Re: Feedback on draft-gregorio-uritemplate-00

DeWitt Clinton wrote:
> [snip]
> On a related topic, we do need to address the bidi issues that James and
> others have raised.  The more we can help codify I18N scenarios here,
> the more likely people are to avoid painting themselves into corners
> later on.
> 

These don't need to be addressed in the URI Template spec.  Definitely
in the IRI Template spec tho (I really need to get that draft done).

For a *URI* Template, folks who want to encode non-ASCII characters into
a template name should likely do so by percent-encoding the characters
(assuming UTF-8) to produce a proper ASCII-only name.  However, given
that the spec does not define equivalence for template-names, template
processors should probably not assume that a template name
"r%C3%A9sum%C3%A9" is equivalent to "résumé".  The application providing
the replacement values, however, likely could treat these as being the same.

For the IRI template specification, my current thinking is that template
names will be expanded to include the and iunreserved / iprivate range
as defined by RFC3987 and MUST be stored/transmitted in logical order
with no bidi control characters.  The IRI templates MAY be rendered
using the same rules defined for IRI's.

- James

Received on Thursday, 5 October 2006 18:53:30 UTC