#microformats 2015-10-26

2015-10-26 UTC
tantek and ben_thatmustbeme joined the channel
#
kevin marks edited /xoxo-sample-code (+55) "add github link" (view diff)
tantek, fuzzyhorns, elf-pavlik, pfefferle, eschnou, adactio, gRegorLove, Garbee, warehouse13, termie, TallTed and KartikPrabhu joined the channel
#
@technotipz Ultimate Guide to Microformats: Reference and Examples http://www.technotipz.com/tips-and-tricks/ultimate-guide-to-microformats-reference-and-examples/?utm_source=ReviveOldPost&utm_medium=social&utm_campaign=ReviveOldPost #tips (twitter.com/_/status/658671863313387520)
Soopaman, fuzzyhorns, gRegorLove, tantek and KartikPrabhu joined the channel
#
@nico_casel Google Structured Data Testing Tool for testing #microdata , #microformats and #RDFa https://developers.google.com/structured-data/testing-tool/ #html5 (twitter.com/_/status/658722303342198785)
Soopaman, fuzzyhorns and tantek joined the channel
#
ben_thatmustbeme aaronpk and I have been working on a jf2
#
ben_thatmustbeme http://microformats.org/wiki/jf2
#
tantek clicks
#
aaronpk most of the quick iterations have been happening on https://github.com/w3c-social/Social-Syntax-Brainstorming/wiki/Minimal-Activity-Stream and in ben's converter implementation
#
tantek perhaps link to those from /jf2
#
ben_thatmustbeme http://stream.thatmustbe.us/?url=<url> and https://github.com/dissolve/socialstream
#
tantek anyone have any better opinions on naming of jf2? or should I just start casually referring to jf2 at TPAC? ;)
#
kylewm fields in jf2 are optionally multivalued?
#
Loqi kylewm: tantek left you a message on 9/18 at 4:02pm: would appreciate your review and comment(s) on the refinement on the updated proposal on this issue: http://microformats.org/wiki/microformats2-parsing-issues#implied_properties_when_an_explicit_class_is_provided
#
Loqi kylewm: tantek left you a message on 9/18 at 4:26pm: all resolved issues with implementation(s) incorporated into the microformats2 parsing spec - take a look, see if you have any qs: http://microformats.org/wiki/index.php?title=microformats2-parsing&diff=65229&oldid=65090&rcid=101616
kevinmarks joined the channel
#
aaronpk kylewm: yeah
#
kylewm well i like that a lot
#
ben_thatmustbeme we are going back and forth on what to do if there is more than 1 item at the top level
#
ben_thatmustbeme so I was just returning an array of objects, aaronpk suggested we contain them in a blank object and put them as children or items within that
#
ben_thatmustbeme i am leaning toward just always containing everything in a top level object representing the page
#
aaronpk my thought is that if there is only one thing on the page, like an h-entry, then it is best if the JSON for the page is just that object
#
aaronpk if there is more than one object on the page, then you've got an implicit collection or feed, so it should be an object with child objects
#
ben_thatmustbeme i would say it makes sense as either just an array or the single object (you are really asking for the item(s) in the page) or as always an object of the page with its items contained in it
#
ben_thatmustbeme alternatively we could make the top level always an array
#
aaronpk we probably need illustrated examples of these options
#
ben_thatmustbeme indeed
tantek joined the channel
#
kylewm top-level array feels icky somehow
#
aaronpk it doesn't play well with strongly-typed languages
#
tantek kylewm: indeed - I think that's why the mf2 json started with items which we expanded to rels etc.
#
tantek lol strongly-typed languages
#
aaronpk java and objective C come to mind as extremely popular typed languages
#
kylewm also python, ruby
#
kylewm puts on pedant hat
#
kylewm mumbles something about static typing and takes off pedant hat
#
aaronpk whatev
#
aaronpk tantek: was curious where "items" vs "children" came from, and why they aren't the same term
#
tantek I think they're both from the microdata JSON design IIRC
#
kylewm top-level document/page object would be a good place to put rel values
#
tantek kylewm: no need for another level of hierarchy
#
aaronpk if rel values apply to the page, and if the page has only one object, doesn't it follow that the rel values apply to that object?
#
tantek that's the point of the mf2 json - the top level of the JSON IS the top-level document/page object
#
tantek aaronpk: it does not
#
tantek also - for usability reasons its best to avoid mixing any kind of rel-based semantics and class-based semantics
#
aaronpk so we do absolutely need the concept that the JSON document represents the page rather than the item on the page?
#
tantek RDFa 1.0 made that mistake (mixing 'rel' and 'property' attributes) and it so totally confused developers that they would markup both incorrectly
#
aaronpk (aka a variation of the httprange14 problem)
#
tantek not really
#
tantek bringing up httprange14 is not actually ever useful
#
tantek as it doesn't impact any real world problems/solutions
#
aaronpk we can talk about it without the name but it's the same thing
#
tantek no it's not
#
tantek you're confusing
#
tantek it
#
tantek range14 is about does a URL represent something real or is it just a URL *of* something real
#
tantek this has nothing to do with that
#
aaronpk it's very similar
#
tantek not at all
#
aaronpk more like htmlrange14
#
tantek hr14 is a philosophical debate - nearly useless
#
aaronpk anyway the question stands
#
ben_thatmustbeme basically range14 is the other side of this, its if it refers to the top level page object or the thing the items[0] represents
#
tantek no its not
#
ben_thatmustbeme s/it refers to /a url refers to /
#
Loqi ben_thatmustbeme meant to say: basically range14 is the other side of this, its if a url refers to the top level page object or the thing the items[0] represents
#
tantek ben_thatmustbeme: the range14 nonsense has to do with the philosophy of real objects vs. representations of them
#
aaronpk fine drop range14
#
aaronpk there is still a question
#
tantek it has nothing to do with is a document the <body> or not
#
tantek etc.
#
aaronpk i don't care what you call it
#
tantek names do matter
#
tantek so do misusing names
#
tantek point being, an HTML doument includes everything you get back in the text/html response body
#
tantek so trying to artificially simplify that to "just one thing" is likely to get you into trouble
#
tantek rel values apply to the whole of the HTML document
#
tantek as in, definitively
#
tantek whereas say <body class="h-entry"> gives a very strong hint that the document is basically there for the purpose of delivering an entry
#
tantek do you see how those two statements are different?
#
aaronpk the problem is that in most common usage, demonstrated in both our indieweb sites as well as twitter and other networks, a URL like http://aaronparecki.com/notes/2015/10/24/1/indieweb represents a single post
#
ben_thatmustbeme as an example been working off of the example on the github page http://stream.thatmustbe.us/?url=https://ben.thatmustbe.me/static/test2.html
#
aaronpk there is a single entry on the page, the entry has a bunch of properties
#
aaronpk it would be nice if the jf2 representation would reflect that
#
tantek right, you can define jf2 in that way if you like
#
tantek that jf2 represents the primary object of the page
#
tantek which is not "the whole page"
#
tantek strictly speaking
#
tantek which is fine
#
tantek mf2 json does attempt to represent all the microformats / rel markup on the entire page - wherever it may be, however many objects there may be
#
tantek it's ok to focus jf2 on a more specific, minimal functionality
#
aaronpk then the follow-up is when there is more than one item on the page, I wanted to make the jf2 representation create an "implicit" top-level object that contains the items
#
ben_thatmustbeme hmm, the issue i can see others complaining it that we were going to do this with rel=alternate to find the jf2 of the page
#
tantek alternates are much more heavily abused than that already ben_thatmustbeme
#
aaronpk the implicit top-level object would have no properties other than "children" or "items" or whatever, which is why i was asking wher ethose names came from
#
tantek e.g. all WordPress installs badly mis-use alternate
#
ben_thatmustbeme if rel relates to the entire document, is rel=alternate to a jf2 that only represents the primary object
#
tantek from post permalinks they say rel=alternate to a feed *of the home page*
#
tantek *to link to
#
ben_thatmustbeme ick
#
tantek aaronpk: hence the problem with any kind of implicit top level object that *sometimes* is there
#
tantek and sometimes not
#
tantek ben_thatmustbeme: yeah
#
tantek aaronpk: let me try to guess a real world use-case - a bunch of h-entry objects on a home page? ;)
fuzzyhorns joined the channel
#
aaronpk yep
#
aaronpk vs someone who has actually marked up a top-levle h-feed
#
ben_thatmustbeme so putting aside the rel alternate issue, perhaps we are going down the wrong track by trying to represent the object at all and should indeed go with only the primary item (first if there is an array of them)
#
aaronpk basically i want those to look the same in jf2, just one doesn't have a name
#
aaronpk since that's basically the first complaint you hear when someone starts consuming microformats feeds
#
ben_thatmustbeme http://stream.thatmustbe.us/?url=https://ben.thatmustbe.me/
#
ben_thatmustbeme tantek, thats exactly what brought this up me parsing my home page
#
aaronpk okay this is updated now https://github.com/w3c-social/Social-Syntax-Brainstorming/wiki/Minimal-Activity-Stream
#
aaronpk specifically check out https://github.com/w3c-social/Social-Syntax-Brainstorming/wiki/Minimal-Activity-Stream#collections
#
aaronpk also added a note on html content https://github.com/w3c-social/Social-Syntax-Brainstorming/wiki/Minimal-Activity-Stream#html-content
fuzzyhorns, Zegnat, KartikPrabhu, Soopaman, gordonb and tantek joined the channel
#
ben_thatmustbeme thinking about the discussion earlier aaronpk, if this is really meant to be a step for social interaction, i think we should contain all top level h-* not just the first is simply for the sake of feed discovery