, ElementRef, etc. Ivy needs to work with these checks.
PR Close#25775
Previously, if ngtsc encountered a VariableDeclaration without an
initializer, it would assume that the variable was undefined, and
return that result.
However, for symbols exported from external modules that resolve to
.d.ts files, variable declarations are of the form:
export declare let varName: Type;
This form also lacks an initializer, but indicates the presence of an
importable symbol which can be referenced. This commit changes the
static resolver to understand variable declarations with the 'declare'
keyword and to generate references when it encounters them.
PR Close#25775
The bootstrap property of @NgModule was not previously compiled by
the compiler in AOT or JIT modes (in Ivy). This commit adds support
for bootstrap.
PR Close#25775
Closure requires @nocollapse on Ivy definition static fields in order
to not convert them to standalone constants. However tsickle, the tool
which would ordinarily be responsible for adding @nocollapse, doesn't
properly annotate fields which are added synthetically via transforms.
So this commit adds @nocollapse by applying regular expressions against
code during the final write to disk.
PR Close#25775
This fix is for the issue below when compiling I18N Angular apps using closure.
For certain locales closure converts the input locale id to a different equivalent locale string. For example if the input locale is 'id'(for Indonesia) goog.LOCALE is set to 'in' and the closure locale data is registered only for 'in'. The Angular compiler uses the original input locale string, 'id' to set the LOCALE_ID token and there is a mismatch of locale used to register and locale used when requesting the locale data.
The fix is for the closure-locale.ts code to register the locale data for all equivalent locales names so that it doesn't matter what goog.LOCALE is actually set to.
PR Close#25867
Add following to your `~/.bazelrc`. This will run the build faster locally
(outside of sandbox), but continue running the builds with sandboxing
on CI.
```
build --spawn_strategy=standalone --strategy=ESM5=sandboxed
```
PR Close#25870
`ngcc` adds marker files to each folder that has been
compiled, containing the version of the ngcc used.
When compiling, it will ignore folders that contain these
marker files, as long as the version matches.
PR Close#25557
defineComponent() and friends can return a flyweight EMPTY object for
specific fields when they contain no data. InheritDefinitionFeature
was attempting to write into these flyweight objects, which have been
protected with Object.freeze().
This commit adds detection to InheritDefinitionFeature to identify the
frozen objects.
PR Close#25755
While creating FESM files, rollup usually drops all unused symbols.
All *__POST_NGCC__ are unused unless ngcc rewires stuff. To prevent this DCE
we reexport them as private symbols. If ngcc is not used, these symbols will
be dropped when we optimize an application bundle.
We don't have an infrastructure to test this fix, so I just manually inspected
the bundles before and after to verify that the fix works.
PR Close#25780
Closure compiler requires that the i18n message constants of the form
const MSG_XYZ = goog.getMessage('...');
have names that are unique across an entire compilation, even if the
variables themselves are local to a given module. This means that in
practice these names must be unique in a codebase.
The best way to guarantee this requirement is met is to encode the
relative file name of the file into which the constant is being written
into the constant name itself. This commit implements that solution.
PR Close#25689
TypeScript has a more modern diagnostic emit function which produces
contextually annotated error information, using colors in the console
to indicate where in the code the error occurs.
This commit swiches ngtsc to use this format for diagnostics when
emitting them after a failed compilation.
PR Close#25647
This commit takes the first steps towards ngtsc producing real
TypeScript diagnostics instead of simply throwing errors when
encountering incorrect code.
A new class is introduced, FatalDiagnosticError, which can be thrown by
handlers whenever a condition in the code is encountered which by
necessity prevents the class from being compiled. This error type is
convertable to a ts.Diagnostic which represents the type and source of
the error.
Error codes are introduced for Angular errors, and are prefixed with -99
(so error code 1001 becomes -991001) to distinguish them from other TS
errors.
A function is provided which will read TS diagnostic output and convert
the TS errors to NG errors if they match this negative error code
format.
PR Close#25647
Previously, benchpress would use `console.time()` and
`console.timeEnd()` to measure the start and end of a test in the
performance log. This used to work over navigations - if you called
`console.time(id)` then navigated to a different page, calling
`console.timeEnd(id)` would still insert an event in the performance
log.
As of Chrome 65, this is no longer the case. `console.timeEnd(id)` will
simply not insert an event in the performance log unless
`console.time(id)` was called on the same page. Likewise, using
`performance.measure()` does not work if the starting mark was on a
different page.
This simple workaround uses `performance.mark()` to insert events in the
performance log at the start and end of the test. Benchpress looks for
'-bpstart' and '-bpend' in the name of the performance mark, and
normalizes that to the start and end events expected by PerflogMetric
PR Close#24114
When using ViewEncapsulation.ShadowDom, Angular will not remove the child nodes of the DOM node a root Component is bootstrapped into. This enables developers building Angular Elements to use the `<slot>` element to do native content projection.
PR Close#24861
By pulling in `compiler` into `core` the `compiler` was not
100% tree-shakable and about 8KB of code was retained
when tree-shaken with closure.
PR Close#25531
In tsc 3.0 the check that enables program structure reuse in tryReuseStructureFromOldProgram has changed
and now uses identity comparison on arrays within CompilerOptions. Since we recreate the options
on each incremental compilation, we now fail this check.
After this change the default set of options is reused in between incremental compilations, but we still
allow options to be overriden if needed.
PR Close#25275
One of the tests was creating TestComponent instance _and_ using
global state making this test not predictable. Fixing the test
by making sure that TestComponent is instantiated only once.
PR Close#25632
This fixes a bug in ngtsc where each @Directive was compiled using a
separate ConstantPool. This resulted in two issues:
* Directive constants were not shared across the file
* Extra statements from directive compilation were dropped instead of
added to the file
This commit fixes both issues and adds a test to verify @Directive is
working properly.
PR Close#25620
Workaround was added in https://github.com/angular/angular/pull/25335. It was necessary for .ngfactory & .ngsummary files to have proper AMD module names starting with @angular when building angular downstream from source using Bazel. The underlying issue has been resolved in the compiler and these files now get proper AMD module names without the need for this workaround. The workaround had an unexpected consequence https://github.com/angular/angular-cli/issues/11835 which is fixed by its removal.
PR Close#25604
This commit adds support for enumeration values. An enumeration value
is now a first-class return value of the resolver, which provides both
a Reference to the enum type itself and the name of the value from that
enum. Resolving an enum itself returns a Map<string, EnumValue>.
PR Close#25619
Ivy definitions in .d.ts files often reference the type of a class.
Sometimes, those classes have generic type parameters. When this is
the case, ngtsc needs to emit generic type parameters in the .d.ts
files (usually by passing 'any').
PR Close#25406
Since non-flat module formats (esm2015, esm5) have different structure
than their flat counterparts (and since we are operating on JS files
inside `node_modules/`, we need to configure TS to include deeply nested
JS files (by specifying a sufficiently high `maxNodeModuleJsDepth`).
Remains to be determined if this has any (noticeable) performance
implications.
PR Close#25406
In some code formats (e.g. ES5) methods can actually be function
expressions. For example:
```js
function MyClass() {}
// this static method is declared as a function expression
MyClass.staticMethod = function() { ... };
```
PR Close#25406
ngtsc's static resolver can evaluate function calls where parameters
have default values. In TypeScript code these default values live on the
function definition, but in ES5 code the default values are represented
by statements in the function body.
A new ReflectionHost method getDefinitionOfFunction() abstracts over
this difference, and allows the static reflector to more accurately
evaluate ES5 code.
PR Close#25406
This is needed to let ts_compile_actions take explicit list of srcs and deps to generate tsc actions from another rule. This is no-op for ngc for now.
PR Close#25558
CanLoad now defines UrlSegment[] as a second parameter of the function.
Users can store the initial url segments and refer to them later, e.g. to go
back to the original url after authentication via router.navigate(urlSegments).
Existing code still works as before because the second function parameter
does not have to be defined.
Closes#12411
PR Close#13127
Provides a runtime and compile time switch for ivy including
`ApplicationRef.bootstrapModule`.
This is done by naming the symbols such that `ngcc` (angular
Compatibility compiler) can rename symbols in such a way that running
`ngcc` command will switch the `@angular/core` module from `legacy` to
`ivy` mode.
This is done as follows:
```
const someToken__PRE_NGCC__ = ‘legacy mode’;
const someToken__POST_NGCC__ = ‘ivy mode’;
export someSymbol = someToken__PRE_NGCC__;
```
The `ngcc` will search for any token which ends with `__PRE_NGCC__`
and replace it with `__POST_NGCC__`. This allows the `@angular/core`
package to be rewritten to ivy mode post `ngcc` execution.
PR Close#25238
At the moment `cacheAge` can we undefined when having `Cache-Control` set to `no-cache` due the mapping method in `needToRevalidate`
Closes#25442
PR Close#25408
A small bug caused base factory variable statements for @Component to
not be emitted properly. At the same time as this is fixed, those
statements are now emitted as const.
PR Close#25425
When an Angular decorated class is inherited, it might be the case that
the entire inheritance chain actually has no constructor defined. In
that event, a factory which simply instantiates the type without any
arguments should be used.
PR Close#25425
This turns on generation of ngfactory.js files when compiling in Ivy
mode in g3. They're not turned on for Bazel users as there appears to
be a strange interaction with the way our tests run in Bazel mode.
PR Close#25392
When @angular/core is compiled by ngtsc, a factory file is generated
for ApplicationModule, that is currently invalid because r3_symbols
does not export NgModuleFactory. This change fixes that issue and
ensures the generated ngfactory file for @angular/core is valid.
PR Close#25392
When generating the 'directives:' property of ngComponentDef, ngtsc
needs to be conscious of declaration order. If a directive being
written into the array is declarated after the component currently
being compiled, then the entire directives array needs to be wrapped
in a closure.
This commit fixes ngtsc to pay attention to such ordering issues
within directives arrays.
PR Close#25392
This commit creates an API for factory functions which allows them
to be inherited from one another. To do so, it differentiates between
the factory function as a wrapper for a constructor and the factory
function in ngInjectableDefs which is determined by a default
provider.
The new form is:
factory: (t?) => new (t || SomeType)(inject(Dep1), inject(Dep2))
The 't' parameter allows for constructor inheritance. A subclass with
no declared constructor inherits its constructor from the superclass.
With the 't' parameter, a subclass can call the superclass' factory
function and use it to create an instance of the subclass.
For @Injectables with configured providers, the factory function is
of the form:
factory: (t?) => t ? constructorInject(t) : provider();
where constructorInject(t) creates an instance of 't' using the
naturally declared constructor of the type, and where provider()
creates an instance of the base type using the special declared
provider on @Injectable.
PR Close#25392
Previously, ngtsc used a new ConstantPool for each decorator
compilation. This could result in collisions between constants in the
top-level scope.
Now, ngtsc uses a single ConstantPool for each source file being
compiled, and merges the constant statements into the file after the
import section.
PR Close#25392
The performCompilation() is always called with an undefined oldProgram option (even in watch mode).
This was regression introduced in: 957be960d2
Partial fix, discovered in: #21361
PR Close#21364
Inside of a nested template, an attempt to generate code for a banana-
in-a-box expression would cause a crash in the _AstToIrVisitor, as it
was not handling the case where a write would be generated to a local
variable.
This change supports such a mode of operation.
PR Close#25321
Pete Bacon Darwin
Alex Rickabaugh
Alex Eagle
Olivier Combe
Pawel Kozlowski
Andrew Kushnir
Matias Niemelä
Miško Hevery
Igor Minar
Jason Aden
George Kalpakas
Kara Erickson
Manduro
Jason Bedard
krzysztof-grzybek
JiaLi.Passion
Ash Kyd
Judy Bogart
JoostK
Alan Agius
Ben Lesh
Brandon Roberts
Frederik Schlemmer
Adam Plumer
Alex Eagle
George Wilde
Greg Magolan
Daniel Wiehl
vincecampanale
Matt Keller
cexbrayat
Greg Estren
Vikram Subramanian
Trotyl
Marc Laval
Mike Giambalvo
Rob Wormald
Wei Huang
Marcus Krahl
Victor Berchet
Kieran Whiteman
Martin Probst
Roope Hakulinen
Ádám Lippai