new :instantiate_only_requested_columns finder option that does not i…

…nstantiate object columns if they were not included in the :select

CHANGELOG

@@ -1,4 +1,7 @@
 0.2.8 (not yet released)
+- new :instantiate_only_requested_columns finder option that does not
+  instantiate object columns if they were not included in the :select
+  - makes large queries against large tables memory-friendly
 - :select option allows column families to be identified using symbols
 - tests and groundwork for mutator flush_interval (asynchronous writes)
   - requires Hypertable 0.9.2.7

lib/active_record/connection_adapters/hypertable_adapter.rb

@@ -251,7 +251,7 @@ def convert_options_to_scan_spec(options={})
                 status, family, qualifier = is_qualified_column_name?(column)
                 family
               end.uniq
-            when :table_name, :start_row, :end_row, :start_inclusive, :end_inclusive, :select, :columns, :row_keys, :conditions, :include, :readonly, :scan_spec
+            when :table_name, :start_row, :end_row, :start_inclusive, :end_inclusive, :select, :columns, :row_keys, :conditions, :include, :readonly, :scan_spec, :instantiate_only_requested_columns
               # ignore
             else
               raise "Unrecognized scan spec option: #{key}"

lib/hyper_record.rb

@@ -357,13 +357,13 @@ def find_by_options(options)
           # puts msg
         end
 
-        convert_cells_to_instantiated_rows(cells)
+        convert_cells_to_instantiated_rows(cells, options)
       end
 
       # Converts cells that come back from Hypertable into hashes.  Each
       # hash represents a separate record (where each cell that has the same
       # row key is considered one record).
-      def convert_cells_to_hashes(cells)
+      def convert_cells_to_hashes(cells, options={})
         rows = []
         current_row = {}
 
@@ -392,7 +392,10 @@ def convert_cells_to_hashes(cells)
             # Make sure that the resulting object has attributes for all
             # columns - even ones that aren't in the response (due to limited
             # select)
+
             for col in column_families_without_row_key
+              next if options[:instantiate_only_requested_columns] && !options[:select].include?(col.name)
+
               if !current_row.has_key?(col.name)
                 if col.is_a?(ActiveRecord::ConnectionAdapters::QualifiedColumn)
                   current_row[col.name] = {}
@@ -410,8 +413,8 @@ def convert_cells_to_hashes(cells)
         rows
       end
 
-      def convert_cells_to_instantiated_rows(cells)
-        convert_cells_to_hashes(cells).map{|row| instantiate(row)}
+      def convert_cells_to_instantiated_rows(cells, options={})
+        convert_cells_to_hashes(cells, options).map{|row| instantiate(row)}
       end
 
       # Return the records that match a specific HQL query.

spec/lib/hyper_record_spec.rb

@@ -280,6 +280,19 @@ module HyperRecord
         qp.misc.should == {}
       end
 
+      it "should only instantiate requested columns when option set" do
+        p = Page.find("page_1")
+        p.name.should == "LOLcats and more"
+        p.url.should == "http://www.icanhascheezburger.com"
+
+        p = Page.find("page_1", 
+          :select => 'name', 
+          :instantiate_only_requested_columns => true)
+
+        p.name.should == "LOLcats and more"
+        lambda {p.url}.should raise_error(::ActiveRecord::MissingAttributeError)
+      end
+
       it "should allow user to specify ROW key as part of initialize attributes" do
         p = Page.new({:ROW => 'row key'})
         p.ROW.should == 'row key'