On 7/13/07, Jakob Fix <[EMAIL PROTECTED]> wrote:
On 7/13/07, Steve Jolly <[EMAIL PROTECTED]> wrote: > There's some confusion over CRIDs IMO - even in RFC 4078 they get > referred to as URLs. I think it's best to think of them as URIs, > designed to be unique and location-independent. TV-Anytime defines the > concept of a CRI service that (amongst other things) provides mappings > between CRIDs and locators, which could include http, rtsp etc *URLs*. > This gives you the benefits of both time-invariant identifiers and > time-varying locators, at the cost of an extra lookup. welcome to CRIDland!
Wha? Huh? Eh? But now you've woken me up - I (as others are in here) am a big fan of human-readable URLs. http://www.bbc.co.uk/radio3/breakfast/pip/jrjen/ - good. http://www.bbc.co.uk/radio3/breakfast/archive/07/07/10/ - better. The 'jrjen' in this URL (no idea, but I suspect it's an internal ID for the PIP system) isn't easily guessable. A date (in this case, a backwards one) is more guessable. Another example (from the same area): http://www.bbc.co.uk/music/artist/x9qv/ - good http://www.bbc.co.uk/music/artist/elton_john/ - better Of course, other benefits are that Google will love these URLs more. Having said that, after five days I'm understanding the reasons for why the URLs currently work the way they do. And I think it might be partially my job to fix that. Just hoping nobody notices quite yet. (In other news, on Friday I found Matt Cashmore's desk. But he wasn't in. I left him a bizarre sticky note on his monitor, though.) j -- http://james.cridland.net/
