"array cannot contain ActiveRecord objects" -- Where is this stated?

[dchapman1988]

👋 Hello all. I just recently wrote a task that iterates over an array, and to my surprised it errored out in production with the error "array cannot contain ActiveRecord objects". My question is, where in the documentation is this stated? I see in the README it states:

collection: return an Active Record Relation or an Array to be iterated over

But I'm not seeing a note that an Array must not contain any AR objects. Am I over looking this somewhere in the documentation?

[etiennebarrie]

It's a safeguard in job-iteration: https://github.com/Shopify/job-iteration/blob/v1.3.6/lib/job-iteration/enumerator_builder.rb#L56
It should error the same way in all environments though, nothing specific to production.

The idea is to catch when one loads Active Record objects and puts them in an Array, because once deployed to production that may lead to loading a few million records and that usually fails at some point.

Ideally you would express the array as a Relation directly, but as a workaround, if you know you're dealing with relatively few records, you can usually make a new relation, e.g.:

def collection
  SomeRecord.where(id: array_of_records.map(&:id))
end

---

Feel free to contribute documentation around this. 🙏

[nvasilevski]

I just wanted to add that often if the task has an array of Active Record objects (array_of_records in the example) - these objects were most likely built using a relation. So for example:

def collection
  canada_users = User.where(country: "Canada").to_a
  User.where(id: canada_users.map(&:id))
end

Can be simplified to

def collection
  User.where(country: "Canada")
end

The only caveat is that in the second example job iteration will append order by id limit 100 to build batches which sometimes may not be the most efficient query due to lack of indexes. But for most cases using an Active Record relation as a collection is the best practise.