The Gradle team is excited to announce Gradle 6.6.1.
This release introduces a major performance optimization as an experimental opt-in. Configuration caching allows Gradle to skip the configuration phase of the build and start executing tasks as soon as possible.
Other improvements in this release include conventions for handling user-provided credentials, support for Java compilation --release flag, and a number of other improvements and bug fixes.
We would like to thank the following community contributors to this release of Gradle:
SheliakLyr, Danny Thomas, Daiki Hirabayashi, Sebastian Schuberth, Frieder Bluemle, Brick Tamland, Stefan Oehme, Yurii Serhiichuk, JunHyung Lim, Igor Dvorzhak, and Leonid Gaiazov.
Switch your build to use Gradle 6.6.1 by updating your wrapper:
./gradlew wrapper --gradle-version=6.6.1
See the Gradle 6.x upgrade guide to learn about deprecations, breaking changes and other considerations when upgrading to Gradle 6.6.1.
NOTE: Gradle 6.6 has had one patch release, which fixed several issues from the original release. We recommend always using the latest patch release.
For Java, Groovy, Kotlin and Android compatibility, see the full compatibility notes.
Fast feedback for local incremental builds is crucial for developer productivity. This is especially true when your IDE uses Gradle to build and run tests for your project, which IntelliJ IDEA does by default. This scenario has been the primary focus of performance improvements since Gradle 6.5 and will continue for the next several Gradle releases.
Before running any task, Gradle needs to run the configuration phase. Currently, this is done on every build invocation and can incur a noticeable overhead, especially in large projects.
The configuration cache significantly improves build performance by caching the result of the configuration phase and reusing it for subsequent builds. Using the configuration cache, Gradle can skip the configuration phase entirely when nothing that affects the build configuration has changed as you can see below.
Additionally, Gradle is able to optimize task execution when configuration caching is enabled and execute more tasks in parallel by default.
Note that configuration caching is different from the build cache, which caches outputs produced by the build. The configuration cache captures only the state of the configuration phase.
IDE sync and import does not currently benefit from configuration caching.
This feature is considered highly experimental and not enabled by default or recommended for production use. Tasks and plugins usually require changes to meet the requirements to use configuration caching. Not all core Gradle plugins have been updated yet and some features are not yet implemented. Your build may likely require changes.
You can enable this experimental feature by supplying the parameter --configuration-cache
on the command-line or adding it to your run configuration. If your build is incompatible with configuration caching, Gradle will generate a report describing the problems found.
Learn more about this new feature and its impact in the Configuration Cache documentation.
Gradle 6.5 introduced an experimental opt-in that improves the performance of local incremental builds by watching for file-system changes.
This release brings a number of stability improvements for file-system watching when used with composite builds or large projects on Windows and macOS. Gradle will now report better errors when you enable file-system watching on unsupported systems.
For up-to-date checks and the build cache, Gradle needs to determine if two task input properties have the same value. In order to do so, Gradle first normalizes both inputs and then compares the result.
Runtime classpath analysis now inspects manifest and META-INF
properties files, ignoring changes to comments, whitespace and order-differences. Moreover, you can selectively ignore attributes or properties that don't impact the runtime classpath.
normalization {
runtimeClasspath {
metaInf {
ignoreAttribute("Implementation-Version")
ignoreProperty("timestamp")
}
}
}
This improves the likelihood of build cache hits when any ZIP file on the classpath is regenerated and only differs by unimportant values or comments. The most common case where this sort of normalization can be useful is with JAR files, but it can be applied to any ZIP file on the classpath--such as AAR, WAR, or APK files.
See the user manual for further information. Note that this API is incubating and will likely change in future releases as support is expanded for normalizing properties files outside of the META-INF
directory.
Builds sometimes require users to supply credentials. For example, credentials might be required to authenticate with an artifact repository in order to publish an artifact. It's a good practice to keep credentials outside the build script.
This release includes a new API for credentials that makes working with credentials easier by establishing a convention to supply credentials using Gradle properties that can be provided to the build as command-line arguments, environment variables, or as values in a gradle.properties
file. It also introduces fail-fast behavior when Gradle knows that the build will need credentials at some point and the credentials are missing.
Starting from this release, you can easily externalize credentials used for authentication to an artifact repository:
repositories {
maven {
name = 'mySecureRepository'
credentials(PasswordCredentials)
// url = uri(<<some repository url>>)
}
}
The credentials for mySecureRepository
will be searched for in Gradle properties with the names mySecureRepositoryUsername
and mySecureRepositoryPassword
.
For more details on using the new API to authenticate with artifact repositories, see the user manual section as well as an updated sample.
You can also use the new provider API directly to supply credentials to an external tool:
tasks.register('login', Exec) {
def loginProvider =
providers.credentials(PasswordCredentials, 'login')
inputs.property('credentials', loginProvider)
doFirst {
PasswordCredentials loginCredentials = loginProvider.get()
// use credentials
}
}
The credentials for the above will be searched for in Gradle properties with the names loginUsername
and loginPassword
.
See the updated sample for more details.
--release
flag in Java compilationJava 9 introduced cross compilation support with the --release
flag on the Java compiler. This option tells the compiler to produce bytecode for an earlier version of Java and guarantees that the code does not use any APIs from later versions.
In previous Gradle versions, it could be achieved through the use of compilerArgs
and making sure that sourceCompatibility
and targetCompatibility
are not set:
compileJava {
options.compilerArgs.addAll(['--release', '7'])
}
With this release, Gradle makes this use case easier by supporting the --release
flag for Java compilation directly on the CompileOptions
of JavaCompile
tasks:
compileJava {
options.release = 7
}
See the section on cross compilation for details.
Gradle Module Metadata is a format used to serialize the Gradle component model, similar to but more powerful than Maven’s POM.
By default, the Gradle Module Metadata file contains a build identifier field which defaults to a unique ID generated during build execution. This behaviour can now be disabled at the publication level, allowing users to opt-in for a reproducible Gradle Module Metadata file. This enables downstream tasks to consider it up-to-date, resulting in faster and reproducible builds.
main(MavenPublication) {
from components.java
withoutBuildIdentifier()
}
See the documentation for more information on Gradle Module Metadata generation.
It’s a common problem in dependency management that the same dependency can appear in a dependency graph but with different attributes. For example, you want to only use the “fat jar” with repackaged dependencies, but the regular jar is pulled in transitively. The far jar may be published under a "fat jar" classifier, while the regular jar has no classifier.
Previously, it wasn't possible for Gradle to substitute a dependency using a classifier with a dependency without a classifier, nor was it possible to substitute a dependency without classifier with a dependency with a classifier.
Similarly, other attributes (typically "platform" dependencies) or capabilities could not be used when describing dependency substitutions.
Gradle now supports declaring substitutions based on classifiers, attributes, or capabilities. Gradle's dependency substitution API has been enriched to cover those cases.
See the documentation on variant-aware substitution for details.
ArchiveOperations
servicePreviously, it was only possible to create a FileTree
for a ZIP or TAR archive by using the APIs provided by a Project
.
However, a Project
object is not always available, for example in worker actions or when using the configuration cache.
The new ArchiveOperations
service has zipTree() and tarTree() methods for creating read-only FileTree
instances respectively for ZIP and TAR archives.
See the user manual for how to inject services and the ArchiveOperations
API documentation for more details and examples.
When using Lazy Properties, it’s common to compute a value by combining the values of two providers. In previous Gradle releases, it wasn’t possible to do this without eagerly reading one of the provider values or losing dependency information. Gradle 6.6 introduces a zip
method which lets you provide the combined value lazily.:
def hello = objects.property(String).convention("Hello")
def world = objects.property(String).convention("World")
def helloWorld = hello.zip(world) { left, right ->
"${left}, ${right}!".toString()
}
// ...
hello.set("Bonjour")
world.set("le monde")
println(helloWorld.get()) // prints “Bonjour, le monde!”
Refer to the API documentation for details.
Debug level logging may expose sensitive information in the build log output, for example in CI server logs. For this reason, Gradle displays a prominent warning when using debug level logging since version 6.4. One example of this risk is leaking secret values such as credentials stored in environmental variables.
Previously, when debug level was enabled, Gradle used to log all environment variables when starting a process such as a test, Gradle daemon, or when using Project.exec
. In practice, this means most of the builds logged environment variables on debug level.
As an additional security precaution, Gradle no longer logs environment variables when starting processes starting with this version.
Note that many CI servers, like Jenkins and Teamcity, mask secrets in the captured logs. Still, we recommend limiting the usage of debug level logging to environments which do not capture the log output, like your local machine. Build scans never capture the debug log as part of the console log even when you enabled debug logging.
As an additional measure, you may want to limit the environment variables passed to the test task or other forked processes by explicitly using ProcessForkOptions.setEnvironment()
. This way the forked processes themselves cannot leak secrets from the environment, since they don't have them available anymore.
Promoted features are features that were incubating in previous versions of Gradle but are now supported and subject to backward compatibility. See the User Manual section on the “Feature Lifecycle” for more information.
The following are the features that have been promoted in this Gradle release.
Known issues are problems that were discovered post release that are directly related to changes made in this release.
We love getting contributions from the Gradle community. For information on contributing, please see gradle.org/contribute.
If you find a problem with this release, please file a bug on GitHub Issues adhering to our issue guidelines. If you're not sure you're encountering a bug, please use the forum.
We hope you will build happiness with Gradle, and we look forward to your feedback via Twitter or on GitHub.