In Angular programs, changing a file may require other files to be
emitted as well due to implicit NgModule dependencies. For example, if
the selector of a directive is changed then all components that have
that directive in their compilation scope need to be recompiled, as the
change of selector may affect the directive matching results.
Until now, the compiler solved this problem using a single dependency
graph. The implicit NgModule dependencies were represented in this
graph, such that a changed file would correctly also cause other files
to be re-emitted. This approach is limited in a few ways:
1. The file dependency graph is used to determine whether it is safe to
reuse the analysis data of an Angular decorated class. This analysis
data is invariant to unrelated changes to the NgModule scope, but
because the single dependency graph also tracked the implicit
NgModule dependencies the compiler had to consider analysis data as
stale far more often than necessary.
2. It is typical for a change to e.g. a directive to not affect its
public API—its selector, inputs, outputs, or exportAs clause—in which
case there is no need to re-emit all declarations in scope, as their
compilation output wouldn't have changed.
This commit implements a mechanism by which the compiler is able to
determine the impact of a change by comparing it to the prior
compilation. To achieve this, a new graph is maintained that tracks all
public API information of all Angular decorated symbols. During an
incremental compilation this information is compared to the information
that was captured in the most recently succeeded compilation. This
determines the exact impact of the changes to the public API, which
is then used to determine which files need to be re-emitted.
Note that the file dependency graph remains, as it is still used to
track the dependencies of analysis data. This graph does no longer track
the implicit NgModule dependencies, which allows for better reuse of
analysis data.
These changes also fix a bug where template type-checking would fail to
incorporate changes made to a transitive base class of a
directive/component. This used to be a problem because transitive base
classes were not recorded as a transitive dependency in the file
dependency graph, such that prior type-check blocks would erroneously
be reused.
This commit also fixes an incorrectness where a change to a declaration
in NgModule `A` would not cause the declarations in NgModules that
import from NgModule `A` to be re-emitted. This was intentionally
incorrect as otherwise the performance of incremental rebuilds would
have been far worse. This is no longer a concern, as the compiler is now
able to only re-emit when actually necessary.
Fixes#34867Fixes#40635Closes#40728
PR Close#40947
Previously, if a trait's analysis step resulted in diagnostics, the trait
would be considered "errored" and no further operations, including register,
would be performed. Effectively, this meant that the compiler would pretend
the class in question was actually undecorated.
However, this behavior is problematic for several reasons:
1. It leads to inaccurate diagnostics being reported downstream.
For example, if a component is put into the error state, for example due to
a template error, the NgModule which declares the component would produce a
diagnostic claiming that the declaration is neither a directive nor a pipe.
This happened because the compiler wouldn't register() the component trait,
so the component would not be recorded as actually being a directive.
2. It can cause incorrect behavior on incremental builds.
This bug is more complex, but the general issue is that if the compiler
fails to associate a component and its module, then incremental builds will
not correctly re-analyze the module when the component's template changes.
Failing to register the component as such is one link in the larger chain of
issues that result in these kinds of issues.
3. It lumps together diagnostics produced during analysis and resolve steps.
This is not causing issues currently as the dependency graph ensures the
right classes are re-analyzed when needed, instead of showing stale
diagnostics. However, the dependency graph was not intended to serve this
role, and could potentially be optimized in ways that would break this
functionality.
This commit removes the concept of an "errored" trait entirely from the
trait system. Instead, analyzed and resolved traits have corresponding (and
separate) diagnostics, in addition to potentially `null` analysis results.
Analysis (but not resolution) diagnostics are carried forward during
incremental build operations. Compilation (emit) is only performed when
a trait reaches the resolved state with no diagnostics.
This change is functionally different than before as the `register` step is
now performed even in the presence of analysis errors, as long as analysis
results are also produced. This fixes problem 1 above, and is part of the
larger solution to problem 2.
PR Close#39923
ngtsc has a robust suite of testing utilities, designed for in-memory
testing of a TypeScript compiler. Previously these utilities lived in the
`test` directory for the compiler-cli package.
This commit moves those utilities to an `ngtsc/testing` package, enabling
them to be depended on separately and opening the door for using them from
the upcoming language server testing infrastructure.
As part of this refactoring, the `fake_core` package (a lightweight API
replacement for @angular/core) is expanded to include functionality needed
for Language Service test use cases.
PR Close#39594
This is a precursor to introducing the Angular linker. As an initial
step, a compiler option to configure the compilation mode is introduced.
This option is initially internal until the linker is considered ready.
PR Close#38938
The `Logger` interface and its related classes are general purpose
and could be used by other tooling. Moving it into ngtsc is a more
suitable place from which to share it - similar to the FileSystem stuff.
PR Close#37114
Previously, when we needed to detect whether a file is external to a
package, we only checked whether the relative path to the file from the
package's root started with `..`. This would detect external imports
when the packages were siblings (e.g. peer dependencies or hoisted to
the top of `node_modules/` by the package manager), but would fail to
detect imports from packages located in nested `node_modules/` as
external. For example, importing `node_modules/foo/node_modules/bar`
from a file in `node_modules/foo/` would be considered internal to the
`foo` package.
This could result in processing/analyzing more files than necessary.
More importantly it could lead to errors due to trying to analyze
non-Angular packages that were direct dependencies of Angular packages.
This commit fixes it by also verifying that the relative path to a file
does not start with `node_modules/`.
Jira issue: [FW-2068](https://angular-team.atlassian.net/browse/FW-2068)
Fixes#36526
PR Close#36559
In #34288, ngtsc was refactored to separate the result of the analysis
and resolve phase for more granular incremental rebuilds. In this model,
any errors in one phase transition the trait into an error state, which
prevents it from being ran through subsequent phases. The ngcc compiler
on the other hand did not adopt this strict error model, which would
cause incomplete metadata—due to errors in earlier phases—to be offered
for compilation that could result in a hard crash.
This commit updates ngcc to take advantage of ngtsc's `TraitCompiler`,
that internally manages all Ivy classes that are part of the
compilation. This effectively replaces ngcc's own `AnalyzedFile` and
`AnalyzedClass` types, together with all of the logic to drive the
`DecoratorHandler`s. All of this is now handled in the `TraitCompiler`,
benefiting from its explicit state transitions of `Trait`s so that the
ngcc crash is a thing of the past.
Fixes#34500
Resolves FW-1788
PR Close#34889
Previously, the compiler performed an incremental build by analyzing and
resolving all classes in the program (even unchanged ones) and then using
the dependency graph information to determine which .js files were stale and
needed to be re-emitted. This algorithm produced "correct" rebuilds, but the
cost of re-analyzing the entire program turned out to be higher than
anticipated, especially for component-heavy compilations.
To achieve performant rebuilds, it is necessary to reuse previous analysis
results if possible. Doing this safely requires knowing when prior work is
viable and when it is stale and needs to be re-done.
The new algorithm implemented by this commit is such:
1) Each incremental build starts with knowledge of the last known good
dependency graph and analysis results from the last successful build,
plus of course information about the set of files changed.
2) The previous dependency graph's information is used to determine the
set of source files which have "logically" changed. A source file is
considered logically changed if it or any of its dependencies have
physically changed (on disk) since the last successful compilation. Any
logically unchanged dependencies have their dependency information copied
over to the new dependency graph.
3) During the `TraitCompiler`'s loop to consider all source files in the
program, if a source file is logically unchanged then its previous
analyses are "adopted" (and their 'register' steps are run). If the file
is logically changed, then it is re-analyzed as usual.
4) Then, incremental build proceeds as before, with the new dependency graph
being used to determine the set of files which require re-emitting.
This analysis reuse avoids template parsing operations in many circumstances
and significantly reduces the time it takes ngtsc to rebuild a large
application.
Future work will increase performance even more, by tackling a variety of
other opportunities to reuse or avoid work.
PR Close#34288
Prior to this commit, the `IvyCompilation` tracked the state of each matched
`DecoratorHandler` on each class in the `ts.Program`, and how they
progressed through the compilation process. This tracking was originally
simple, but had grown more complicated as the compiler evolved. The state of
each specific "target" of compilation was determined by the nullability of
a number of fields on the object which tracked it.
This commit formalizes the process of compilation of each matched handler
into a new "trait" concept. A trait is some aspect of a class which gets
created when a `DecoratorHandler` matches the class. It represents an Ivy
aspect that needs to go through the compilation process.
Traits begin in a "pending" state and undergo transitions as various steps
of compilation take place. The `IvyCompilation` class is renamed to the
`TraitCompiler`, which manages the state of all of the traits in the active
program.
Making the trait concept explicit will support future work to incrementalize
the expensive analysis process of compilation.
PR Close#34288
When ngcc is analyzing synthetically inserted decorators from a
migration, it is typically not expected that any diagnostics are
produced. In the situation where a diagnostic is produced, however, the
diagnostic would not be reported at all. This commit ensures that
diagnostics in migrations are reported.
PR Close#34014
A class that is provided as Angular service is required to have an
`@Injectable()` decorator so that the compiler generates its injectable
definition for the runtime. Applications are automatically migrated
using the "missing-injectable" schematic, however libraries built for
older version of Angular may not yet satisfy this requirement.
This commit ports the "missing-injectable" schematic to a migration that
is ran when ngcc is processing a library. This ensures that any service
that is provided from an NgModule or Directive/Component will have an
`@Injectable()` decorator.
PR Close#33362
In ngcc's reflection hosts for compiled JS bundles, such as ESM2015,
special care needs to be taken for classes as there may be an outer
declaration (referred to as "declaration") and an inner declaration
(referred to as "implementation") for a given class. Therefore, there
will also be two `ts.Symbol`s bound per class, and ngcc needs to switch
between those declarations and symbols depending on where certain
information can be found.
Prior to this commit, the `NgccReflectionHost` interface had methods
`getClassSymbol` and `findClassSymbols` that would return a `ts.Symbol`.
These class symbols would be used to kick off compilation of components
using ngtsc, so it is important for these symbols to correspond with the
publicly visible outer declaration of the class. However, the ESM2015
reflection host used to return the `ts.Symbol` for the inner
declaration, if the class was declared as follows:
```javascript
var MyClass = class MyClass {};
```
For the above code, `Esm2015ReflectionHost.getClassSymbol` would return
the `ts.Symbol` corresponding with the `class MyClass {}` declaration,
whereas it should have corresponded with the `var MyClass` declaration.
As a consequence, no `NgModule` could be resolved for the component, so
no components/directives would be in scope for the component. This
resulted in errors during runtime.
This commit resolves the issue by introducing a `NgccClassSymbol` that
contains references to both the outer and inner `ts.Symbol`, instead of
just a single `ts.Symbol`. This avoids the unclarity of whether a
`ts.Symbol` corresponds with the outer or inner declaration.
More details can be found here: https://hackmd.io/7nkgWOFWQlSRAuIW_8KPPwFixes#32078
Closes FW-1507
PR Close#32539
JoostK
Alex Rickabaugh
Pete Bacon Darwin
Joey Perrott
George Kalpakas