In recent versions we have started to collect more registration data from production sites. These changes predominantly relate to usage of certain features such as AI, Databases, authentication mechanisms, and so on.

Historically these have been treated as new features, but I would argue that we should backport them to stables and security branches in the same way that we do for testing changes.

These changes are relatively rare, but contain important information which helps us to make strategic decisions.

One recent example is the decision to drop support for Oracle. Registration data relating to databases was added to Moodle 4.2, but the types of site predominantly using Oracle were likely those running Moodle 4.1 LTS who likely follow the LTS to LTS upgrade path. We therefore have no stats for those users using Moodle 4.1 LTS on Oracle and won't get any until those sites upgrade to Moodle 4.5 LTS.

I would propose that we automatically backport issues relating to registration data in the same that we do for PHPUnit and Behat, including to security releases. This data impacts the way that we discover, design, and deliver meaningful changes to Moodle.