- From: Roy T. Fielding <fielding@apache.org>
- Date: Tue, 29 Apr 2003 15:04:49 -0700
- To: "David Orchard" <dorchard@bea.com>
- Cc: "'Tim Bray'" <tbray@textuality.com>, "'WWW-Tag'" <www-tag@w3.org>
> Seems to me it is in the URI designers choice. If they want widest > possible > global appeal, they use ascii. If they want the widest possible local > appeal, they use localization. This is the ever present lowest(worst?) > common denominator situation. Exactly. I agree that it should be possible for the technology to support. However, I don't believe that use of such identifiers should be recommended for namespaces (though I wouldn't stop someone from doing so because it only hurts them), and I absolutely refuse to accept the notion that IRIs are a "no-brainer". The syntax for URIs is restricted because of the lack of interoperability of identifiers that use characters other than ascii, not because we weren't aware of people wishing to put other characters in identifiers. Even with utf-8 now making such use consistent, we still have the problems of local fonts and local input devices that makes such identifiers significantly less useful. ASCII is still the only set of characters that we know is both displayable and enterable on every general-purpose computing device. It's okay to choose some other design trade-off, but we should not be portraying it as anything other than a risky decision that must be made with the brain fully-engaged. Please don't tell me this is an anglo-centric view. The people I live and work with who are most vocal on this viewpoint are chinese, including my fiancee, who have to deal with such interoperability problems on a regular basis in order to communicate with relatives. ....Roy
Received on Tuesday, 29 April 2003 18:27:41 UTC