[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: Re: Non-infoset

• To: mike@s...
• Subject: RE: Re: Non-infoset
• From: Alessandro Triglia <sandro@m...>
• Date: Mon, 11 Apr 2005 10:16:11 +0200 (CEST)
• Cc: xml-dev@l...

> -----Original Message-----
> From: Michael Kay [mailto:mike@s...] 
> Sent: Monday, April 11, 2005 03:46
> To: 'Alessandro Triglia'; bob@o...
> Cc: xml-dev@l...
> Subject: RE:  Re: Non-infoset
> 
> 
> > Certainly not.  I am trying to get a feel of the existing 
> > views on how much relevance the non-infoset part of an XML 
> > document has for the creator of the document (be it a program 
> > or a person).
> 
> I'm puzzled about the way you ask the question. Surely the 
> things that are
> relevant to the writer of the document are the things that the writer
> believes will be relevant to the readers of the document: 
> nothing more and
> nothing less?


The reason I am referring to the writer is because a reader, in some cases, cannot make any assumptions on the significance of the non-infoset features of a given document, and therefore it has to treat everything as significant.

For example, it may happen that a given human writer, when creating an XML document, defines and uses generic entities but does not place any significance in her use of generic entities, that is, she may define and then reference some entities purely for her editorial convenience, with no special semantics attached to the position and usage of entity references.  For this particular writer and document, the ultimate consumer is supposed to simply expand all the entities (if any) and not infer any meaning from the way entities are used in the document.  From the point of view of the writer, a transformation of the document that lost general entity definition and references could still be "lossless" with respect to the part of the information that the writer considers significant.

However, an intermediate reader will often be unable to determine what non-infoset features (if any) in the document were insignificant for the writer, and therefore it may have to preserve the entire content of the document (including all the non-infoset features). Lossiness seems to depend on the role.  

Alessandro


> 
> Michael Kay
> http://www.saxonica.com/
> 
> 
> 

• Prev by Date: XML Schema for unordered elements
• Next by Date: Re: Fast Infoset: does the name fit?
• Previous by thread: XML Schema for unordered elements
• Next by thread: RE: Re: Non-infoset
• Index(es):
  • Date
  • Thread