Character entities (was: Re: XHTML2 MIME type)

On Sunday, May 18, 2003, at 15:52 Europe/Helsinki, David Woolley wrote:

>> If you think they are useless, would you accept not declaring any
>> character entities in the presumably upcoming XHTML 2 DTD?
>
> I don't think they are useless, because I don't agree with your 
> hypothesis.

What hypothesis?

> If I did, I would want & type named entities removed from the XML 
> specification.
> It's not then strictly necessary to say that numeric character 
> references
> must be used to represent < etc., but a non-normative note to that 
> effect
> might be advisable.
>
> Including a whole syntactic construct, with a significant amount of
> supporting text, only to be used for 4, redundant, special cases,
> doesn't make sense.  (On that ground alone, I use the legal document
> construction rule that every word has a meaning, to infer that named
> entities can be used in XML applications.)

They can be used when the document author has enough knowledge about or 
control over the application reading the document. For example, named 
entities are used for separating user interface strings from the user 
interface widget descriptions in Mozilla. Also, when a book is typeset 
as a batch run it doesn't matter if the DTD takes some time to parse.

My point is that in the context of Web browsing (no direct control over 
choice of UA, interactive UA, possibly slow network connection, 
possibly low memory and CPU power on client device) requiring UAs to 
process the external DTD subset is a bad idea. The XML spec even 
defines things like  "non-validating XML processor" and "standalone XML 
document" for use in Web browser-like apps. OTOH, overriding the XML 
spec in order to get the effect of processing the DTD without actually 
processing it would be tag soupish.

-- 
Henri Sivonen
hsivonen@iki.fi
http://www.iki.fi/hsivonen/

Received on Sunday, 18 May 2003 11:26:21 UTC