From: martin@... (Martin v. Loewis)
Date: 2001-11-12T16:38:34+09:00
Subject: [ruby-talk:24909] Re: XML libraries (Re:  Re: ruby and webservices)

Sean Russell <ser@efn.org> writes:

> How can you possibly read UTF-16 and not be able to write it back out.  

Why is it a problem? Nobody uses XML documents in UTF-16. So *if* you
come across a document in UTF-16, the spec mandates that you should be
able to process it. There is nothing wrong with writing it back out as
UTF-8.

> Further, you have to be able to give it to the User in some useable format. 
> Not all UTF-16 maps to UTF-8

??? It certainly does. It used to be the other way 'round: Not all
UTF-8 maps to UTF-16, but today, UTF-8 and UTF-16 encode the exact
same characters.

> Yeah, I agree.  If Java got one thing right, it was detaching
> characters from bytes, and supporting UTF out of the box.

That is also the approach Python took; you also need a powerful
encoding library to go with the character string type, to convert
between byte strings and character strings.

Regards,
Martin