From: Brian Candler <b.candler@...>
Date: 2011-02-09T05:01:16+09:00
Subject: Re: making hash key from arrays

botp wrote in post #980085:
> just joining the fun here..  an inject version
>
> keys.zip(values).inject(Hash.new{|h,k|h[k]=[]}) {|hsh,(k,v)|
> hsh.tap{|h|h[k] << v}}
> #=> {300=>[1, 2], 301=>[1, 3, 4], 302=>[3, 4]}

That's a prime example of using inject purely for obfuscation.

The principal property of inject is that the value which is the output 
of one iteration is fed in as in input to the next iteration. You're not 
using that property here, because you're mutating the same object each 
time. In that case, it's just a complicated way of saying 'each'. You 
may as well just write:

h = Hash.new { |h,k| h[k] = [] }
keys.zip(values).each { |k,v| h[k] << v }
h

Furthermore, if you insist on using #tap, at least do it on the outside 
where it belongs in this case:

Hash.new { |h,k| h[k] = [] }.tap do |hsh|
  keys.zip(values).each { |k,v| hsh[k] << v }
end

Now, if you wanted to do this in a functional way using inject, then you 
would mutate nothing and instead build a new Hash object in each 
iteration:

keys.zip(values).inject({}) do |hsh,(k,v)|
  if hsh.has_key?(k)
    hsh.merge(k=>hsh[k]+[v])
  else
    hsh.merge(k=>[v])
  end
end

That, in my opinion, is the correct way to use inject here - and it's 
reasonably clear when written that way too. The Ruby runtime will end up 
doing a lot of hash copying though.

Regards,

Brian.

-- 
Posted via http://www.ruby-forum.com/.