How are feature flags managed under IEC 62304 compliance?

Question

I am interested in how feature flags fit into the context of IEC 62304 for medical device software. Specifically, if a feature is deployed but hidden behind a feature flag, does enabling the flag constitute a new software release, or is it considered a configuration change? Would software with a feature flag enabled be considered a different version from when it is disabled?

Dr. Oliver Eidel · Accepted Answer

Feature flags are generally considered a configuration, not a new release of the software. However, you must ensure that all possible configurations enabled by feature flags are tested and validated before release. This means you can't push untested features to production just because they're hidden behind a flag—regulatory approval and validation are still required for any new feature, whether the flag is on or off. For A/B testing or gradual rollouts, both configurations (A and B) must be proven safe and effective. So, the software version itself doesn't necessarily change with the flag, but your validation must cover all intended combinations.

Answer

Feature flags can help with integrating unfinished functionality without exposing it to users, making it easier to manage code branches and releases. But you'll need to show that any configuration you ship is safe, even if the new feature is disabled. So, you still have to do thorough testing and validation before releasing code with new feature flags, even if the features are 'dead code.'

How are feature flags managed under IEC 62304 compliance?

1 comment

2 Answers