[Ietf-not43] part way through the matrix

[Andrew Newton]

Edward Lewis wrote:
> 
> Talking to Andy a few days ago about CRISP in a context outside the 
> matrix (I'm trying to do the review relying only on the docs as much as 
> a possible), he mentioned that CRISP is defining a lookup protocol and 
> not a search protocol.  

Uh?  I said that?  I don't think the CRISP working group has made this 
distinction in that its output will be a lookup protocol vs. a search 
protocol.

In my mind (which is a very messy place), it is an access protocol 
consisting of well defined searches and lookups, with a distinction 
being given between a search and a lookup.  And a lookup being a 
distinct value query into a single index.  Anything more, such as 
multiple hits to a single index, non-distinct values (substrings, 
etc...), hits across multiple indexes, is a search.

> I don't see that this distinction is clearly described in the documents 
> (requirements or proposal).  Once Andy explained that, it made a lot of 
> things make more sense.

Well, you might be right about the requirements doc.  IRIS does divide 
the 'searches' from the things can be done with 'lookupEntity'.

> Given that, returning to the matrix, there is a lot of verbiage about 
> "search" as in 3.2.2.1 of the requirements.  I can understand that if I 
> want to look up a book in a library collection, I need to search the 
> shelves for the location assigned it.  But I found the words there 
> confusing (until talking with Andy).

Section 3.2.1.1 says lookup.  Section 3.2.2.1 says search.

> On another topic, I can't figure out one requirement - like, what it 
> means.  It's the one on a query distribution mechanism.  (3.1.11.1 in 
> the requirements.)  What's that?  TCP?  The accompanying text in the 
> requirements isn't very specific.

Higher up the stack.  Something like CIP.

> PPS - And Andy, if you are working all weekend to get the new version 
> out anyway, (;)), can you include some client-server message examples 
> (in dreg)?

Other than Appendix A?  Is there something you want illustrated in 
better detail?

-andy