Reference names termination

[Ygg01]

How long do we want to reference names?

Right now current XML5 implementation parse until they hit a ;. That's in my honest opinion too much. Error recovery in that case is moot because if for example I wrote:

   <z link = "&amp forgot to close">...

The parser will consider amp forgot to close">... as character reference. I think it would be ok, that we either:

• Have a set of allowed characters like XML 1.1 definition of allowed char names
• There is a set of delimiting characters that terminate the reference. E.g. ,\n,\t,<,=,>,&.

I'm leaning towards first solution since we allowed unquoted value attributes like <a attr=&amp foo=bar>, but in theory both could work.

[annevk]

Both would allow unquoted attributes as far as I can tell.

[Ygg01]

This issue is essentially closed, since we're implementing full support for HTML5 named entities and character references.