#microformats 2014-10-06

2014-10-06 UTC
shaners, gRegor`, KevinMarks, Soopaman, tantek, KartikPrabhu and Atamido_ joined the channel
#
shaners Has anyone moved hMedia into mf2-land as h-media?
pfefferle joined the channel
#
tantek ah - sorry - I shoud have answered that here
#
tantek copying from #indiewebcamp for the logs: in practice hMedia never found any interesting consuming applications, so it didn't get moved forward to microformats2
#
tantek basically, any classic microformat that has no consuming applications got dropped.
#
tantek and only those with publicly visible / useful consuming applications got upgraded to microformats2, and even those had every property scrutinized.
#
shaners 👍
kez joined the channel
#
tantek edited /representative-hcard-authoring (+58) "needs update for microformats2" (view diff)
#
tantek edited /representative-hcard-parsing (+58) "needs update for microformats2" (view diff)
kensanata joined the channel
#
tantek edited /microformats2-parsing (+26) "/* parsing a u- property */ or audio for src attribute per KevinMarks realworld publishing http://feed.unmung.com/feed?feed=http%3A%2F%2Ffeeds.wnyc.org%2Fonthemedia. mark as {{new}} for review." (view diff)
#
tantek !tell barnabywalters, tommorris, KartikPrabhu please review "new" http://microformats.org/wiki/microformats2-parsing#parsing_a_u-_property
#
Loqi Ok, I'll tell them that when I see them next
#
tommorris tantek: will have a look at shortly when in office
#
Loqi tommorris: tantek left you a message 29 seconds ago: please review "new" http://microformats.org/wiki/microformats2-parsing#parsing_a_u-_property
#
tommorris actually, looking now. yup. seems reasonable.
#
tommorris any reason audio and not video too?
#
tantek tommorris: see edit comment. real world use-case driven
#
tantek it's under consideration here: http://microformats.org/wiki/h-entry#Properties
#
tantek * u-video - consider special u- parsing rules for <video>
#
tantek but no one is publishing that yet so...
KevinMarks__ joined the channel
#
KartikPrabhu tantek: looks good
#
Loqi KartikPrabhu: tantek left you a message 5 minutes ago: please review "new" http://microformats.org/wiki/microformats2-parsing#parsing_a_u-_property
shaners joined the channel
#
tantek ok I'll give it day for barnabywalters before I remove the {{new}}
pfefferle, robmorrissey, alanpearce, Atamido_, krendil, kez_ and kez joined the channel
#
pradeep edited /hatom-examples-in-wild (+79) "/* new and uncategorized examples */" (view diff)
pfefferle, kez_, xtof and kez joined the channel
#
xtof neuro`michel_v bonjour. Pour info, meetup indiewebcamp improvisé et ouvert par Auli nouvelle venue et motivée iwc. On s’y retrouve ? á la mutinerie 29 rue de Meaux, Paris Bise. http://indiewebcamp.com/IndieWebCampParis
pfefferle, alanpearce, barnabywalters, TallTed, alanpearce_, kez, kez_, gRegor` and Soopaman joined the channel
#
barnabywalters edited /representative-hcard-parsing (+598) "/* representative hCard algorithm */ raised parsing issues" (view diff)
#
KevinMarks See http://feed.unmung.com/feed?feed=http%3A%2F%2Ffeeds.feedburner.com%2Fputthison for an example with video
#
barnabywalters KevinMarks: an example of what?
#
Loqi barnabywalters: tantek left you a message 7 hours, 20 minutes ago: please review "new" http://microformats.org/wiki/microformats2-parsing#parsing_a_u-_property
#
barnabywalters KevinMarks: ah okay I understand now :)
#
barnabywalters what about <audio><source>? and preserving codec information for each of the links?
#
barnabywalters just grabbing <audio src=“”> is easy enough but it doesn’t cover real-world <audio> element usage
#
barnabywalters unless you put the u-audio class on <source> children, and apply the parsing rule to those as well
pfefferle joined the channel
#
KevinMarks How do we cue your feed generator to make podcast catching friendly output?
#
KevinMarks You can have multiple audio formats in a podcast too?
#
barnabywalters I don’t know anything about podcast markup, but if someone can spec out what the RSS <entry> should look like for a post with audio elements in, I’ll update it
KevinMarks, KartikPrabhu, barnabywalters and tantek joined the channel
#
barnabywalters good morning tantek
#
barnabywalters for your review: tantek: for your review:
#
barnabywalters argh
#
barnabywalters pastings
#
barnabywalters http://microformats.org/wiki/index.php?title=representative-hcard-parsing&diff=64580&oldid=64577&rcid=100944
#
tantek second question was answered last night in #indiewebcamp
#
barnabywalters link?
#
tantek short answer: group blogs, where people put the blog in all their h-cards
#
tantek so there are multiple people with the URL of the home page, on the home page
#
barnabywalters tantek: real-world example?
#
tantek and none of them are the representative h-card
#
tantek the point is to avoid such false positives
#
tantek you want examples of group blogs?
#
tantek there's tons of them
#
barnabywalters tantek: one with all of the author’s profiles on the homepage
#
barnabywalters may as well document it if you’re going to base parsing behaviour off that assertion
#
tantek and if/when they all markup their abbreviated bios with h-card then they'll all have URLs to the home page
#
tantek unfortunately I don't have it offhand - from experience at Technorati 2004-2007. KevinMarks can also validate.
#
tantek isn't comparing URLs specified by the URL standard? http://url.spec.whatwg.org/
#
tantek (for the first issue)
#
barnabywalters if so it should be linked to
#
tantek true
#
barnabywalters I can’t see a “comparison”, or even “normalization” heading in there
KartikPrabhu joined the channel
#
barnabywalters RE group blog with profiles on homepage example — you don’t have a monopoly on demanding real-world evidence ;) other people are allowed to require it too
#
barnabywalters there are none representing the group blog case already on http://microformats.org/wiki/representative-hcard-examples
#
barnabywalters the closest thing is multiple conference speakers, which is not equivalent as speakers wouldn’t add the conference site as a url to their h-card
#
tantek in this case I'm trying to avoid false positives, rather than make up a property for something without evidence
#
tantek it's an aspect of being more conservative with protocol/format design
#
tantek the only indirect circumstantial evidence I have is people using URLs on the Twitter profiles that point to group blogs, or their companies, rather than *themsevles*
#
barnabywalters tantek: that’s a useful distinction
#
tantek and look at this another way - if we do allow mere page = u-url matching, then we explicitly *disallow* the behavior of group-blogs with mini-profiles for the people on the home page which link to the home page.
#
tantek maybe that's ok? and I do feel that's one of those generation 2-3 use-cases that we're not going to see a lot of until indieweb stuff is more popular
#
tantek like a family site
#
tantek where only one person is maintaining the site, but everyone in the family has an account on it
#
tantek FWIW - Known has such assumptions built in - that one install might be used by a small group like that.
xtof joined the channel
#
barnabywalters !tell tantek ah yes I understand better now, it’s the liberal-in-what-you-accept conservative-in-what-you-publish principle applied to formats. makes total sense
#
Loqi Ok, I'll tell him that when I see him next
KevinMarks_ and shaners joined the channel
#
KevinMarks__ here's a multi-author known blog wiht hcards: http://launch.orbit.do/
#
KevinMarks__ so http://launch.orbit.do/profile/jason should have u-uid on it for being the profile of the author?
#
KartikPrabhu KevinMarks__ that seems to be what rep-h-card algo says
tantek, caseorganic, Exploter and barnabywalters joined the channel
#
tantek barnabywalters: I'm willing to go with your proposal, while documenting my reservations / fears as an outstanding issue, and see if anything breaks.
#
Loqi tantek: barnabywalters left you a message 2 hours, 48 minutes ago: ah yes I understand better now, it’s the liberal-in-what-you-accept conservative-in-what-you-publish principle applied to formats. makes total sense
#
tantek since no one else seems to confirm my concerns
#
tantek dropping the u-uid requirement makes it easier for authors/publishers to "get it right"
#
tantek as well as gets more real world examples to "just work" immediately, e.g. adactio.com
#
tantek so that's a concrete real world benefit now, that we have to measure against the hypothetical even if highly expected problem scenario
#
tantek does that reasoning make sense to you?
#
barnabywalters tantek: actually your concerns make a lot of sense
#
barnabywalters I wonder if there’d be any downsides to speccing the stricter version but noting that implementations may choose looser requirements if they want
#
tantek that's worse IMO
#
barnabywalters as far as I know there are very few real-world h-cards with u-uid properties
#
tantek leads to less interop
#
tantek it's good that I was able to explain my concerns so that they made sense
#
tantek HOWEVER, your point about no real world examples still stands.
#
tantek well that's another challenge
#
tantek and the counter-point to " very few real-world h-cards with u-uid " is what I said about adactio.com - dropping the u-uid requirement gets more real world examples to "just work" immediately
#
barnabywalters how about simply requiring there to only be one h-card on the page with the same u-url as the page for it to be representative
#
barnabywalters if there’s one with u-url and u-uid == url, that’s the rep. h-card
krendil joined the channel
#
barnabywalters if there are multiple with u-url == url, one with u-uid == u-url == url, that’s the rep. h-card
#
barnabywalters if there are multiple with u-url == url, none are the rep. h-card
#
barnabywalters unless the u-url == rel-me case overrides that
#
tantek that might actually be sufficient - uniqueness of u-url = page
#
barnabywalters I think that covers all the cases
#
tantek assuming no u-uid == u-url == page
#
tantek in summary , just adding one step
#
tantek after the u-uid == u-url == page test
#
tantek if there is only one h-card with u-url == page, use that h-card
#
barnabywalters tantek: yep, if no u-uid == u-url == page found, look for SINGLE u-url == page
#
tantek otherwise there is no definitive representative h-card
#
barnabywalters tbh I’d rather people used the u-url == rel-me markup instead. That way implementations don’t have to worry about URL normalisation and matching
#
tantek ok that sounds like a good very deliberate incremental step forward that both addresses real world examples like adactio.com, *and* provides some guarding against the multi-author blog case
#
barnabywalters that’s what I’m recommending people do in the indiewebify.me sample code if there’s no h-card found
#
tantek even u-url = rel-me markup needs URL normalisation and matching
#
tantek agreed
#
barnabywalters tantek: not if the markup is <a class=“u-url” rel=“me”>
#
tantek but you can't tell that from the parse results
#
barnabywalters the normalisation and matching is to handle discrepancies between what’s on the page and what the user types in
#
barnabywalters tantek: you can’t, but they’re guaranteed to be an exact match
#
tantek and you still have to check == page URL
#
barnabywalters regardless of the URL which was used to locate the page
#
barnabywalters tantek: not according to http://microformats.org/wiki/representative-hcard-parsing
#
barnabywalters the second step doesn’t mention matching the actual page URL
#
barnabywalters talking of normalisation rules, did anyone find a spec for that? it’s actually quite a common topic
#
barnabywalters and ideally one which should be applied consistently
#
barnabywalters tantek: thanks for asking in #whatwg, you just beat me to it :)
#
tantek lol ok
#
tantek so for now, we can go with "parse them first and then compare the serialization" apparently :)
#
barnabywalters as in, parse the URLs? and compare the parts individually?
#
barnabywalters that seems sane
#
tantek yeah
caseorganic joined the channel
#
barnabywalters predictably, PHP’s parse_url function doesn’t parse the URLs as specified
barnabywalters and shaners joined the channel
#
tantek barnabywalters: fyi: http://krijnhoetmer.nl/irc-logs/whatwg/20141006#l-843
#
tantek making sure we capture a permlink for the logs
#
barnabywalters edited /h-card (+70) "/* Properties */ clarified u-uid, u-logo as set apart from u-photo" (view diff)
#
tantek thanks barnabywalters
#
tantek edited /representative-hcard-parsing (+376) "/* Issues */ add proposed resolutions" (view diff)
#
barnabywalters tantek: don’t have the right focus for adding shrewdness features right now so going to write up improved, mf2-compliant representative h-card parsing algo
#
tantek edited /representative-hcard-parsing (+85) "/* Issues */ add IRC log citation of discussion re: comparing URLs" (view diff)
#
barnabywalters I assume the back compat is handled by the h-card backcompat parsing and there doesn’t need to be a separate backcompat codepath in the algo itself?
#
tantek barnabywalters: feel free to write a clean version at /representative-h-card-parsing
#
barnabywalters okay, will do that
#
tantek correct re: backcompat
#
tantek barnabywalters: and then you can replace the "NEEDS UPDATE for microformats2 h-card. " with a link to the new spec
#
barnabywalters yup
#
tantek similar to how http://microformats.org/wiki/hcard says "See latest version: h-card" at the top in a big yellow box
#
tantek cool
#
tantek thanks!
#
barnabywalters edited /representative-hcard-parsing (-16) "Linked to updated version" (view diff)
#
KevinMarks__ empirical URL normalisation can be useful too
#
barnabywalters e.g. should URL comparison involve redirect resolution?
#
barnabywalters if http://waterpigs.co.uk redirects to https://, my site has an h-card with url uid of https://, but someone types in the http:// to a service which does representative h-card parsing, should it correctly follow the redirect to the https:// site?
#
tantek barnabywalters: anything that *loads* the URL should use the destination of the redirect
#
tantek not what someone typed in
#
barnabywalters that’s worth noting in the URL Matching section of the algo
#
tantek why? nothing in there loads a page
#
tantek assumes you already have the page loaded
#
tantek that's outside UI stuff
#
tantek maybe as an FAQ?
#
barnabywalters tantek: then it should be noted that the resolved URL of the page should be used rather than the original fetched URL
#
tantek rather than the requested URL
#
barnabywalters edited /representative-h-card-parsing (+375) "added mf-cleaner implementation" (view diff)
#
barnabywalters barnabywalters/mf-cleaner now implements the new representative h-card algorithm ^^^
#
tantek barnabywalters: with backcompat - does it work for hCards as well?
#
barnabywalters tantek: it just works on the output of my Mf2 parser, so yes
#
tantek as in - is it also an implementation of http://microformats.org/wiki/representative-hcard-parsing ?
#
tantek might as well add it to the top of http://microformats.org/wiki/representative-hcard-parsing#open_source_implementations also :)
#
barnabywalters (or indeed any other PHP microformats2 parser)
Soopaman joined the channel