Re: XML-* [was: ... XML subsetting...]

> Roy T. Fielding wrote [in private]:
>> It seems to me that simple, non-recursive character-replacement
>> entity names are a fine thing and don't have DoS issues.  I don't
>> know why recursive processing is needed in XML.

and Tim Bray replied with this excerpt:

> It would be easy enough to fix too; just rejig the grammar of entity 
> references to disallow any '&' that's not followed by '#', that way you 
> could still give characters nice names.
>
> This would allow another common appliication of entities in documents, 
> which is to have something at the top like
>
> <!ENTITY home 'example.com'>
> <!ENTITY server 'server1'>
> <!ENTITY root 'base'>
>
> and it makes it easier to maintain big XHTML docs by filling them up with 
> many instances of
>
>  <a href='http://&server;.$home;/&root;/foo.html'>.
>
> Or you could be even more puritan and allow entity defs ONLY of the form
>
> '&#' (x[0-9a-fA-F]+ | [0-9]+)
>
> Go ahead and propose it to www-tag  -Tim

Okay, consider it proposed.

....Roy

Received on Thursday, 5 December 2002 21:29:56 UTC