Hi,

At Sun, 20 Jan 2008 18:17:54 +0900,
Martin Duerst wrote in [ruby-core:15155]:
> 
> For those not reading ruby-dev, I just wanted to point to
> a series of mails today. In [ruby-dev:33189], Akira Tanaka
> pointed out a non-linear performance problem with
> String#each_char.
> 
> It turned out that this iterator started at the start of the
> string for each next character, very inefficient indeed. This
> is now fixed. Some programs may run much faster now, for others,
> it will not make a difference. It's a typical example of the fact
> that the new String implementation still may need some tuning.

The example code in [ruby-dev:33189] is one of worst cases.  It
needs to scan backward to the head each time in order to detect
char boundaries.

-- 
Nobu Nakada