- From: Michiel de Jong <michiel@unhosted.org>
- Date: Wed, 18 Jul 2012 12:47:10 +0200
- To: Markus Sabadello <markus.sabadello@gmail.com>
- Cc: Melvin Carvalho <melvincarvalho@gmail.com>, public-fedsocweb@w3.org
On Wed, Jul 18, 2012 at 11:29 AM, Markus Sabadello <markus.sabadello@gmail.com> wrote: > What is the "HTML5 data layer"? > - There are the data-* attributes, which are meant for use by JavaScript > within an HTML5 page. > - There is also HTML5 microdata, i.e. simple semantic markup for embedding > data in a page. > - Or do you mean Linked Data, RDF, JSON-LD, that sort of thing. i've started implementing useraddress.net (working title), my cross-origin search engine for the fedsocweb. it works pretty well so far, and while programming yesterday, i realised it doesn't matter which format people use, we should support all of them. In particular, there is an example in the webfinger spec that uses webfinger.net/rel/avatar and some people have adopted this, others haven't. Even when foaf is used, the way to link to it is different across implementations. Other domains don't even seem to support any generic discovery mechanisms, but the data on there is still machine-readable so with a few host-specific rules we can still extract it. All in all, this is not a problem, and i was able to get full name + avatar lookup working on statusnet, twitter, facebook, and google without much trouble. Buddycloud will probably be a bit more work because i probably need to get an xmpp client from somewhere, but friendica and diaspora also look quite straightforward at first glance. about open knowledge vs privacy i'm still in two minds, i think this search engine should be a collaborative cache whose contents is owned by the public domain, much like ur1.ca, but at the same time i don't want to make data findable that people were relying on to be unfindable (even though it was available over public http). Anyway, my conclusion is that even if people use webfinger there are different formats we have to consider, and even if they don't, we can still quite easily route around that with either per-host rules for the identity format, or the user remembering their full http URI, or by discovering new users by going through public friend lists in foaf profiles of others. So yes, useraddress.net will (once i finish all the code for this) be able to deal with pretty much any crazy way people can invent to publish their machine-readable identity, not only by-the-book webfinger. Cheers, Michiel
Received on Wednesday, 18 July 2012 10:47:38 UTC