Re: Context info for datatype libraries

[Murata Makoto <mura034@attglobal.net>]

James Clark wrote:

> I think the issue is simply this: should the element-level context used by a
> datatype library be allowed to contain anything other than the namespace map
> and base URI?  Murata-san believes not (correct me if I have misrepresented
> your position).  I am not yet convinced that it is essential to so limit it
> (though I can certainly live with it being so limited).

Here are some examples, which I think are fine.

- a datatype library implementation accesses to an external database (e.g., a list of 
  part numbers) to determine if a given string is really a part number

- ... accesses to the system clock so as to determine 
  if a given string is concerned about the future

- .. accesses to general parsed entity declarations 
  in the document so as to determine if a given string is the name of an external 
  parsed entity.

Some examples, which I would like to disallow.

- a datatype library implementation accesses the path from the root.  If the length 
  of the path is higher than 3, the content cannot have more than 100 characters.

- a datatype library implementation accesses the contents of siblings.  The 
  result of concatenating the contents of elder siblings, the given string, 
  and the content of younger siblings does not exceet 300 characters.

IMHO, it is risky to allow datatype libraries to access any information 
from the document, since the boundary between RELAX NG and datatypes become 
unclear.

James Clark wrote:

> 1) Information that is the same for all elements in a document (let's call
> this document-level context).

I would say "Information independent from the document".  For example, 
I do not want to allow a datatype libary to examine the XML version number.

Cheers,

Makoto