- From: David Brownell <david-b@pacbell.net>
- Date: Fri, 18 Jan 2002 08:36:21 -0800
- To: Joseph Kesselman <keshlam@us.ibm.com>, www-dom@w3.org
> > > ... XML did not officially define a processor type/mode in which no > > > attempt is made to use the external subset. > > > > Eh? I sure think it did. The external subset is an unnamed entity, > > so it falls under the options for a non-validating parser. That was > > the clear intent. > > They said that this was legitimate behavior for a nonvalidating processor. > As far as I can tell, they didn't provide any terminology for > distinguishing this case from that where the attempt is made, and that lack > of terminology has been an ongoing problem for XML users. The parse would be non-validating, whether an attempt was made or not. The problem was ... what? > Especially if we decide to actually make it directly requestable through > the API... which I think might be a good idea; the parser itself is much > better equipped to know which external entity requests are intended for use > as External Subsets than the EntityResolver is. The "org.xml.sax.ext.EntityResolver2" interface (see the SAX website, http://www.saxproject.org, for such draft "extensions 1.1" APIs) makes that information available to the resolver. Not that it can cause PEs to be "skipped"; it can only resolve them to empty content, which is a much simpler thing. Having thought about this a bit, I can still say that I think it's a lot more usable to just have controls that affect all external entitiees of a given type: PE or GE. Attempting more granular control, particularly of PEs, creates needless confusion and potential for error. - Dave
Received on Friday, 18 January 2002 11:37:55 UTC