- From: Melvin Carvalho <melvincarvalho@gmail.com>
- Date: Mon, 9 Jul 2012 19:26:08 +0200
- To: Evan Prodromou <evan@status.net>
- Cc: public-fedsocweb@w3.org
- Message-ID: <CAKaEYhJM9fJbsKjhf_p6DYLmbygHRFLbV0rpFVORdjFra8hRdQ@mail.gmail.com>
On 9 July 2012 18:15, Evan Prodromou <evan@status.net> wrote: > On 09/07/12 11:41 AM, Kingsley Idehen wrote: > >> On 7/9/12 11:20 AM, Evan Prodromou wrote: >> >>> There's obviously some differences on the subject, but if you insist on >>> HTTP URIs for account identifiers, feel free to change "evan@example.com" >>> to "http://example.com/evan" in the examples. >>> >> you are conflating a URI with an http: scheme URI just because http: URLs >> are prevalent on the Web. >> > I explicitly said "HTTP URIs", so I'm pretty sure I'm not conflating > anything with anything. > > Melvin balked at the Webfinger identifiers I had in my examples, so I > suggested he think of them in terms of HTTP URIs instead. > > I'd rather not stop the discussion at step 1. Makes sense. Lots of practical things we can do here. It's important to get identifiers right, because that's the basis on which everything is built. The good news is that identi.ca is doing this pretty much spot on, already. In a nutshell, a modern HTML document can contain one or more pointers (aka anchors) using the # identifier. The plain # (as facebook users) is a pointer to the top of a document. #me is the most common pointer to someone in a document. #i is used by timbl and toby inkster (probably the 2 most knowlegable people in the web). #this is used by kingsley which is a clever analogy to javascript's 'this' keyword. But you can use whatever you want so long as you are consistent. Each # identifier in a modern html5 doc can have key values associated with it. Understand that and you've got the most important parts of the web of data which is hopefully going to be the next big wave of the web, and vital to http based discovery. It used to be the case that you'd need a separate document in XML for this, and that's exactly what identi.ca does. However, more recently in html5 / rdfa 1.1 you can put the identifiers in the html itself, which makes life a lot easier, in many cases. There's also a serialization in JSON (LD) which developers find useful. As I said identi.ca does this already in quite sophisticated way. sioc:follows <http://identi.ca/user/3#acct> was taken from evan's foaf file. This is exactly the kind of association that works very well. Similarly we can say foaf:name "Evan Prodromou" Now with a bit of dexterous programming we can suddenly do a lookup of "Evan" to get to his data record. Alternatively foaf:nick "evan@identi.ca" Now we can lookup in this way too and also leave this nickname in blog posts etc. See how this gets extended? Note: I didnt mean to come across that I balk at webfinger identifiers, but rather, life is so much easier when you use HTTP. This is essentially the feedback that the IETF is giving the webifinger spec, e.g. with Mark Nottingham's suggestion example.com/.well-known/user/evan as a uniform place to discover things. Truth is we dont yet know what a webfinger identifier will be yet, the best guess is acct:user@host but there's still some ways to go before that is finalized. As I've said it's going to be very useful in dealing with many millions on webmail right now, so im supportive of webfinger. Just the FSW can go so much further. > > > -Evan > > -- > Evan Prodromou, CEO and Founder, StatusNet Inc. > 1124 rue Marie-Anne Est #32, Montreal, Quebec, Canada H2J 2B7 > E: evan@status.net P: +1-514-554-3826 > > >
Received on Monday, 9 July 2012 17:26:35 UTC