> Just checking, I believe that the following are all legal URI Templates according to the latest revision:
> Which does lead to the question of whether
> the following are equivalent templates:
I think they should be equivalent -- will need to note that
in the variable name parser.
> Previous drafts used
> varname = (ALPHA / DIGIT)*(ALPHA / DIGIT / "." / "_" / "-" )
> which avoided this ambiguity.
Yes, I gave in to the requests for full i18n support since
some people might want to use them in non-LTR character encoded
On Mar 7, 2010, at 6:20 PM, Martin J. Dürst wrote:
> On 2010/03/07 10:15, Roy T. Fielding wrote:
>> On Mar 6, 2010, at 4:07 PM, Joe Gregorio wrote:
>> Yes, I gave in to the requests for full i18n support since
>> some people might want to use them in non-LTR character encoded
>> documents. *shrug*
> Should that be non-ASCII characters instead of non-LTR characters (which would be RTL, i.e. right-to-left, i.e. Arabic and Hebrew and such)?
Yes and no. Yes, it applies to all non-ASCII encodings.
No, I was not convinced it was useful to have non-ASCII
variable names until I considered the effect it would have
on documenting IRI templates in an RTL language.
That was just my personal tipping point. There were
already requests for the feature in general, but it adds
complexity to all parsers and thus needed justification.