Rollouts

import ScaleBun from '@scalebun/react-native';
ScaleBun.rollouts.<method>(…);

Members#

Reference#

isOn#

ScaleBun.rollouts.isOn(key: string)

Notes#

Rollouts are percentage-based releases: a feature enabled for a share of users rather than all or none. Same delivery path as flags — evaluated on the device from delivered configuration — with a bucket decision instead of a straight boolean.

Bucketing must be stable per user#

This is the property that makes a rollout usable. The same user must land in the same bucket on every evaluation, or they see the feature appear and disappear between sessions — which is a worse experience than either state and makes any measurement meaningless.

Stability depends on the identifier being bucketed. If that identifier changes — because identify has not run yet, or because it is derived from something per-session — the bucket changes with it.

Ordering#

1. 1

   init(), then identify(), then read the rollout

   In that order, for the stability reason above.

2. 2

   Stage the percentage upward, checking between steps

   1%, 10%, 50%, 100% — watching crash rate and performance at each step. A problem found at 1% affects a hundredth as many people.

3. 3

   Keep the kill switch separate from the rollout

   A rollout percentage is a release tool. Turning something off in an emergency is a flag, and you want that to be one action rather than a percentage edit.

Limits and edge cases#

• Propagation is not instant. A percentage change reaches a device with its next configuration fetch.

• Percentages are not exact at small n. A 10% rollout across 50 users will not be five users.

• Reducing the percentage does not reliably remove the same users. Do not use a rollout decrease as a rollback; use a kill switch.

• Evaluated client-side, so the rollout configuration is visible to anyone inspecting the app.

Related#

• Flags & config

• Feature flags SDK guide