Re: [xdi] colon for values only?

[Markus Sabadello <markus.sabadello@xdi.org>]

Hmm I'll play with it a bit, but in all the discussions we have ever had about syntax, we never had a case where a certain symbol would only sometimes (at the end) have XDI semantics associated with it, and sometimes (in the middle) not.

Or we leave colon as a valid identifier in arcs (e.g. UUIDs) and use something else for value.

Markus

On Tue, Apr 16, 2013 at 9:42 AM, Drummond Reed <drummond.reed@xdi.org> wrote:

Markus, I thought about this and started working on it in the ABNF. The key challenge with not using colons is: a) they are the standard delimiter in IPv6 addresses, and b) they align nicely with URI scheme syntax.

Since they are not allowed as the first character in any subsegment EXCEPT the terminating literal subsegment, it shouldn't be hard to exclude them in parsing, no?

=Drummond 

On Tue, Apr 16, 2013 at 12:33 AM, Markus Sabadello <markus.sabadello@xdi.org> wrote:

While doing some implementation work I ran into a parsing challenge around colons

We have been using them to indicate a value, but we're also using them in UUIDs.

To me this creates ambiguity in the parsing process.

So I propose to use colons ONLY for values.

One idea would then be to use dots . instead in UUIDs, since dots have no syntactical meaning in XDI anyway.

This is completely orthogonal to the current discussions around paired vs. leading vs. trailing symbols.

Example with dots in UUIDs according to "Second Strawman":

=markus<+email>!.uuid.9ce739f0-7665-11e2-bcfd-0800200c0002:/:/
   "markus@example.net"

Joseph would your version look like this? Please correct

=markus+email&.uuid.9ce739f0-7665-11e2-bcfd-0800200c0002:/:/
   "markus@example.net"

Markus