test please delete

[NickFitz]

Quiet again

I'm off out for a bit; see you later, denizens

[NickFitz]

FFS people, are you so obsessed with post counts that you won't hang out here any more if it "doesn't count"?

I thought we were better than that.

I can forgive Diver, because for all I know he's been rushed into hospital with a gangrenous lump of mould hanging from his left shoulder. And BI's probably got dog hairs in his Ethernet port.

But really, all the other TPDenizens who used to flock here, even on a Sunday night - where are you?

I'm getting lonely

I shall use testpleasedelete.com to identify you all, and email (or tweet, or whatever) you when you fail to keep me company

(It's coming along nicely now, since you ask )

(And of course I couldn't email you - the only user data it currently collects is username, CUK user page URL, and CUK avatar URL - and I need to devise a caching system before I can display avatars, to avoid leeching off CUK by deep linking, which would be a horrid thing to do. I am constantly thinking of the good people of CUK as I implement this, to ensure it doesn't damage their interests.)

[NickFitz]

It'll have lovely URLs, though

[NickFitz]

URLs like

http://example.com/NickFitz/posts/from/2008-02-29T18:00:00/to/2008-03-03-02T09:00:07

or

http://example.com/sasguru/post/1

or

http://example.com/posts/mentioning?term=goats&user=BrowneIssue

and so forth... I haven't designed the full URL structure yet, but it will allow for the output of the TPD-QL engine to be a uniquely-addressable URL, meaning that once a query has been defined, a simple HTTP GET will always return the data corresponding to that query in a way that reflects the actual resources. The corollary of this is that it will be possible (if you're geeky enough) to query it by constructing the appropriate URL on the fly, rather than using the TPD-QL UI to build the query

In fact, the TPD-QL will actually be the URL... that which one perceives as the TPD-QL UI will merely be a URL generator

(Although you should build your own clients so as to use the links and/or forms returned in the representation, rather than second-guessing URI construction schemes and building URLs on the fly. I'll be offering representations of resources in HTML (with chrome), XHTML (with chrome, and well-formed), and JSON to start with. Atom is another possibility. Oh, and UTF-8 encoding.)

[NickFitz]

(Oh, and XML - that went without saying to such a degree that I didn't say it )

[NickFitz]

Not forgetting http://example.com/NickFitz/posts/from/yesterday

(http://example.com/NickFitz/posts/from/tomorrow wouldn't fail, but really ought to return no results )

[NickFitz]

Actually, all those examples will have to be along the lines of /user/NickFitz/dot/dot/dot/, but you probably get the idea - this isn't a definitive spec, just drivel

Although I'm serious - just stick /user in front of the username in the examples above, and you'll get a good idea of how it'll work

Still a bit more to do yet, but the URLs are the easiest, yet most satisfying, part

[NickFitz]

BTW, such things as http://example.com/smiley/:wave/usage/by/user/NickFitz are being planned (I was thinking through the /smiley API URLs and standard queries down the pub tonight) but will probably have to wait until version 2. That will probably take another weekend or four.

[NickFitz]

Anyway, having redirected all the work-in-progress to the current home page (you can guess what it says) I'd better crash out in anticipation of the big piece of macadamized road I need to traverse tomorrow

Night, denizens - don't be strangers

[NickFitz]

the URLs are the easiest, yet most satisfying, part

"For by names and images are all powers awakened, and re-awakened" - from the Neophyte Initiation Ritual of The Hermetic Order of the Golden Dawn