Re: Re: sxml:sxml->xml

From: MJ Ray <[email protected]>
Subject: Re: [ssax-sxml] Re: sxml:sxml->xml
Date: Fri, 17 Oct 2003 01:21:43 +0100

> Speaking of bad characters, I recently ran into the problem with the 
> PLT Scheme edition's ssax:xml->sxml hitting a UTF-8 numeric &#6419; or 
> similar and barfing.  IMO, the real bug is that PLT's integer->char 
> doesn't handle anything over 255, but could that sort of entity be 
> handled in another way?

Not exactly a bug, since R5RS doesn't precisely define the
behavior of integer->char.

In order to treat the entity reference portably, we need at
least the following API that has to be customized
by the implementations:

  ucs->char-list :: Integer -> [Char]

    Given the integer of ucs-4 codepoint, returns either
      * a list of characters that encode the unicode character in
        utf-8, if the implementation only supports single-byte characters.
      * or, a list of characters that encode the unicode character
        in utf-16, if the implementation supports up to two-byte width
        characters.
      * or, a list of one character, if the implementation supports
        full UCS-4.

On single-byte character implementations, the output happens to
work, since writing out the character sequence obtained by
ucs->char-list produces a valid utf-8 string.  However, to produce
a valid character entity reference, we need some more APIs.

--shiro




  


-------------------------------------------------------
This SF.net email sponsored by: Enterprise Linux Forum Conference & Expo
The Event For Linux Datacenter Solutions & Strategies in The Enterprise 
Linux in the Boardroom; in the Front Office; & in the Server Room 
http://www.enterpriselinuxforum.com