Re: ERH on the xml:id test suite

On Sat, Apr 02, 2005 at 06:00:37PM -0500, Paul Grosso wrote:
> 
> Elliotte is posting a bunch of messages to 
> public-xml-id@w3.org.  Rather than forward
> all of them, I suggest people--especially 
> those involved in xml-id implementations
> (e.g., Norm, Richard, Daniel)--take a look at
> the xml-id archives (if you aren't subscribed).

  His message about non-fatal errors raise a serious point:

 http://lists.w3.org/Archives/Public/public-xml-id/2005Apr/0005

  XML parsing is relatively low in the software stack, and we
cannot expect non-fatal error to cross boundaries. Apparently
XOM don't even have any "non-fatal" error reporting API, and
Elliotte hence seems to think XOM cannot be made xml:id compliant.

 maybe the sentence
  "Such errors are not fatal, but must be reported by 
   the xml:id processor to the application invoking it."

 should be softened into
  "Such errors are not fatal, but must be reported if possible
   by the xml:id processor to the application invoking it."


I am not sure "must ... if possible" really fit the existing semantic
of MUST. I would really like the error to be reported, but I understand
that it may not fit all frameworks.

Daniel

-- 
Daniel Veillard      | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
daniel@veillard.com  | Rpmfind RPM search engine http://rpmfind.net/
http://veillard.com/ | 

Received on Sunday, 3 April 2005 11:20:44 UTC