Optimization and Fixes

[akodkod]

Hi Girish,

I made some changes for you gem, if you do not against.

1. Little bit optimized the speed of methods pluck_h and pluck_s
   1.1 Moved keys.size == 1 from loop
   1.2 Keys was converted to string because HashWithIndifferentAccess always convert keys into string inside self
2. Fixed some RSpec warnings.
3. Changed behaviour of returning value if block given. For example:

result = Todo.pluck_h(:id) do |todo|
  "TODO ID: #{todo[:id]}"
end

# result = ["TODO ID: 1", "TODO ID: 1", "TODO ID: 1"]
# Instead of [{ id: 1 }, { id: 2 }, { id: 3 }]

On my mind, it's more flexible behaviour.

Benchmarks

Main benchmark code (o - means optimized).

require 'benchmark/ips'

require_relative 'config/environment'


Benchmark.ips do |x|
  
  COUNT = 100000
  x.config(time: 10, warmup: 5)


 

  x.report('pluck_to_hash') { Todo.limit(COUNT).pluck_to_hash(:id) }

  x.report('o_pluck_to_hash') { Todo.limit(COUNT).o_pluck_to_hash(:id) }



  x.compare!

end

Warming up --------------------------------------
       pluck_to_hash     1.000  i/100ms
     o_pluck_to_hash     1.000  i/100ms
Calculating -------------------------------------
       pluck_to_hash      1.773  (± 0.0%) i/s -     18.000  in  10.162897s
     o_pluck_to_hash      2.013  (± 0.0%) i/s -     21.000  in  10.439589s

Comparison:
     o_pluck_to_hash:        2.0 i/s
       pluck_to_hash:        1.8 i/s - 1.14x slower

Pluck with lot of keys

x.report('pluck_to_hash') { Todo.limit(COUNT).pluck_to_hash(:id, :user_id, :text, :done, :created_at, :updated_at) }

x.report('o_pluck_to_hash') { Todo.limit(COUNT).o_pluck_to_hash(:id, :user_id, :text, :done, :created_at, :updated_at) }

Warming up --------------------------------------
       pluck_to_hash     1.000  i/100ms
     o_pluck_to_hash     1.000  i/100ms
Calculating -------------------------------------
       pluck_to_hash      0.117  (± 0.0%) i/s -      2.000  in  17.073606s
     o_pluck_to_hash      0.124  (± 0.0%) i/s -      2.000  in  16.182610s

Comparison:
     o_pluck_to_hash:        0.1 i/s
       pluck_to_hash:        0.1 i/s - 1.06x slower

Pluck to Struct

x.report('pluck_to_struct') { Todo.limit(COUNT).pluck_to_struct }

x.report('o_pluck_to_struct') { Todo.limit(COUNT).o_pluck_to_struct }

Warming up --------------------------------------
     pluck_to_struct     1.000  i/100ms
   o_pluck_to_struct     1.000  i/100ms
Calculating -------------------------------------
     pluck_to_struct      0.124  (± 0.0%) i/s -      2.000  in  16.192719s
   o_pluck_to_struct      0.129  (± 0.0%) i/s -      2.000  in  15.466436s

Comparison:
   o_pluck_to_struct:        0.1 i/s
     pluck_to_struct:        0.1 i/s - 1.05x slower

[girishso]

Merged. Thanks @MrEmelianenko for good work!