Ruby on Rails的独特记录

Question

What is the best way to return the unique records from the database, please consider the following :

@users = User.joins('LEFT JOIN subscriptions s ON users.id = s.user_id').includes(:profile).with_deleted.where("...", params[:conditions]).order("users.#{sort_column}" + ' ' + sort_direction).page params[:page]

It has fair amount of joins and conditions and paging. So for now the users are not unique. This is one of the ways to make it unique :

@users = @users.select('DISTINCT(users.id), users.created_at, users.deleted_at , ...')

However this seems to be very slow and I see a lot of explains in the log, which tells me it's not a good query.

I also tried using the uniq method like :

@users = @users.uniq{|u| [u.email]}

This seems to be running the while longer (timeouts the web worker) than the above statement. What is the correct way to de-duplicate the records? Or what would be optimal thing to do in this kind of situation?

There is about 120k users, however only 25 should be displayed at one time, hence the .page method in the first/second statements.

Answer 1

uniq is a method of Array , so it returns the whole bunch of 120k users and iterates through them one by one using ruby to check the condition. This is definetely the wrong way to do filtering.

On the other hand, DISTINCT(users.id) is SQL condition which is handled by your PostgreSQL server. This one should be executed pretty fast. In case it takes some significant time, you should double-check your indexes ( users.id , subscriptions.user_id , profiles.user_id and basically all of the primary and foreign keys as well as attributes which can be queried in your where clause).

ActiveRecord has distinct method to speify a uniqueness constraint, but its implementation simply uses arel to do the same SQL DISTINCT query, so there should be no perfomance difference.

PS: just as a sidenote, there is no need to enumerate all the desired fields of users in your select query. The following should select all the fields of users table for you:

@users = @users.select('DISTINCT(users.id), users.*')

Answer 2

Check the documentation for distinct

Also note that in your third example you are loading all the elements to memory and then making the operations, which is slow and memory hungry.

You should opt to instruct the DBMS for unique records, by using distinct.