From: "NARUSE, Yui" <naruse@...>
Date: 2009-02-11T22:16:19+09:00
Subject: [ruby-core:21987] Re: UTF-16BE is alias for UCS-2BE - is this correct?

Hi,

Rados�aw Bu�at wrote:
> I've played with new encoding stuff from ruby1.9.1 and have noticed
> that UTF-16BE is alias for UCS-2BE. I wonder if it's correct.
> Wikipedia describes ucs-2 and utf-16 on one page but it's clearly
> stated that ucs-2 is only 16bit subset of utf-16 so if I'm thinking
> correctly not all characters represented by utf-16 could be
> represented in ucs-2?

Strictly it is of course incorrect.
But we think "UCS-2BE" encoding is for old data that are tagged as UCS-2BE.

Ruby 1.9.1 doesn't support output UCS-2BE.

> I noticed it because I did some benchmarks and expected that ruby
> optimize indexed access (String#[]) for fixed-length character
> encodings but it looks like it doesn't do that for ucs-2:
> (snip)
> Should I file a bug for that?

So this is intentional.

-- 
NARUSE, Yui  <naruse@airemix.jp>