From: David Garamond <lists@...6.isreserved.com>
Date: 2004-07-03T16:40:43+09:00
Subject: Re: slow method searching?

George Ogata wrote:
> g@crash:~$ time ruby -e 'a=(1..2000000).to_a; a.classx'
> -e:1: undefined method `classx' for #<Array:0x401cfebc> (NoMethodError)
> 
> real    0m6.100s
> user    0m5.897s
> sys     0m0.104s
> g@crash:~$ time ruby -e 'a=(1..2000000).to_a; begin a.classx; rescue NoMethodError; end'
> 
> real    0m0.458s
> user    0m0.449s
> sys     0m0.008s
> 
> 
> I'm guessing the time is spent "stringifying" the array when
> outputting exception message.  Note that for shorter arrays, the array
> is actually printed.

I guess you're right.

$ time ruby -e'a=(1..1000000).to_a; a.classx'
-e:1: undefined method `classx' for #<Array:0x4027c18c> (NoMethodError)

real    0m19.999s
user    0m16.680s
sys     0m0.060s

$ time ruby -e'a=(1..1000000).to_a; class Array; def inspect; 
"Array[#{self.length}]" end end; a.classx'
-e:1: undefined method `classx' for Array[1000000]:Array (NoMethodError)

real    0m0.877s
user    0m0.710s
sys     0m0.040s

So I think what needs to be done is make Array#inspect smarter, right? 
There's no way a 1-mil element array will have a string representation 
of < 65 chars.

-- 
dave