Le lundi 10 septembre 2007, Tomasz Sterna a écrit : > It was brought on a list but I will repeat it: > It WILL break existing client. > > HTML-like way of thinking will not work here because message stanza is > not HTML. > > Let me rewrite your example (skipping the attributes and adding newlines > and whitespace for readability): > > <message> > <body>Hello, > <span>:D</span> > . > </body> > </message> > > There is no element <span/> in the message/body schema, so unknown > sub-element 'span' will be ignored by the client (or the whole message, > by the more strict ones), leaving us with <body>Hello, .</body> CDATA.
Of course there is an element <span/> It's even specified in the XEP-0071 Now, there is no 'rel' attribute. But what does implementation with unknown attribute? They probably ignore them. The 'class' attribute can be used instead We could define some predefined imll-* class <body>Hello <span class="imml-emoticon">:-D</span>. </body> Personally i don't see the interest of the parent emoticon. What garrent the parent emoticon actually exist. Emoticon that doesn't exist should be rendered as text. The text of an emoticon should have enough meaning by itself. The image is just a plus. If the image has a meaning by itself, maybe inband image should be used instead.
signature.asc
Description: This is a digitally signed message part.
