Bundle standalone runtime in dd-openfeature

[leoromanovsky]

Motivation

Customers using the Datadog OpenFeature provider without APM still have to load dd-java-agent.jar because configuration polling and event delivery live in agent-owned modules. That keeps an agent deployment and lifecycle in the path even when the customer only wants Feature Flagging.

Changes

This packages the agentless UFC poller, evaluator support, exposure delivery, and flag-evaluation delivery into the published dd-openfeature artifact. The artifact shades its internal Datadog and HTTP implementation dependencies while preserving the OpenFeature SDK as its sole published dependency. When no Java agent is installed and DD_FEATURE_FLAGS_CONFIGURATION_SOURCE=agentless is explicit, the provider claims the standalone runtime, polls configuration, and sends events directly with DD_API_KEY.

The provider still activates an installed Java agent first. If the agent claims the runtime, the bundled implementation does not start, so transport and lifecycle remain agent-owned and no duplicate poller or event writer is created.

Decisions

Standalone activation remains explicit: absence of DD_FEATURE_FLAGS_CONFIGURATION_SOURCE=agentless does not start network work, and remote_config remains agent-only. Direct event delivery deliberately does not probe a local EVP proxy because the standalone process must not imply an Agent dependency.

Validation covers the runtime lifecycle, direct-only transport selection, formatting, SpotBugs, Javadocs, and all API/library tests. A published-artifact test compiles and launches a child JVM with no -javaagent, serves UFC from a local HTTP endpoint, and verifies a real OpenFeature evaluation resolves from the polled configuration.

[datadog-datadog-prod-us1-2]

🎯 Code Coverage (details)
• Patch Coverage: 78.95%
• Overall Coverage: 78.41% (-0.00%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 37893e4 | Docs | View more details | Give us feedback!}

[dd-octo-sts]

🟢 Java Benchmark SLOs — All performance SLOs passed

Suite	Status
Startup	🟢 pass

SLO thresholds are defined here based on automatically generated metrics. A warning is raised when results are within 5% of the threshold.

PR vs. master results

Scenario	Candidate	master	Δ (95% CI of mean)
startup:insecure-bank:iast:Agent	14.84 s	14.64 s	[+0.5%; +2.2%] (maybe worse)
startup:insecure-bank:tracing:Agent	13.66 s	13.77 s	[-1.5%; -0.1%] (maybe better)
startup:petclinic:appsec:Agent	17.52 s	17.33 s	[+0.2%; +1.9%] (maybe worse)
startup:petclinic:iast:Agent	17.45 s	17.55 s	[-1.2%; +0.1%] (no difference)
startup:petclinic:profiling:Agent	17.40 s	17.06 s	[+0.9%; +3.1%] (maybe worse)
startup:petclinic:sca:Agent	17.39 s	17.27 s	[-0.3%; +1.7%] (no difference)
startup:petclinic:tracing:Agent	16.64 s	16.62 s	[-0.8%; +1.1%] (no difference)

Commit: 37893e46 · CI Pipeline · Benchmarking Platform UI

---

Load and DaCapo benchmarks can be triggered manually in the GitLab pipeline. Results will appear in the Benchmarking Platform UI after completion.