Age | Commit message (Collapse) | Author |
|
RELNOTES=`com.google.auto.common` is annotated for null-safety.
PiperOrigin-RevId: 382311770
|
|
RELNOTES=n/a
PiperOrigin-RevId: 377297134
|
|
in Class.
These changes continue the effort to move away from referencing classes from the user API directly in processors, which can cause issues especially with Android dependencies.
Note: I decided not to do anything with isTypeOf yet because I'm not sure of a great way to support that.
RELNOTES=Add String and TypeElement versions of auto-common APIs that currently take in Class.
PiperOrigin-RevId: 377287397
|
|
reformatting is excluded.
RELNOTES=n/a
PiperOrigin-RevId: 377145832
|
|
ancestor.
RELNOTES=Fixed a crash when `getLocalAndInheritedMethods` is called with a raw-type ancestor.
PiperOrigin-RevId: 373583104
|
|
Throwing `IllegalArgumentException` can mask the real source of a problem, or prevent code that would have handled the `ErrorType` from running. The `ErrorType` typically (maybe always) corresponds to an identifier in the source code, so it is in fact a class, just one that doesn't exist.
RELNOTES=MoreTypes.isTypeOf returns false for ErrorType rather than throwing.
PiperOrigin-RevId: 369464743
|
|
PiperOrigin-RevId: 326930055
|
|
String-based API.
Context: Per discussion in [] this change adds a String-based `Step` class as a replacement for `ProcessingStep`. The new `Step` contains the same two methods (annotations() & process()) which return and accept fully-qualified annotation names. To support the potential absence of the annotation classes from the processor classpath the corresponding methods were changed to no longer depend on explicit Class types.
RELNOTES=Adding String-based `Step` as a replacement for `ProcessingStep` to BasicAnnotationProcessor. Allows for fully-qualified Annotation names to be specified in a processing Step in order to remove the requirement that implementation processors depend directly on their Annotation classes.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=316968142
|
|
In Google's internal java8 compilation environment, problems with symbolic links mean that ecj can't find the standard system classes without this workaround. In other environments, including Google's java11, this problem doesn't arise.
RELNOTES=Make OverridesTest work with Google's internal java8 compilation environment.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=300664328
|
|
given type, whether casting Object to that type will elicit an "unchecked" warning from the compiler.
RELNOTES=Added MoreTypes.isConversionFromObjectUnchecked to test whether casting to a type will elicit an "unchecked" compiler warning.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=288896286
|
|
This CL tries to avoid using expensive method calls like Elements#getTypeElement, TypeElement#getKind(), and Types#directSupertypes. Of all of these, TypeElement#getKind() had the largest cost because it requires symbol completion. In addition, getKind() was being called on unnecessarily on all supertypes, even though only the first type can possibly be an ElementKind.CLASS.
RELNOTES=N/A
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=280196697
|
|
RELNOTES=Adding helper methods for getting different types of annotation values with AnnotationValueVisitor.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=275921910
|
|
errors in the processing round. The extra errors don't add any information and end up hiding the real errors.
I found this substantially reduced the issue highlighted in b/141176717.
RELNOTES=Suppress error noise in `com.google.auto.common.BasicAnnotationProcessor`
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=270454309
|
|
Tested:
TAP --sample ran all affected tests and none failed
[]
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=259328073
|
|
This CL adds a new method, MoreElements#getAllMethods(), which returns all local and inherited methods (including static methods), but does not include any methods that have been overridden.
Note: we are still relying on an explicit ordering for static methods to ensure we don't include hidden static methods from super classes.
RELNOTES=Adds MoreElements#getAllMethods(), which returns all local and inherited methods (including static methods), but does not include any methods that have been overridden.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=257649768
|
|
We're removing it from open-source Truth, since we're about to permanently commit to the remaining open-source Truth APIs, and we're not sure if assert_().fail(...) offers positive value when assertWithMessage(...).fail() is more in line with normal Truth practices.
MOE_MIGRATED_REVID=254043172
|
|
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=249869914
|
|
The two behave identically, and containsAllOf is being removed.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=245840719
|
|
The two behave identically, and containsAllIn is being removed.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=245837617
|
|
(The exact change is slightly different in some cases, like when using custom subjects or check(), but it's always a migration from named(...) to [assert]WithMessage(...).)
named(...) is being removed.
This CL may slightly modify the failure messages produced, but all the old information will still be present.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=245027656
|
|
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=244250754
|
|
ThrowableSubject.hasMessageThat(). This replaces assertions of the form assertThat(e).hasMessage(...) and assertThat(e.getMessage()) with assertThat(e).hasMessageThat().
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=239181584
|
|
RELNOTES=n/a
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=226067929
|
|
versions of the JDK.
RELNOTES=n/a
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=223350143
|
|
non-deprecated replacements.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=223170828
|
|
RELNOTES=n/a
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=219712298
|
|
when there had been deferred elements.
RELNOTES=Fix bug where `ProcessingStep.process(...)` was called with too many elements when there had been deferred elements.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=218513715
|
|
RELNOTES=(auto.common) Add `MoreElements.asTypeParameter()`
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=205449221
|
|
RELNOTES=N/A
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=185615476
|
|
internal code to auto-common
RELNOTES=Added `SimpleAnnotationMirror` and `SimpleTypeAnnotationValue` to easily create new `AnnotationMirror`s in code.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=179840730
|
|
Without this, effectiveVisibilityOfElement() will always return at least DEFAULT, since there are no modifiers on a module.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167283498
|
|
with an ECJ that works properly there.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167210566
|
|
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=154221087
|
|
containing type was not being compared correctly.
Closes https://github.com/google/auto/pull/476
Thanks to Dusan Bartos <doodeec@gmail.com> for pointing out the bug. His one-character change might reasonably have been expected to fix the problem, but alas things are rarely so simple.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=153296509
|
|
overrides, by reworking the logic of the explicit override implementation.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=135954107
|
|
MoreElements.getLocalAndInheritedMethods with a reimplementation that should work with Eclipse. Fixes https://github.com/google/auto/issues/372.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=133716523
|
|
the new simplified import rules.
One test, AutoFactoryProcessorTest, had to be updated with modified line numbers because unused imports in bad/MixedFinals.java were deleted.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=129643318
|
|
that Set<Set<Object>> is the same as Set<Set<String>>.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=108368220
|
|
nested type boundaries. We deferred the type because some annotated element within that type itself, and not within a nested type, had to be deferred; don't overtraverse when looking for those elements again.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=100847273
|
|
into MoreElements.
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=98943442
|
|
appropriate)
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=98172107
|
|
not ready for processing, but might be in a later round.
This lets us impose a partial order over annotation processors when required. For instance, if a Dagger component can't be generated until some AutoFactory class is generated, the Dagger processor could use this mechanism instead of failing arbitrarily when the processors happen to run out of order.
API difference notes: This changes `ProcessingStep.process` from a void method to one that returns `Set<Element>`. For implementing classes to compile with no behavioral changes, they must simply return an empty set.
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=97822074
|
|
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=94596593
|
|
Its logic to test that GeneratesCodeProcessor had run before RequiresGeneratedCodeProcessor was incorrect. Specifically:
1. Filer.getResource() does not throw for nonexisting files if the location is the output location.
2. The resource to test should have a ".java" extension.
3. The filer fails if you even try to read a resource that exists.
Instead, we just keep track of whether the GeneratesCodeProcessor has run on elements and check that explicitly.
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=92832468
|
|
but can be called statically. Then, clean up all of the places that callers to MoreTypes.asTypeElement were passing through a Types instance just for that reason.
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=84753119
|
|
JDK 7.
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=84581037
|
|
types (supporting, for instance, generics provides methods in dagger.
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=84577408
|
|
returns an Optional<DeclaredType>, in support of deriving superclasses with
proper type parameters.
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=83959294
|
|
considered equal. Also add MoreElements.isType(Element) and
MoreTypes.isType(TypeMirror) as parallels to MoreElements.asType &
MoreTypes.isTypeOf (to guard against exceptions from calling them).
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=83672418
|
|
intersection-type bounds, even though the representation of those differs between Java 7 and Java 8.
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=83567846
|