[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: arbitrary bytes/characters?

• From: John Cowan <cowan@l...>
• To: Peter.Cardona@u... (Cardona, Peter)
• Date: Wed, 17 May 100 23:35:33 -0400 (EDT)

Cardona, Peter scripsit:
> 
> 
> All-
> 
> (Sorry if this is a FAQ; I just subscribed to this list and the link to the
> archives on xml.org is broken. I'm also very new to the XML stuff!)
> 
> I'm building a system that needs to deliver arbitrary byte streams and
> descriptions of those streams to client software. Most of the conversations
> between client and server use XML to encode requests and replies. My
> approach was to embed the byte stream inside a CDATA section of an XML
> document. But this appears to be impossible to do reliably:

It's unreliable for other reasons.  Bytes 0x0-0x1F (except CR, LF, TAB)
aren't legal in XML documents at all.  The only safe strategy is to
encode the bytes into characters; Base64 encoding (4 characters for
each 3 bytes) is probably the best option.

If you won't or can't do that:

> 1. The sequence "]]>" may not appear in byte stream, so I'd have to escape
> it with "]]>". That, of course, adds processing time to both the client
> and server in this communication.

Within CDATA sections, ">" isn't understood.  The best approach
is to encode "]]>" as "]]>]]<[!CDATA[>" or the like.
This terminates the CDATA section, provides "]]" as ordinary character
data, and then starts a new CDATA section beginning with ">".

> 2. If the sequence "]]&gt;" were to appear in my byte stream as actual
> content, then what? I don't see a way to escape the escape.

Within a CDATA section, "]]&gt;" represents itself.

-- 
John Cowan                                   cowan@c...
	Yes, I know the message date is bogus.  I can't help it.
		--me, on far too many occasions

***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@x...&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************

• Prev by Date: Re: arbitrary bytes/characters?
• Next by Date: Re: Syntax and semantics
• Previous by thread: Re: arbitrary bytes/characters?
• Next by thread: land specific and international use of common labels
• Index(es):
  • Date
  • Thread