melvsterrhiaro: from my experience person = website tends to work well enough to start with, but in more complex cases such as complex clients that involve clients, some stuff breaks when applying HTTP headers such as etags, to the wrong thing
melvsterso i wonder do we have a practical way forward, I know what to do with your profile, since it is a dual indieweb / linked data profile (gold star! :)) but let's say I want to link to tantek or aaronpk, can you suggest some ideas?
melvsterso i could say tantek.com : <#this> a Person ; <homepage> <> . ... I just fear <#me> foaf:knows <tantek.com> is going to break lots of things in the future and require having to change the links ... so id rather get it right earlier than later ... it's a bit like getting a tattoo, you dont really want to make a mistake at the start :)
melvsterrhiaro: tl;dr if we agreed a simple heuristic for indieweb such that say <> owl:sameAs <#this> ... life could be a lot easier and we could start to merge the two together ...
melvsterif the #me or the #card or # is there already, we can simply respect that and use it, the question is what happens when there isnt anything there, can we guess intelligently?
melvsterrhiaro: "it'll be a stretch to ask people to add rdf to their page themselves" -- I dont doubt you here, but could you say why? People dont want to follow standards, or it's not fashionable, or the markup is too hard? If we know the reason it's easier to find solutions ...
rhiaroif you have your h-card data in html, why would you add it again as rdf, especially in say a separate [turtle] file? Invisible data is more fragile.
melvsterok well i think you have a good idea to take a survey of what people do already, then maybe we can see what people want to do, and are prepared to do, then maybe make a consensus
melvstereventually everything will have libraries for everything else (whatever survives that is) ... the question in the short term is what makes sense to tackle first with limited resources
rhiaroYeah. I don't think parsing a h-card from a homepage is too resource intensive. Especially as there are several good, well-maintained microformats2 parsers already
melvsterthe thing i like to focus on at the start is (1) because once you link to something it's hopefully relatively stable, changing URIs later is a lot of work
ben_thatmustbemeputting h-* on <body> tag was a really interesting simplification proposed, since a url should take the first h-* item as what it represents putting h-card on the body means it has to and will always be the first and only item on the page
rhiaroben_thatmustbeme: I use h-feed, which contains my h-card and h-entries, which in theory means it should be inferred that the author of the h-feed is the author of all the h-entries
rhiaroso the url that actually returns foaf is rhiaro.co.uk/about (still should return foaf unless I broke it more) and the url I use in author u-url is rhiaro.co.uk/about#me
melvsterthe #amy starts to matter when you do caching, which isnt now for indieweb, and maybe never, but for my client side JS apps it's quickly becoming essential
bblfish it seems to be just for me http://downforeveryoneorjustme.com/www.w3.org
melvster so i could say tantek.com : <#this> a Person ; <homepage> <> . ... I just fear <#me> foaf:knows <tantek.com> is going to break lots of things in the future and require having to change the links ... so id rather get it right earlier than later ... it's a bit like getting a tattoo, you dont really want to make a mistake at the start :)
rhiaro a la view-source:http://hmans.io/following
ben_thatmustbeme reads back the log