prepare 6.0.0 release (#285)

* javadoc fixes

* fix test code

* fix test code some more

* move component APIs to new subsystems package

* javadoc fixes

* fix benchmarks

* use streaming JSON parsing for incoming LD data

* fix tests

* rm unused

* rm unused

* use okhttp-eventsource 2.6.0

* update eventsource to 2.6.1 to fix pom/manifest problem

* increase efficiency of summary event data structures (#335)

* make reusable EvaluationDetail instances as part of flag preprocessing (#336)

* make evaluator result object immutable and reuse instances

* comment

* avoid creating List iterators during evaluations

* remove unnecessary copy

* update data model for U2C schema changes, without using contexts yet

* more model updates + tests

* update benchmark code

* more model updates

* use LDContext in Evaluator; add per-kind targets logic

* rm redundant test

* skip inapplicable contract tests

* implement new context kind logic for clauses

* implement new context kind logic for rollouts

* skip inapplicable contract tests

* add context kind logic for regular segments and big segments

* allow segments to reference other segments

* guard against prerequisite cycles

* implement event context redaction logic

* misc fixes

* implement the rest of U2C event logic

* misc fixes

* fix benchmarks

* add contextKinds to summary + misc fixes

* events schema 4

* update client API to use contexts; enable v2 contract tests

* misc fixes

* javadoc & test app fixes

* javadoc fix

* javadoc fix

* unskip some tests since the behavior was fixed in java-sdk-common

* bad attribute in clause should be reported as an error

* update TestData to be context-aware

* misc fixes + javadoc

* fix allFlagsState to not generate prereq eval events

* factor context deduplication logic out of DefaultEventProcessor

* fix JSON serialization test

* refactor diagnostic events logic to abstract out server-side-specific info

* separate inner events implementation from public interface, don't expose event types

* fix benchmarks

* fix benchmarks

* rm unused

* rm unused

* separate Java-SDK-specific HTTP functionality from events logic

* fix tests

* don't close the underlying Writer when we stop using the JsonWriter

* rm unused

* remove references to Java 8 types and Guava from shared event/HTTP code

* fix benchmarks

* also try not to use java.time.Duration in the event tests

* also avoid using lambdas in event test code

* move shared code into internal packages before splitting it off

* add "...ForAll" TestData methods to replace "...ForAllUsers"

* replace all component factory interfaces with a single generic one

* change HttpConfiguration and LoggingConfiguration to concrete classes

* bump okhttp & okhttp-eventsource dependencies

* update comment to clarify that level() doesn't apply to SLF4J

* update readme to mention different logging examples in hello-java

* switch to use snapshot build of java-logging, pending next release

* level setting does not apply to SLF4J and JUL

* use java-logging 1.1.0 release

* make sure META-INF files are never mistaken for classes and relocated

* update shared data store test logic to pass ClientContext with logger

* enable external javadoc links for com.launchdarkly.logging types

* use variable for dependency version

* remove another Guava usage

* make events test code independent of server-side data model

* fix flaky big segment status polling tests

* fix benchmark code

* update benchmarks

* javadoc formatting

* javadocs

* undo accidental reversion

* remove some more stray references to java-server-sdk test helper code

* more efficient equals() and hashCode() for Operator

* method can be static

* clarify comment

* typo

* remove event logic, migrate to using java-sdk-internal

* adjust for API change

* use constant

* fix packaging tests

* remove inapplicable benchmarks

* add error message if makefile can't run

* force result to be "not in experiment" if bucketing calculation failed due to context kind not found

* use newer HTTP/TCP test helpers

* Update Windows orb, fix Windows JDK install in CI (#372)

* update snakeyaml for CVE-CVE-2022-25857

* latest snakeyaml is 1.31

* bump snakeyaml version for CVE-2022-38752

* disable Windows Java 11 build

* minor test updates for java-sdk-internal API changes

* fix packaging of com.launchdarkly.logging classes

* rm debugging

* reconsidered - let's include the logging classes in the jars

* fix packaging test logic

* correct documentation

* remove secondary meta-attribute

* rm obsolete references to UserAttribute

* support passing LDUser instead of LDContext in all SDK methods (#379)

* support passing LDUser instead of LDContext in all SDK methods

* actually we should use default methods

* rm unused

* enable test capability for user type

* update java-sdk-common dependency

* use okhttp-eventsource 3.0.0-SNAPSHOT

* use non-snapshot okhttp-eventsource

* include nested segment references in dependency update checks

* use java-sdk-internal 1.0.0

* remove SLF4J dependency, remove "all" jar, simplify build (#381)

* don't generate evaluation events for invalid context

Co-authored-by: Eli Bishop <[email protected]>
Co-authored-by: LaunchDarklyReleaseBot <[email protected]>
Co-authored-by: Alex Engelberg <[email protected]>

Author: 4 people

Makefile

@@ -10,6 +10,10 @@ test:
 
 TEMP_TEST_OUTPUT=/tmp/sdk-test-service.log
 
+# Add any extra sdk-test-harness parameters here, such as -skip for tests that are
+# temporarily not working.
+TEST_HARNESS_PARAMS=
+
 build-contract-tests:
 	@cd contract-tests && ../gradlew installDist
 
@@ -21,8 +25,8 @@ start-contract-test-service-bg:
 	@make start-contract-test-service >$(TEMP_TEST_OUTPUT) 2>&1 &
 
 run-contract-tests:
-	@curl -s https://raw.githubusercontent.com/launchdarkly/sdk-test-harness/v1.0.0/downloader/run.sh \
-      | VERSION=v1 PARAMS="-url http://localhost:8000 -debug -stop-service-at-end $(TEST_HARNESS_PARAMS)" sh
+	@curl -s https://raw.githubusercontent.com/launchdarkly/sdk-test-harness/v2/downloader/run.sh \
+      | VERSION=v2 PARAMS="-url http://localhost:8000 -debug -stop-service-at-end $(TEST_HARNESS_PARAMS)" sh
 
 contract-tests: build-contract-tests start-contract-test-service-bg run-contract-tests
 

README.md

@@ -15,21 +15,24 @@ This version of the LaunchDarkly SDK works with Java 8 and above.
 
 ## Distributions
 
-Three variants of the SDK jar are published to Maven:
+Two variants of the SDK jar are published to Maven:
 
-* The default uberjar - this is accessible as `com.launchdarkly:launchdarkly-java-server-sdk:jar` and is the dependency used in the "[Getting started](https://docs.launchdarkly.com/sdk/server-side/java#getting-started)" section of the SDK reference guide as well as in the [`hello-java`](https://github.com/launchdarkly/hello-java) sample app. This variant contains the SDK classes, and all of the SDK's dependencies except for SLF4J (the SLF4J API is assumed to be brought in automatically as Maven dependencies, or otherwise made available in the classpath of the host application). All third-party bundled dependencies have shaded package names (and are not exported in OSGi), so they will not interfere with any other versions of the same packages.
-* The extended uberjar - add `<classifier>all</classifier>` in Maven, or `:all` in Gradle. This is the same as the default uberjar except that SLF4J is also bundled, without shading (and is exported in OSGi).
-* The "thin" jar - add `<classifier>thin</classifier>` in Maven, or `:thin` in Gradle. This contains _only_ the SDK classes. Applications using this jar must provide all of the dependencies that are in the SDK's `build.gradle`, so it is intended for use only in special cases.
+* The default uberjar - this is accessible as `com.launchdarkly:launchdarkly-java-server-sdk:jar` and is the dependency used in the "[Getting started](https://docs.launchdarkly.com/sdk/server-side/java#getting-started)" section of the SDK reference guide as well as in the [`hello-java`](https://github.com/launchdarkly/hello-java) sample app. This variant contains the SDK classes and all of its required dependencies. All bundled dependencies that are not surfaced in the public API have shaded package names (and are not exported in OSGi), so they will not interfere with any other versions of the same packages.
+* The "thin" jar - add `<classifier>thin</classifier>` in Maven, or `:thin` in Gradle. This contains only the SDK classes, without its dependencies. Applications using this jar must provide all of the dependencies that are in the SDK's `build.gradle`, so it is intended for use only in special cases.
+
+Previous SDK versions also included a third classifier, `all`, which was the same as the default uberjar but also contained the SLF4J API. This no longer exists because the SDK no longer requires the SLF4J API to be in the classpath.
 
 ## Getting started
 
 Refer to the [SDK reference guide](https://docs.launchdarkly.com/sdk/server-side/java#getting-started) for instructions on getting started with using the SDK.
 
 ## Logging
 
-By default, the LaunchDarkly SDK uses [SLF4J](https://www.slf4j.org/). SLF4J has its own configuration mechanisms for determining where output will go, and filtering by level and/or logger name.
+By default, the LaunchDarkly SDK uses [SLF4J](https://www.slf4j.org/) _if_ the SLF4J API is present in the classpath. SLF4J has its own configuration mechanisms for determining where output will go, and filtering by level and/or logger name.
+
+If SLF4J is not in the classpath, the SDK's default logging destination is `System.err`.
 
-The SDK can also be configured to use other adapters from the [com.launchdarkly.logging](https://github.com/launchdarkly/java-logging) facade instead of SLF4J. See `LoggingConfigurationBuilder`. This allows the logging behavior to be completely determined by the application, rather than by external SLF4J configuration.
+The SDK can also be configured to use other adapters from the [com.launchdarkly.logging](https://github.com/launchdarkly/java-logging) facade. See `LoggingConfigurationBuilder`. This allows the logging behavior to be completely determined by the application, rather than by external SLF4J configuration.
 
 For an example of using the default SLF4J behavior with a simple console logging configuration, check out the [`slf4j-logging` branch](https://github.com/launchdarkly/hello-java/tree/slf4j-logging) of the [`hello-java`](https://github.com/launchdarkly/hello-java) project. The [main branch](https://github.com/launchdarkly/hello-java) of `hello-java` uses console logging that is programmatically configured without SLF4J.
 

benchmarks/Makefile

@@ -4,13 +4,13 @@ BASE_DIR:=$(shell pwd)
 PROJECT_DIR=$(shell cd .. && pwd)
 SDK_VERSION=$(shell grep "version=" $(PROJECT_DIR)/gradle.properties | cut -d '=' -f 2)
 
-BENCHMARK_ALL_JAR=lib/launchdarkly-java-server-sdk-all.jar
+BENCHMARK_SDK_JAR=lib/launchdarkly-java-server-sdk.jar
 BENCHMARK_TEST_JAR=lib/launchdarkly-java-server-sdk-test.jar
 SDK_JARS_DIR=$(PROJECT_DIR)/build/libs
-SDK_ALL_JAR=$(SDK_JARS_DIR)/launchdarkly-java-server-sdk-$(SDK_VERSION)-all.jar
+SDK_JAR=$(SDK_JARS_DIR)/launchdarkly-java-server-sdk-$(SDK_VERSION).jar
 SDK_TEST_JAR=$(SDK_JARS_DIR)/launchdarkly-java-server-sdk-$(SDK_VERSION)-test.jar
 
-benchmark: $(BENCHMARK_ALL_JAR) $(BENCHMARK_TEST_JAR)
+benchmark: $(BENCHMARK_SDK_JAR) $(BENCHMARK_TEST_JAR)
 	rm -rf build/tmp
 	../gradlew jmh
 	cat build/reports/jmh/human.txt
@@ -21,16 +21,16 @@ clean:
 
 sdk: $(BENCHMARK_ALL_JAR) $(BENCHMARK_TEST_JAR)
 
-$(BENCHMARK_ALL_JAR): $(SDK_ALL_JAR)
+$(BENCHMARK_SDK_JAR): $(SDK_JAR)
 	mkdir -p lib
 	cp $< $@
 
 $(BENCHMARK_TEST_JAR): $(SDK_TEST_JAR)
 	mkdir -p lib
 	cp $< $@
 
-$(SDK_ALL_JAR):
-	cd .. && ./gradlew shadowJarAll
+$(SDK_JAR):
+	cd .. && ./gradlew shadowJar
 
 $(SDK_TEST_JAR):
 	cd .. && ./gradlew testJar

benchmarks/build.gradle

@@ -20,7 +20,7 @@ ext.versions = [
 ]
 
 dependencies {
-    implementation files("lib/launchdarkly-java-server-sdk-all.jar")
+    implementation files("lib/launchdarkly-java-server-sdk.jar")
     implementation files("lib/launchdarkly-java-server-sdk-test.jar")
     implementation "com.google.code.gson:gson:2.8.9"
     implementation "com.google.guava:guava:${versions.guava}" // required by SDK test code

benchmarks/src/jmh/java/com/launchdarkly/sdk/server/EventProcessorBenchmarks.java

@@ -1,10 +1,10 @@
 package com.launchdarkly.sdk.server;
 
-import com.launchdarkly.sdk.LDUser;
+import com.launchdarkly.sdk.LDContext;
 import com.launchdarkly.sdk.LDValue;
 import com.launchdarkly.sdk.server.DataModel.FeatureFlag;
-import com.launchdarkly.sdk.server.interfaces.DataStore;
 import com.launchdarkly.sdk.server.interfaces.LDClientInterface;
+import com.launchdarkly.sdk.server.subsystems.DataStore;
 
 import org.openjdk.jmh.annotations.Benchmark;
 import org.openjdk.jmh.annotations.Scope;
@@ -13,7 +13,7 @@
 import java.util.Random;
 
 import static com.launchdarkly.sdk.server.TestComponents.initedDataStore;
-import static com.launchdarkly.sdk.server.TestComponents.specificDataStore;
+import static com.launchdarkly.sdk.server.TestComponents.specificComponent;
 import static com.launchdarkly.sdk.server.TestUtil.upsertFlag;
 import static com.launchdarkly.sdk.server.TestValues.BOOLEAN_FLAG_KEY;
 import static com.launchdarkly.sdk.server.TestValues.CLAUSE_MATCH_VALUE_COUNT;
@@ -22,8 +22,8 @@
 import static com.launchdarkly.sdk.server.TestValues.FLAG_WITH_TARGET_LIST_KEY;
 import static com.launchdarkly.sdk.server.TestValues.INT_FLAG_KEY;
 import static com.launchdarkly.sdk.server.TestValues.JSON_FLAG_KEY;
-import static com.launchdarkly.sdk.server.TestValues.NOT_MATCHED_VALUE_USER;
-import static com.launchdarkly.sdk.server.TestValues.NOT_TARGETED_USER_KEY;
+import static com.launchdarkly.sdk.server.TestValues.NOT_MATCHED_VALUE_CONTEXT;
+import static com.launchdarkly.sdk.server.TestValues.NOT_TARGETED_CONTEXT_KEY;
 import static com.launchdarkly.sdk.server.TestValues.SDK_KEY;
 import static com.launchdarkly.sdk.server.TestValues.STRING_FLAG_KEY;
 import static com.launchdarkly.sdk.server.TestValues.TARGETED_USER_KEYS;
@@ -41,7 +41,7 @@ public class LDClientEvaluationBenchmarks {
   public static class BenchmarkInputs {
     // Initialization of the things in BenchmarkInputs does not count as part of a benchmark.
     final LDClientInterface client;
-    final LDUser basicUser;
+    final LDContext basicUser;
     final Random random;
 
     public BenchmarkInputs() {
@@ -51,14 +51,14 @@ public BenchmarkInputs() {
       }
 
       LDConfig config = new LDConfig.Builder()
-        .dataStore(specificDataStore(dataStore))
+        .dataStore(specificComponent(dataStore))
         .events(Components.noEvents())
         .dataSource(Components.externalUpdatesOnly())
         .logging(Components.noLogging())
         .build();
       client = new LDClient(SDK_KEY, config);
 
-      basicUser = new LDUser("userkey");
+      basicUser = LDContext.create("userkey");
 
       random = new Random();
     }
@@ -127,13 +127,13 @@ public void jsonVariationForUnknownFlag(BenchmarkInputs inputs) throws Exception
   @Benchmark
   public void userFoundInTargetList(BenchmarkInputs inputs) throws Exception {
     String userKey = TARGETED_USER_KEYS.get(inputs.random.nextInt(TARGETED_USER_KEYS.size()));
-    boolean result = inputs.client.boolVariation(FLAG_WITH_TARGET_LIST_KEY, new LDUser(userKey), false);
+    boolean result = inputs.client.boolVariation(FLAG_WITH_TARGET_LIST_KEY, LDContext.create(userKey), false);
     assertTrue(result);
   }
 
   @Benchmark
   public void userNotFoundInTargetList(BenchmarkInputs inputs) throws Exception {
-    boolean result = inputs.client.boolVariation(FLAG_WITH_TARGET_LIST_KEY, new LDUser(NOT_TARGETED_USER_KEY), false);
+    boolean result = inputs.client.boolVariation(FLAG_WITH_TARGET_LIST_KEY, LDContext.create(NOT_TARGETED_CONTEXT_KEY), false);
     assertFalse(result);
   }
 
@@ -146,14 +146,14 @@ public void flagWithPrerequisite(BenchmarkInputs inputs) throws Exception {
   @Benchmark
   public void userValueFoundInClauseList(BenchmarkInputs inputs) throws Exception {
     int i = inputs.random.nextInt(CLAUSE_MATCH_VALUE_COUNT);
-    LDUser user = TestValues.CLAUSE_MATCH_VALUE_USERS.get(i);
-    boolean result = inputs.client.boolVariation(FLAG_WITH_MULTI_VALUE_CLAUSE_KEY, user, false);
+    LDContext context = TestValues.CLAUSE_MATCH_VALUE_CONTEXTS.get(i);
+    boolean result = inputs.client.boolVariation(FLAG_WITH_MULTI_VALUE_CLAUSE_KEY, context, false);
     assertTrue(result);
   }
 
   @Benchmark
   public void userValueNotFoundInClauseList(BenchmarkInputs inputs) throws Exception {
-    boolean result = inputs.client.boolVariation(FLAG_WITH_MULTI_VALUE_CLAUSE_KEY, NOT_MATCHED_VALUE_USER, false);
+    boolean result = inputs.client.boolVariation(FLAG_WITH_MULTI_VALUE_CLAUSE_KEY, NOT_MATCHED_VALUE_CONTEXT, false);
     assertFalse(result);
   }
 }