RE: [xri] Potential breakthrough

["Drummond Reed" <drummond.reed@cordance.net>]

>>> Wil wrote:
>>>
>>> I agree that not having the sticky star rules does simplify things a 
>>> lot. In effect, we're kind of back to compact syntax, which means that a

>>> global-xref is a two-subseg-XRI-where-first-subseg-is-GCS construct. 
>>
>> Drummond wrote:
>>
>> Just to clarify, it's a "global-xref is
>> n-subseg-XRI-where-first-subseg-is-GCS construct".
>>   
>
>Wil wrote:
>How so? Without sticky stars and bangs, wouldn't the global-xref in 
>@foo+bar*baz be just "+bar"?

[=Drummond] Agreed. I haven't had time to do the ABNF yet, but I think
you're right, global-xref would just be a single subsegment.

>A related question which I don't think was answered: do we support 
>relative subsegments? I.e. can you have @foo refer to @foo*bar by just 
>"*bar"?

[=Drummond] You've asked that question before, and it's a really good one.
My gut says yes, that *subsegments and !subsegments should be relative to
other absolute XRI authorities (i.e., *west or !1234 would be relative to
@ootao, @ootao*office, and =drummond). But we need to think this through.

>If so, we wouldn't have the ability to reference a global-xref 
>relatively because there is no LCS (* or !) in the absolute reference, 
>i.e. @foo cannot refer to @foo+bar just by saying "+bar" because it 
>could be mistaken as an absolute reference.

[=Drummond] Not "mistaken" -- +bar *is* an absolute reference. In fact we've
never had the ability to reference a global-xref relatively because that's
the whole point of it -- it's an absolute identifier by itself. In other
words, XRI lets you join two or more absolute XRIs into another absolute
global XRI, whereas with URIs, you can only join a relative URI to an
absolute URI (the base URI) to create another absolute URI. 

=Drummond