Suggestion: Option to raise error when record has non-enumerized value?

[jzisser9]

TLDR: A configuration option to cause Enumerize to raise an error when a record with an invalid enumerized attribute would be very helpful!

---

Hello! My team and I were recently investigating an issue that turned out to be due to the way Enumerize behaves with database records with an invalid value. For example, we have a database table with a string column, example, with enumerized values being ['foo', 'bar', 'baz']. We had some records where example was set to baw. We then made an ActiveRecord query:

Thing.where(example: 'baw').count
#=> 70

Thing.where(example: 'baw').last
<Thing:#abc123
...
example: nil
>

This was caused by a data migration we ran on this table before adding the enumerize keyword to the model. The data migration populated records in the database with a value we didn't realize at the time was invalid. So once we added Enumerize to the model, we introduced a bug that causes the behavior to break silently.

It would be really cool if there was a configuration option that causes Enumerize to raise an error when fetching a record which has an invalid value for an enumerized attribute. It probably makes the most sense to default it to false to preserve current behavior, but I'd leave that up to the implementer.

Thanks for your consideration!

[nashby]

Hey @jzisser9! Thanks for the suggestion, but it feels like an edge case. You would encounter pretty much the same issue with a regular model using, for example, an inclusion validation. In both cases, the record wouldn't be valid after the migration.

I think the only difference is that with Enumerize, the example attribute would become nil, whereas with a regular inclusion validation, it would be baw. However, record.valid? would still return false either way.

Personally, I don't think this is something we need to support. During a migration, you can always run record.valid? or something similar to catch issues like this.