- From: Roy T. Fielding <fielding@gbiv.com>
- Date: Mon, 15 Sep 2008 23:37:41 -0700
- To: "Phillips, Addison" <addison@amazon.com>
- Cc: Mark Nottingham <mnot@mnot.net>, URI <uri@w3.org>, Joe Gregorio <joe@bitworking.org>, David Orchard <orchard@pacificspirit.com>, Marc Hadley <Marc.Hadley@Sun.COM>
On Sep 15, 2008, at 8:21 PM, Phillips, Addison wrote: >> varname = ALPHA *( ALPHA | DIGIT | "_" ) > > We have pretty good knowledge of what makes a good Unicode identifier. No, we have almost no knowledge of what makes a good Unicode identifier, zero successful deployments of Unicode-based protocol parsing (judging from the security reports), and the most completely bolloxed conceptions of what interoperability means across real distributed applications that don't require every system to have been implemented since the most recent Unicode consortium pub. As such, the sane approach is to restrict the URI templates to very nearly the same set of printable ASCII characters as the URI syntax is restricted, and for the same reasons: we know that every computer in the world can display it in the same context as a URI. That's what matters for the template. What matters for i18n is that the output of the template be in IRI-compatible form, which is what the draft currently defines. ....Roy
Received on Tuesday, 16 September 2008 06:38:21 UTC