fetlife/rollout
58.3
Adequate · 5 November 2026
1.5k
lines of production code
Ruby
primary language
4
measurements over time
What this system is
Features
Add gem entry point for rollout-redis-adapter
The rollout-redis-adapter gem now includes a top-level entry point file that requires the underlying Redis adapter implementation, allowing users to load the adapter via the gem's main namespace.
rollout-redis-adapter/lib · high confidence
Introduce dedicated Redis adapter with history tracking and export capabilities
The rollout-redis-adapter library now provides a dedicated Redis adapter for the Rollout feature flag system. This change introduces a new \Rollout::Adapters::Redis\ class that manages feature states and events using Redis sorted sets, enabling the recording of feature deletions and mutations in a global history log. The adapter includes a codec for serializing feature states and events, and supports exporting feature states along with their associated history, allowing users to audit changes and migrate data more effectively.
rollout-redis-adapter/lib/rollout · high confidence
Introduce standalone Active Record adapter for feature flags
A new \rollout-active\_record-adapter\ gem is available, allowing feature flag state and history to be stored in application database tables (PostgreSQL, MySQL, SQLite) instead of Redis. This adapter depends on Active Record 7.1–8.x and provides a Rails generator to create the necessary \rollout\_features\ and \rollout\_events\ tables. It supports an optional in-process cache (disabled by default) to reduce database reads, and includes a migration tool to copy existing feature state from a Redis adapter to the new database tables.
_rollout-active\record-adapter · high confidence
New Active Record adapter for feature flag storage
This release introduces a new \rollout-active\_record-adapter\ gem, allowing feature flags to be persisted in a relational database instead of Redis. The adapter includes a Rails generator to create the necessary \rollout\_features\ and \rollout\_events\ tables, with specific schema optimizations for MySQL (such as \utf8mb4\_bin\ collation and \mediumtext\ payload limits). It supports an optional in-process feature-state cache to reduce database load, handles feature deletion with global history recording, and provides a migration tool to import existing feature states from other sources.
_rollout-active\record-adapter/lib · high confidence
Behavioural changes
Project structure modernized with explicit Ruby version and CI configuration
The project now explicitly targets Ruby 2.3.0 via a new .ruby-version file and configures Travis CI to test against Ruby versions 2.0 through 2.6 using both Redis 3 and Redis 4 gemfiles. The build system has been refactored from Jeweler to standard RSpec rake tasks, adding specific tasks for running tests in the Redis and Active Record adapter subdirectories. Documentation has been converted from RDoc to Markdown, and the copyright holder has been updated to BitLove, Inc.
(repo-wide) · high confidence
Refactored feature activation API and introduced bulk operations
The Rollout library has refactored its core API to use explicit, granular methods for managing feature states, replacing previous implicit behaviors. Users now interact with \activate\, \deactivate\, \activate\_group\, \deactivate\_group\, \activate\_user\, and \deactivate\_user\ for precise control, alongside new bulk operations like \activate\_users\, \deactivate\_users\, and \set\_users\ to manage multiple users at once. Additionally, a \set\ method allows toggling a feature's global state with a boolean, and \define\_group\ enables custom group logic. The internal state is now encapsulated in \Rollout::Feature\ objects, and persistence is handled via an adapter pattern, supporting features like \multi\_get\ for efficient batch status checks and \clear!\ for full state reset.
lib · high confidence
Rollout 3.2.0 introduces FeatureState boundary and optional logging
This release updates the library to version 3.2.0 and introduces a new \FeatureState\ class that serves as the canonical persistence boundary for feature data, ensuring metadata is canonicalized through JSON. The \Feature\ class is now isolated in its own file and uses \FeatureState\ to manage state, improving ergonomics for feature activation checks. Additionally, an optional logging module is added, allowing users to track feature changes with context support, history limits, and the ability to temporarily disable logging per thread.
lib/rollout · high confidence
Test coverage
Added shared test contracts for rollout backend and history implementations; Added test coverage for Rollout feature state, logging, and backend contracts; Added test suite for the Redis adapter; Added test suite for the new Active Record adapter gem; Updated test suite to RSpec 3 syntax and added comprehensive specs.
Dependencies
Extract Redis and Active Record adapters into separate gems
The project has been restructured to extract the Redis and Active Record storage backends into their own independent gems: \rollout-redis-adapter\ (v0.4.0) and \rollout-active\_record-adapter\ (v0.2.0). The core \rollout\ gem now only depends on the \observer\ library, while the new adapter gems handle their specific dependencies (e.g., \redis\ \>= 4.0 for the Redis adapter, \activerecord\ \>= 7.1 for the Active Record adapter). This allows users to install only the storage backend they need, reducing the core gem's footprint.
(dependencies) · high confidence
Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.
How this codebase got here
Score
- CAI 55 → 58 (+3.8)
- Rubric changed (rubric-2026.10.5 → rubric-2026.11.2) — scores are not directly comparable.
Lenses
- Code Health 74 → 74 (+0.0)
- Architecture 69 → 85 (+16.2)
- Maturity 46 → 63 (+17.0)
- Readiness 54 → 54 (+0.0)
- Security 69 → 54 (-15.0)
Resolved (6)
- Documentation: no project overview (README.md)
- Inconsistent naming pattern: activate_percentage and deactivate_percentage are distinct operations, but they lack a symmetric counterpart to set or update. More importantly, activate_percentage implies setting a specific value, while deactivate_percentage implies removing it. However, Rollout.set(feature, desired_state) exists. It is unclear if activate_percentage is just a convenience for set with a specific state structure. If set handles all state changes, these specific methods are redundant.
- Off-boarding risk: anonymized user #1
- Semantic duplication: Same issue as above but at the Feature object level. user_in_active_users? is likely redundant with active?.
- Semantic duplication: user_in_active_users? appears to be a redundant alias for active? or a subset check that is not clearly distinguished from the general activation state check. Given active? is the standard idiom, the existence of a more verbose, specific-sounding method suggests confusion about whether it checks the user list specifically vs. the overall feature state.
- While adapters are expected to have similar interfaces, the inconsistency lies in the return type or behavior not being visible here, but the naming fetch_feature vs get_feature (if it existed) would be an issue. Here, both use fetch_feature, which is good. However, compare with Rollout.get(feature) vs Rollout.multi_get(features). The adapter layer uses fetch while the top-level API uses get. This is a minor naming inconsistency between the domain object and the adapter interface.
New (1)
- Off-boarding risk: anonymized user #1
Changes since last survey
- 5 commits — 5 feature/other, 0 fixes
By area
- (repo) — 2 commits
- (root) — 2 commits
- lib/rollout — 1 commit
Notable commits
- change: Bump rollout version to 3.3.0
- change: Load feature definitions from config/rollouts
- change: Look for definitions under the current directory, not Rails.root
- change: Merge pull request #205 from fetlife/feature/yaml-definitions
- change: Merge pull request #206 from fetlife/bump-rollout-3.3.0
Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.
Survey your own repository
fetlife/rollout was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.
About this page
- The score is its most recent published measurement, taken on 5 November 2026 at a pinned commit. It is not a live figure and does not change until the project is measured again.
- Measured at commit 147e1c8fa5b32703d616f75ee50561971e178d61 — the exact code this score is about.
- Scored under rubric-2026.11.2 — the same rubric and the same method as every other entry in this index.
- Measured by Canine Code Analyzer (cca.canine.dev) using cca measure-2026.11.2+preprod-5c4e66d9318b.