Wed, Jul 07, 2004
The Future of Purple: Distributed Transclusions
#
One of my main interests is collaborative tool interoperability. In that vein, I've
participated in many conversations, but I've been actively involved in
two initiatives: PurpleNumbers and IdentityCommons. On the surface,
these two technologies solve different problems, but when you dig a
bit deeper, you can actually combine the two to do some very
interesting things. (1N1)
PurpleNumbers enable granular addressability, which enables you to
attach metadata to finer-grained chunks of data. Instead of having
an author of a document, you have authors for each paragraph of a
document. This is an especially useful concept for collaborative
authoring tools, such as Wikis (hence PurpleWiki). (1N2)
If you can address granular chunks of data, you can also transclude
them. This is generally more useful than transcluding an entire
document. (You are more likely to want to quote a sentence from a
paper than the entire paper.) So, one consequence of tools that
support PurpleNumbers is that they also support granular
transclusions. (1N3)
One of the limitations of PurpleNumbers as currently implemented is
that addresses are only unique in a local context. What we really
want are globally unique, persistent PurpleNumbers. That would allow
me to transclude a paragraph from another blog into this one. If
that paragraph moves, the transclusion should still work. (1N4)
Guess what. XRIs are globally unique, persistent addresses. In fact,
they were specifically designed to address data. (See the excellent
whitepaper,
"The
Dataweb: An Introduction to XDI" for more on this.) In other words,
we can use XRIs for PurpleNumbers. (1N5)
Even better, IdentityCommons is developing schemas for link contracts
that allow you to attach permissions with your data. While
IdentityCommons will use these link contracts for protecting profile
data, they could certainly be used for other things as well, such as
protecting purpled data. (1N6)
Suppose Alice writes a paper, and she only wants Bob and Carl to see
it. She could attach permissions to the entire paper that says as
much. Now Bob wants to write a paper, and he wants to quote a
paragraph from Alice's paper via a transclusion. However, he also
wants to make his paper available to the public. So, he transcludes
Alice's paragraph and makes his paper publically available. When
most people view the paper, they'll see everything except for Alice's
paragraph because the link contract associated with that paragraph
denies them permission. However, when Bob or Carl view the paper,
they see everything, including Alice's paragraph, because they have
permission to view that. (1N7)
More thinking needs to be done. This originally started as a fancy,
but a conversation with GabeWachob over IRC a few months helped convince me
that using XRIs for PurpleNumbers was not only possible, but a good
application. Two issues are: (1N8)
- What should the PurpleNumbers look like? It's the age-old question.
XRI e-numbers are definitely not pleasing to the eye. (1N9)
- What should the numbers do when you click on them? The way it works
right now is that the addresses are named anchors. But this won't
work out-of-the-box with XRIs, because most XRIs are not valid XML
IDs. (1NA)
ChrisDent has
mentioned handles on several occasions as
another direction we could go. We need to examine this possibility
as well. (1NB)
/tech/purple |
Posted at 5:32pm
EEK Speaks
A blog about collaboration, community-building, and the various
goings-on at Blue Oxen
Associates, with occasional digressions on food and other vital
matters.
Subscribe
Comments
Comments disabled until future notice. If you'd like to contact me, use my i-name (=eekim).