- From: Yin Leng Husband <yin-leng.husband@compaq.com>
- Date: Wed, 9 May 2001 12:47:18 +1000
- To: Daniel Barclay <Daniel.Barclay@digitalfocus.com>, David Clay <david.clay@oracle.com>
- Cc: xml-dist-app@w3.org
> I believe that this should be added to the issues list. The issue > is that unnormalized payloads cannot predictably be routed based > upon their content. I think the implied solution is to have normalized payloads, and (my assumption is) that normalization be according to the normalization-algorithms (Unicode TR#15) endorsed by W3C's I18N WG. David, I wasn't at the Boston f2f, so please correct me if my assumption is wrong. Regards, Yin Leng -- Yin-Leng Husband Enterprise Integration Practice Compaq Global Services Compaq Computer Corporation -----Original Message----- From: Daniel Barclay [mailto:Daniel.Barclay@digitalfocus.com] Sent: Friday, 4 May 2001 1:22 To: David Clay Cc: xml-dist-app@w3.org Subject: Re: Internationalization issue. David Clay wrote: > ... > I believe that this should be added to the issues list. The issue > is that unnormalized payloads cannot predictably be routed based > upon their content. (You're talking about things like "test" vs. "test" or CDATA sections in XML, or "test" vs. "t%65st" in URIs, right?) I don't think that any filtering can _ever_ be reliable without interpreting the raw data the way it is meant to be interpreted. (It's like trying to check for ".." segments in URLs that will be passed to a shell without checking for backslashes that the shell will interpret. You might catch the URL "http://host/subsection/../.." but you'll miss "http://host/subsection/\.\./\.\.".) Daniel -- Daniel Barclay Digital Focus Daniel.Barclay@digitalfocus.com
Received on Tuesday, 8 May 2001 22:43:07 UTC