Calling #update on a dataset doesn't update the updated_at column

[AlasdairWallaceMackie]

As the title says, this is something I noticed today. Wanted to know if this was intended behavior before I opened an issue. Right now my workaround is is to use #map on a dataset to individually update each row, in order to get their updated_at columns updated as expected.

Snippets:

# Update dataset
ruby-dev >> User.where(id: ids).update(is_admin: false)
=> 5

# Check values for `updated_at` - no change
ruby-dev >> User.where(id: ids).map(&:refresh).map(&:updated_at)
=> [2023-01-13 14:24:46.571127 UTC, 2023-09-28 21:38:28.442335 UTC, 2024-03-19 19:40:03.279699 UTC, 2024-05-14 19:24:06.390158 UTC, 2024-03-19 19:40:03.235151 UTC]

# Update dataset by mapping through each row
ruby-dev >> User.where(id: ids).map { |user| user.update(is_admin: true) }

# Check values for `updated_at` - All have been updated
ruby-dev >> User.where(id: ids).map(&:refresh).map(&:updated_at)
=> [2024-05-29 17:30:12.076985 UTC, 2024-05-29 17:30:14.376477 UTC, 2024-05-29 17:30:13.464115 UTC, 2024-05-29 17:30:15.172423 UTC, 2024-05-29 17:30:16.104969 UTC]

[adam12]

I'm assuming you're using the timestamps model plugin? If so, I'm almost certain this is intended behaviour.

An alternative, if you're using Postgres, is sequel_postgres_triggers plugin from Jeremy. You could do this with your own triggers as well if you're on another database.

[jeremyevans]

As @adam12 wrote, this is expected behavior. Dataset#update is a dataset-level update (no model save hooks), Model#update is a model-level update (with model save hooks).

[AlasdairWallaceMackie]

Thanks for the quick reply! I will close this discussion