Re: DOCBOOK: Re: Entity Reference Resolution on an HTTP Server

["Steven T. Hatton" <hattons@speakeasy.net>]

On Thursday 26 September 2002 11:29, Norman Walsh wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> / "Steven T. Hatton" <hattons@speakeasy.net> (by way of Steven T. Hatton 
<hattons@speakeasy.net>) was heard to say:
> | Whose job is it to resolve entities referenced within an XML document
> | served out from an http server.
>
> The XML parser.

I guess that begs the question of where the XML is/should be parsed.  The 
browser is to my mind, a good place to do most of this.

> | If the browser is expected to resolve this reference, then how?  Can I do
> | that with CSS?
>
> Yes, the browser is expected to. No, you can't do it with CSS.

I guess some people might frown on a lot of secondary http traffic generated 
by resolving entity references over the web.  Especially the security 
conscious.  It might be nice to have a 'one shot' method.  That is, someone 
hits a DocBook document requiering the DocBook related DTDs, and the user 
agent could say, "Hey, I don't have this. Should I go fetch it?".  If the 
user says "go for it", the browser could then fetch it, and store it either 
in a browser specific location such as ~/.mozilla or in a user specific 
location such as ~/.sgml, or someting like that.

> | My doctype definition looks thus:
> | <!DOCTYPE chapter PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN"
> | "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd";>
>
> In that case, it seems pretty likely that the parser in Mozilla
> doesn't read the external subset (sigh).

I'll run it by the folks on the Mozilla news server.  There may be some things 
I don't know regarding this functionality.  There may be reasons things are 
this way.

> | Now, in a perfect world Mozilla would just go out and read that from
> | www.oasis-open.org.  That don't seem to be happ'nin'.  What's plan 'B'? 
> | I'm running apache.
>
> Off the top of my head, run the document through an XSLT identity
> transformation before sending it out, that'll expand the entity
> references to numeric character refs which should work even without
> reading the external subset.

Scratching head... I was hoping to put of learning XSLT for a week or so.  
This is probably pretty easy. I might just do that.

XEmacs (and I assume Emacs) does a real nice job of convertng to and from 
entity references in HTML.  Do you know if it will do the same with DocBook?

>                                         Be seeing you,
>                                           norm

STH