[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: intelligent searches

Carl Ellison wrote:
> What does it take to ensure SPKI certs are searchable?
> 1) expressed in readable text rather than binary encoding

Yes and no. Available in text is a better expression. It is possible
to index and allow searching on trees of Gzipped files. The fact that
there is some searchable text underlying the binary file is all that's
necessary. We must simply make sure that there's a simple, well
documented way to index the parts of keys that is worth indexing.

> 2) available by URL

Well, yes and no, again. I'm not all that keen on URL's, myself. They
are location dependent, etc. It's important that the thing that
indexes an object can tell you how to pick up an object. But URL's
aren't the be-all-and-end-all of this problem. URN's (and all the
associated political baggage) are the right way to solve this. On the
way to URN utopia, we may need to travel through the URL slums.

> but there needs to be some convention on content -- some format for what I
> call Meaning fields -- to let the searching user specify searches.

This is tricky... but I agree it would need to be done to be able to
do queries against the global key repository like I originally
suggested. Many of the fields will be straightforward (i.e. name,
keyid), and can be derived from existing models for directories and
from the obvious text encoding of certain parts of the key. But the
Meaning field (which would likely be broken out into many pieces, via
Patrik's clustering idea) will take some serious operations experience
to get right.

> I'm inclined to leave that problem for another day -- after we have
> acquired experience with SPKI certs in actual practice.  Do you have
> specific ideas here?

I think standardizing the Meaning field's syntax and semantics is not
an option at this point. We'll definately need operational experience
to even start on it. That's not to say we can't start trying to make
applications that use it -- I believe that in extensible systems like
Whois++ and (hopefully) SPKI, the "client" authors control what goes
into the templates by the ways they make their clients react to
various attributes. However, things need to evolve towards some standards to
get anything accomplished in the long term.

    Jeff Allen <jeff@bunyip.com>   |   For information about Bunyip
Bunyip Information Systems, Inc.   |   send e-mail to <info@bunyip.com>