UPDATE.md

3/23/17

• Adding Hash#each! See the branch each-bang-hash
• It seems to be significantly slower than the native each. hmmm why?

3/18/17

• Created hyper_iterator-rails as the adapter to Rails, since ActiveRecord implements its own each method
• Added HyperIterator.each!(array) as a backup method. each is a widely implemented method in many gems (e.g. the one we need to deal with is the pg gem). Because of that adding this method that takes an array as argument can be more versatile without monkey patching for each individual library.

3/16/17

• Tweak the two benchmark file and made it work
• This time around, I am dynamically creating new arrays and counting the number of iterations before it blows up. And it's clear that the each! can complete more iteration before it uses all the memory
• Changed each_slice! implementation to avoid using slice!. The execution time comp looks good (seems even faster than the native each_slice). And the memory usage comp looks very good too!

3/14/17

• Testing benchmark in containerized environment to simulate limited allocated memory.
• each! seems to be working fine. I can see gradual memory release in docker stat even not by a large margin.
• Next thing to test would be creating more objects within the iteration to use up more memory. Given that with 4MB allocation, the current ./benchmark/each.rb and ./benchmark/each_bang.rb is running at the edge of using up all memory, creating more objects supposedly would kill the process using each while each! should survive.
• each_slice! implementation is very slow. Possibly because slice! is O(n)?