Commits on Jun 15, 2016

1. Prefer Hash#[] over Set#.include? for speed …

   Playing with stackprof against codetriage and for an initial run with no cache `Set#include?` was the top called method, something around 8% of execution time spent there.
   
   Did a microbenchmark to see if it would be faster to use a hash:
   
   ```
   require 'benchmark/ips'
   
   require 'set'
   
   set  = Set.new [:foo, :bar]
   hash = {foo: true, bar: true}
   
   Benchmark.ips do |x|
     x.report("set ") {|num|
       i = 0
       while i < num
         set.include?(:foo)
         i += 1
       end
   
     }
     x.report("hash") {|num|
       i = 0
       while i < num
         hash[:foo]
         i += 1
       end
     }
     x.compare!
   end
   
   # Warming up --------------------------------------
   #                 set    215.314k i/100ms
   #                 hash   219.939k i/100ms
   # Calculating -------------------------------------
   #                 set      11.715M (±15.5%) i/s -     56.843M in   5.010837s
   #                 hash     20.119M (±18.2%) i/s -     96.333M in   5.010977s
   
   # Comparison:
   #                 hash: 20118880.7 i/s
   #                 set : 11714839.0 i/s - 1.72x slower
   ```
   
   Yes, it is faster.
   
   Anecdotally when running `RAILS_ENV=production time rake assets:precompile` against codetriage:
   
   Before patch:
   
   ```
   eal    0m18.325s
   user    0m14.564s
   sys    0m2.729s
   ```
   
   After patch:
   
   ```
   real    0m17.981s
   user    0m14.461s
   sys    0m2.716s
   ```

   

   schneems committed Jun 15, 2016
   Copy the full SHA

   f032821 View commit details

   Browse the repository at this point in the history