[microformats-discuss] Re: XHTML-REST -> APIs

[Dr. Ernie Prabhakar]

Hi all,

On Oct 13, 2005, at 12:56 PM, Dr. Ernie Prabhakar wrote:
> Is this making sense yet? :-)  I'm still filling it in, but  
> hopefully people can follow along...

Thanks to all of you who showed up last night for dinner.  I've  
updated the base accordingly:

http://microformats.org/wiki/rest-brainstorming

Its cool to see Dimitri and I have been thinking along similar  
lines.  Two major issues:

a) Its the data[base], stupid!
[no offense, just an American slang phrase, sorry]

I realized that a series of key-value pairs following a common schema  
are merely (duh) database tables, with URIs being 'primary keys'  
allowing relations between them.  The difference is that we don't  
need to completely describe all the data ourselves -- we merely  
provide the basic info and let the server handle all the  
normalization and default values.

Given that, I propose the following normative URI format for XHTML-REST:

http://host:port/base/table/record

I realize the SQL terminology could be slightly misleading, but it  
seems a much closer mapping than classes/instance, and (at least  
naively) encourages the right mindset.  Reactions?

b) OpenSource 1.1

One of the biggest challenges is figuring out the right way to return  
search results. As Tantek pointed out, this is precisely what  
OpenSource 1.1 is supposed to do:

http://opensearch.a9.com/spec/opensearchresponse/1.1/

However, contrary to what I though he said, I don't see any  
specifications for HTML encodings of results.   Does anybody know of  
such, or do we need invent (transpose) them?

-- Ernie P.

------------
Ernest N. Prabhakar, Ph.D. <drernie at opendarwin.org>
Ex-Physicist, Marketing Weenie, and Dilettante Hacker
Probe-Hacker blog: http://www.opendarwin.org/~drernie/