Fix that #exists? can produce invalid SQL: "SELECT DISTINCT DISTINCT"

The combination of a :uniq => true association and the #distinct call
in #construct_limited_ids_condition combine to create invalid SQL, because
we're explicitly selecting DISTINCT, and also sending #distinct on to AREL,
via the relation#distinct_value.

Rather than build a select distinct clause in #construct_limited_ids_condition,
I set #distinct! and pass just the columns into the select statement.
This requires introducing a #columns_for_distinct method to return the
select columns but not the statement itself.

3 files changed, 15 insertions, 12 deletions

diff --git a/activerecord/lib/active_record/connection_adapters/abstract/schema_statements.rb b/activerecord/lib/active_record/connection_adapters/abstract/schema_statements.rb
index 9c0c4e3ef0..6e1f43cce6 100644
--- a/activerecord/lib/active_record/connection_adapters/abstract/schema_statements.rb
+++ b/activerecord/lib/active_record/connection_adapters/abstract/schema_statements.rb
@@ -706,12 +706,20 @@ module ActiveRecord
       end
 
       # SELECT DISTINCT clause for a given set of columns and a given ORDER BY clause.
-      # Both PostgreSQL and Oracle overrides this for custom DISTINCT syntax.
       #
-      #   distinct("posts.id", "posts.created_at desc")
+      #   distinct("posts.id", ["posts.created_at desc"])
       #
       def distinct(columns, order_by)
-        "DISTINCT #{columns}"
+        "DISTINCT #{columns_for_distinct(columns, order_by)}"
+      end
+
+      # Given a set of columns and an ORDER BY clause, returns the columns for a SELECT DISTINCT.
+      # Both PostgreSQL and Oracle overrides this for custom DISTINCT syntax - they
+      # require the order columns appear in the SELECT.
+      #
+      #   columns_for_distinct("posts.id", ["posts.created_at desc"])
+      def columns_for_distinct(columns, orders)
+        columns
       end
 
       # Adds timestamps (+created_at+ and +updated_at+) columns to the named table.
diff --git a/activerecord/lib/active_record/connection_adapters/postgresql/schema_statements.rb b/activerecord/lib/active_record/connection_adapters/postgresql/schema_statements.rb
index 98916b06a5..8feee23df0 100644
--- a/activerecord/lib/active_record/connection_adapters/postgresql/schema_statements.rb
+++ b/activerecord/lib/active_record/connection_adapters/postgresql/schema_statements.rb
@@ -467,14 +467,9 @@ module ActiveRecord
           end
         end
 
-        # Returns a SELECT DISTINCT clause for a given set of columns and a given ORDER BY clause.
-        #
         # PostgreSQL requires the ORDER BY columns in the select list for distinct queries, and
         # requires that the ORDER BY include the distinct column.
-        #
-        #   distinct("posts.id", ["posts.created_at desc"])
-        #   # => "DISTINCT posts.id, posts.created_at AS alias_0"
-        def distinct(columns, orders) #:nodoc:
+        def columns_for_distinct(columns, orders) #:nodoc:
           order_columns = orders.map{ |s|
               # Convert Arel node to string
               s = s.to_sql unless s.is_a?(String)
@@ -482,7 +477,7 @@ module ActiveRecord
               s.gsub(/\s+(ASC|DESC)\s*(NULLS\s+(FIRST|LAST)\s*)?/i, '')
             }.reject(&:blank?).map.with_index { |column, i| "#{column} AS alias_#{i}" }
 
-          [super].concat(order_columns).join(', ')
+          [super, *order_columns].join(', ')
         end
       end
     end
diff --git a/activerecord/lib/active_record/relation/finder_methods.rb b/activerecord/lib/active_record/relation/finder_methods.rb
index 72e9272cd7..ff825e52c1 100644
--- a/activerecord/lib/active_record/relation/finder_methods.rb
+++ b/activerecord/lib/active_record/relation/finder_methods.rb
@@ -245,9 +245,9 @@ module ActiveRecord
 
     def construct_limited_ids_condition(relation)
       orders = relation.order_values.map { |val| val.presence }.compact
-      values = @klass.connection.distinct("#{quoted_table_name}.#{primary_key}", orders)
+      values = @klass.connection.columns_for_distinct("#{quoted_table_name}.#{quoted_primary_key}", orders)
 
-      relation = relation.dup.select(values)
+      relation = relation.dup.select(values).distinct!
 
       id_rows = @klass.connection.select_all(relation.arel, 'SQL', relation.bind_values)
       ids_array = id_rows.map {|row| row[primary_key]}