Hello,

Le 30 juil. 07 à 21:35, Peter Saint-Andre a écrit :

Sergei Golovan wrote:
Hi!

Today I've found some interesting commits in ejabberd SVN. It became
using <[CDATA[ in XML output generation.

Is this acceptable according to RFC or no? I can't find any reference
in RFC-3920 except the phrase that character data containing any
predefined entity must be escaped.

RFC 3920 does not forbid CDATA sections. Neither does rfc3920bis. I
don't know whether that's good or bad. Probably it's not especially good...

What about RFC3923 ?
RFC3923 is "End-to-End Signing and Object Encryption for the Extensible Messaging and Presence Protocol (XMPP)". Reading this document carefully, I think CDATA must have at least been implicitely part of XMPP:

From section 3.1 of RFC3923:
   In order to sign and/or encrypt a message, a sending agent MUST use
   the following procedure:
     [snip]
3. Provide the resulting signed and/or encrypted object within an
       XML CDATA section (see Section 2.7 of [XML]) contained in an
       <e2e/> child of a <message/> stanza, where the <e2e/> element is
       qualified by the 'urn:ietf:params:xml:ns:xmpp-e2e' namespace as
       specified more fully in Section 9 below.

In example 3:
    Example 3: Sender generates XMPP message stanza:

   |   <message to='[EMAIL PROTECTED]/orchard' type='chat'>
   |     <e2e xmlns='urn:ietf:params:xml:ns:xmpp-e2e'>
   |   <![CDATA[
   |   Content-Type: multipart/signed; boundary=next;
   |                 micalg=sha1;
   |                 protocol=application/pkcs7-signature

It seems to me that it implies and requires CDATA to be part of XMPP.

You can check by yourself:
http://www.xmpp.org/rfcs/rfc3923.html

--
Mickaël Rémond
 http://www.process-one.net/


Reply via email to