[Feature flags] Send exposures as batches by jonathanmos · Pull Request #2895 · DataDog/dd-sdk-android

jonathanmos · 2025-09-25T08:45:14Z

What does this PR do?

This is part of the feature flags project.
• Takes the Flags Feature name from the core, to match the rest of our features
• GSON added as a transitive dependency.
• Writes feature flags exposures to batches and sends them to the endpoint using the core mechanism.
• Increases test coverage past 80% for the module.
• Exposures deduplication lrucache set at 4mb max size.
• Sending exposures batches is controlled by a configuration option in FlagsConfiguration of trackExposures that is set to default true.

Motivation

What inspired you to submit this pull request?

Additional Notes

Anything else we should know when reviewing?

Review checklist (to be filled by reviewers)

Feature or bugfix MUST have appropriate tests (unit, integration, e2e)
Make sure you discussed the feature or bugfix with the maintaining team in an Issue
Make sure each commit and the PR mention the Issue number (cf the CONTRIBUTING doc)

codecov-commenter · 2025-09-25T11:32:59Z

Codecov Report

❌ Patch coverage is 94.07895% with 9 lines in your changes missing coverage. Please review.
✅ Project coverage is 70.69%. Comparing base (b641740) to head (44bdded).

Files with missing lines	Patch %	Lines
.../flags/featureflags/internal/DatadogFlagsClient.kt	76.19%	3 Missing and 2 partials ⚠️
...com/datadog/android/flags/internal/FlagsFeature.kt	75.00%	2 Missing and 2 partials ⚠️

Additional details and impacted files

@@                     Coverage Diff                      @@
##           feature/feature-flagging    #2895      +/-   ##
============================================================
+ Coverage                     70.36%   70.69%   +0.33%     
============================================================
  Files                           826      830       +4     
  Lines                         30084    30203     +119     
  Branches                       5124     5128       +4     
============================================================
+ Hits                          21167    21349     +182     
+ Misses                         7465     7409      -56     
+ Partials                       1452     1445       -7

Files with missing lines	Coverage Δ
.../kotlin/com/datadog/android/api/feature/Feature.kt	`100.00% <ø> (ø)`
...src/main/kotlin/com/datadog/android/flags/Flags.kt	`88.89% <100.00%> (+0.21%)`	⬆️
...in/com/datadog/android/flags/FlagsConfiguration.kt	`100.00% <100.00%> (ø)`
...ernal/repository/net/DefaultFlagsNetworkManager.kt	`49.22% <ø> (ø)`
.../android/flags/internal/ExposureEventsProcessor.kt	`100.00% <100.00%> (ø)`
...adog/android/flags/internal/model/ExposureEvent.kt	`100.00% <100.00%> (ø)`
...roid/flags/internal/net/ExposuresRequestFactory.kt	`100.00% <100.00%> (ø)`
...lags/internal/storage/ExposureEventRecordWriter.kt	`100.00% <100.00%> (ø)`
...android/flags/internal/storage/NoOpRecordWriter.kt	`100.00% <100.00%> (ø)`
...com/datadog/android/flags/internal/FlagsFeature.kt	`80.00% <75.00%> (+37.14%)`	⬆️
... and 1 more

... and 35 files with indirect coverage changes

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

typotter

Looks great!
Just a couple small name/refactor comments.

0xnm

I reviewed production files and left some comments.

0xnm · 2025-10-06T07:55:24Z

+        private var enableExposureLogging: Boolean = false

        /**
         * Sets whether exposures should be logged to RUM. This is disabled by default.


just curious: what is the impact if it is enabled? billing?

You've got it. It's RUM ingestion and eventually, experiments ingestion events that would hit billing. The other impact is that feedback for Feature Flags won't be available in parts of the Datadog Feature Flags & Experimentation app, so the customer loses observability.

We probably need to put more focus on explaining what it is, because it impacts billing, so it is kind of important parameter to change.

0xnm · 2025-10-06T08:04:49Z

+import com.google.gson.JsonElement
+import com.google.gson.JsonObject
+
+internal data class ExposureEvent(


should we maybe simply create json schema locally and then generate event from there?

Added a todo to do this in another pr

0xnm · 2025-10-06T14:56:14Z

+    private data class CacheKey(
+        val targetingKey: String,
+        val flagName: String,
+        val allocationKey: String,
+        val variationKey: String
+    )


with 4 unrelated properties making a cache key do we expect a high rate of cache miss?

The number of misses should be low once all of the Flags in an app have been evaluated for the first time.
targetingKey is likely to remain stable as it typically references the user. The length of the cache is essentially the number of Flags that are evaluated times the churn in assignment.
The allocationKey & variationKey only change if the evaluationContext changes is reset and the resulting flag assignment changes. The flux in assignment varies depending on the kind of feature flag (i.e. a rollout vs A/B test) but would generally be pretty stable during a user's session in the app. As a result, churn on Allocation and Variation would be very low.

typotter

LGTM

typotter · 2025-10-06T16:11:20Z

+        private var enableExposureLogging: Boolean = false

        /**
         * Sets whether exposures should be logged to RUM. This is disabled by default.


You've got it. It's RUM ingestion and eventually, experiments ingestion events that would hit billing. The other impact is that feedback for Feature Flags won't be available in parts of the Datadog Feature Flags & Experimentation app, so the customer loses observability.

typotter · 2025-10-06T16:19:46Z

+    private data class CacheKey(
+        val targetingKey: String,
+        val flagName: String,
+        val allocationKey: String,
+        val variationKey: String
+    )


The number of misses should be low once all of the Flags in an app have been evaluated for the first time.
targetingKey is likely to remain stable as it typically references the user. The length of the cache is essentially the number of Flags that are evaluated times the churn in assignment.
The allocationKey & variationKey only change if the evaluationContext changes is reset and the resulting flag assignment changes. The flux in assignment varies depending on the kind of feature flag (i.e. a rollout vs A/B test) but would generally be pretty stable during a user's session in the app. As a result, churn on Allocation and Variation would be very low.

jonathanmos force-pushed the jmoskovich/flags-feature-batching branch from 48599ba to 50bbdd6 Compare September 25, 2025 08:46

jonathanmos requested a review from typotter September 25, 2025 08:51

jonathanmos force-pushed the jmoskovich/flags-feature-batching branch 2 times, most recently from 5f1386d to bedf595 Compare September 25, 2025 10:55

jonathanmos changed the title ~~Send exposures as batches~~ [Feature flags] Send exposures as batches Sep 25, 2025

jonathanmos marked this pull request as ready for review September 29, 2025 12:43

jonathanmos requested review from a team as code owners September 29, 2025 12:43

aleksandr-gringauz reviewed Sep 29, 2025

View reviewed changes

aleksandr-gringauz reviewed Sep 30, 2025

View reviewed changes

Comment thread ...-android-flags/src/main/kotlin/com/datadog/android/flags/internal/ExposureEventsProcessor.kt Outdated

aleksandr-gringauz previously approved these changes Sep 30, 2025

View reviewed changes

typotter previously approved these changes Oct 1, 2025

View reviewed changes

Comment thread ...lags/src/main/kotlin/com/datadog/android/flags/featureflags/internal/DatadogFlagsProvider.kt Outdated

Comment thread ...-android-flags/src/main/kotlin/com/datadog/android/flags/internal/net/FlagsRequestFactory.kt Outdated

jonathanmos dismissed stale reviews from typotter and aleksandr-gringauz via 96114fc October 5, 2025 10:29

jonathanmos force-pushed the jmoskovich/flags-feature-batching branch from 86c1e46 to 96114fc Compare October 5, 2025 10:29

aleksandr-gringauz previously approved these changes Oct 6, 2025

View reviewed changes

0xnm requested changes Oct 6, 2025

View reviewed changes

ambushwork reviewed Oct 6, 2025

View reviewed changes

Comment thread ...ures/dd-sdk-android-flags/src/main/kotlin/com/datadog/android/flags/internal/FlagsFeature.kt

jonathanmos dismissed aleksandr-gringauz’s stale review via 05e5a4a October 6, 2025 13:10

jonathanmos force-pushed the jmoskovich/flags-feature-batching branch 3 times, most recently from e8dc8d2 to 63f5129 Compare October 6, 2025 13:48

jonathanmos requested review from 0xnm, aleksandr-gringauz, ambushwork and typotter October 6, 2025 14:51

0xnm reviewed Oct 6, 2025

View reviewed changes

typotter previously approved these changes Oct 6, 2025

View reviewed changes

typotter mentioned this pull request Oct 7, 2025

Flags.enable and FlagsClient APIs #2900

Merged

3 tasks

jonathanmos added 5 commits October 8, 2025 10:16

Write exposures to batches

7ebd52c

Improved thread safety

8fba0f3

Fix rebase issues

f0fdb70

Rename to ExposuresRequestFactory

f7c201e

Change exposuresSentCache to an lruCache

6ee424a

jonathanmos dismissed typotter’s stale review via 16b6e3a October 8, 2025 12:02

jonathanmos force-pushed the jmoskovich/flags-feature-batching branch from 63f5129 to 16b6e3a Compare October 8, 2025 12:02

Set exposure lrucache to 4mb

a1e7ee1

jonathanmos force-pushed the jmoskovich/flags-feature-batching branch from 16b6e3a to a1e7ee1 Compare October 8, 2025 12:07

change enableExposureLogging to trackExposures

5e27688

jonathanmos force-pushed the jmoskovich/flags-feature-batching branch from e53c83c to 5e27688 Compare October 8, 2025 12:23

jonathanmos requested review from 0xnm and typotter October 8, 2025 13:30

0xnm reviewed Oct 8, 2025

View reviewed changes

Add forgery factory for flagsConfiguration

ce2957b

jonathanmos requested a review from 0xnm October 9, 2025 09:51

0xnm previously approved these changes Oct 9, 2025

View reviewed changes

Small fixes

44bdded

jonathanmos dismissed 0xnm’s stale review via 44bdded October 9, 2025 11:25

jonathanmos merged commit 09aacbf into feature/feature-flagging Oct 9, 2025
25 checks passed

jonathanmos deleted the jmoskovich/flags-feature-batching branch October 9, 2025 16:00

This was referenced Nov 4, 2025

Flags Feature #2979

Closed

Flags Feature #2982

Merged

Uh oh!

Conversation

jonathanmos commented Sep 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What does this PR do?

Motivation

Additional Notes

Review checklist (to be filled by reviewers)

Uh oh!

codecov-commenter commented Sep 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

typotter left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

0xnm left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

typotter left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

jonathanmos commented Sep 25, 2025 •

edited

Loading

codecov-commenter commented Sep 25, 2025 •

edited

Loading