xml-dev - Re: Possible changes for XML 2nd Edition

[Rick JELLIFFE <ricko@geotempo.com>]

Eldar Musayev wrote:
 
> In short, non-valid characters are errors, but they should not be fatal.

I think they *should* be fatal but at the moment they *cannot* be fatal
due to "friendly" libraries.

I wonder if this is the kind of thing where text/xml should have a slack
behaviour and application/xml should have a Draconian behaviour?

We found that a lot of our Chinese data had bad codes because it
regularly included chunks of ASCII HTML, cut an pasted. Howeever, it
turned out that that "ASCII" HTML in fact  frequently had many A0  (=
  in iso 8859-1)
characters which are not part of a legit Big5 code sequence. It makes me
think that it is good practise to always encode blank characters > 127
using some kind of reference.

Rick Jelliffe

***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************