#kylewmi wonder if the python parser is more problematic than the others in this regard
#KevinMarksit may be an artefact of line-oriented parsing
#kylewmso... another option would be to collapse whitespace in the test harness, when doing string comparisons
#kylewmand basically punt on the right way for parsers to do it
#tantekedited /picoformats (-122) "removals: hashtags.org corp service site, microsyntax(.)org looks like expired & registered by nothing do with this topic, twitterdata(.)org similarly appears to have been parked after being abandoned" (view diff)
#kylewmand I'd probably vote to go ahead and do that for now, so we can worry about more interesting parsing issues
#tantekedited /picoformats (-4) "/* Generic */ civilities.net 404 - thus removing - assumed dead site, note @-mentions different from @-replies, add #-hashtags as generic example" (view diff)
#tantekedited /microformats2-parsing (+172) "remove leading/trailing whitespace per part of issue whitespace collapsing revisited consensus, and implementation in mf2py" (view diff)
#tantekbtw one way to explicitly include leading/trailing whitespace with with the data element and value attribute
#tantekI'm leaving that in because the only reason you would have whitespace in a quoted attribute value is deliberate
#tantekedited /microformats2-parsing-issues (+78) "note whitespace collapsing revisited 2015-06-08 option 2 resolved by consensus/mf2py impl - drop leading/trailing whitespace" (view diff)
#tantekok I think I'm caught up to consensus + implementation issues
#tanteknext I'm going to start incorporating the strong consensus issues that are TBI
#tantekin particular: " uf2 children inside a classic microformats root class name", " any h- root class name overrides and stops backcompat root", " backcompat classic microformats should only see backcompat properties", " microformats2 root class names should only see microformats2 properties", " implied properties on backcompat parsing unlikely to be intended", " implied properties when an explicit class is provided", " link elements and u-
#tantek!tell csarven re: https://twitter.com/csarven/status/608194508711444480 what's been your personal experience in using microformats2 to markup your HTML, vs. RDFa with/without schemaorg etc.? difficulty / impact on your markup / # of changes / time to do etc.
#csarventantek To be absolutely clear, *for me*, there is hardly any significant difference between writing mf2 and RDFa. In fact, mf2 is so close to RDFa (at least the way I see it), they are virtually interchangeable. Yes, there are plenty of differences if we look closely, but I don't think those are fundamental to picking one over the other. *As I see it*, if one can do either one of those, they can handle the other.
#Loqicsarven: tantek left you a message 4 minutes ago: re: https://twitter.com/csarven/status/608194508711444480 what's been your personal experience in using microformats2 to markup your HTML, vs. RDFa with/without schemaorg etc.? difficulty / impact on your markup / # of changes / time to do etc.
#csarvenDifficulty of vocab use (e.g., looking up a term to use) is about equivalent.
#csarvenclass="p-name" vs. property="foaf:name" --- hardly calls for a debate.
#csarvenAt the end of the day, whether you tell the machine "foo" or "foo:bar", it doesn't make a difference. Bunch of sufficiently unique-enough strings
#tanteksince mf folks don't have much direct experience with using qnames, others have documented the problems
#Loqitantek meant to say: but that's not a qname - qnames are bound with URIs (somewhere else) which is what makes them fragile
#tanteksaying p-name is a "simple" QName is as much of a lie / deception as saying that Facebook's OGP is "simple" RDFa.
#csarvenWell, if you treat it as an URN, the problem goes away.
#tantek"treat as an URN" - something no one does in actual practice with anything
#tantekso that's like saying "if (this hypothetical thing), ..."
#csarvenYour definition of "actual practice" is bound to what's on webpages. I would say that, that is not the best or complete (although I don't disagree that it is a good sample).
#tantekright, the web is my dataset, all else is handwaving
#csarvenTo be clear: Open/Public/Visible Web is your dataset.
#tantekI don't believe in angels on the head of a pin either. What's your point?
#csarvenJust because you can't access or don't want to work with, it doesn't mean that others are not.
#csarvenI'm happy to get back to the main discussion point.
#tanteknah - I like stuff that is citable. that typically means open/public/visible web. have pretty much given up on armchair architecture about uncitable things.
#tantekscience prefers the citable, all else is good material for art
#csarvenIf the QName breaks or fragile, information can be extracted just the same. Even if http://example.org/ behind that example:foo disappears you can still make sense of it. Just as you make sense of p-name in different context, i.e., offloading that interpretation to the scripts.
#tantekthen just use prefixes in the first place, and forget about qnames. it makes the model simpler.
#csarvenUnfortunately we live in a world where not every phenomenon is easily reproducible or even replicable.
#tantekbtw - lots more citations in that Hixie email
#csarvenI think whether the model is simpler or to what extent it is, it is not that clear. The model is arguably "simpler" for your dataset, but I don't think that simplicity is obvious nor is it the case that the fragility of QNames is something to worry about (in a force-able future) - if you think that's the case, show me what breaks and everything falls apart.
#tantekqnames are in the basic category of "YAGNI"
#tanteksome aspects of classic microformats were very simple, other aspects (needing to write special parsing code per new microformat) were not
#tantekwe thought that was a reasonable trade-off, turned out with some more thinking / experience / innovation, it was a trade-off we didn't need to actually make
#csarvenHow is the parsing of mf2 fundamentally different than RDFa?
#tantekI agree there is much potential compatibility between mf2 and RDF(a)
#csarvenDo you think that ... say the World Bank or.... you know NSA and other teams which wear black all day publish their data on webpages? It is even commonly accepted that there is more "invisible" data than visible data.
#tantek"commonly accepted" - science is not a democracy - so "commonly accepted" is never a good argument
#csarvenRight. Is that why /triples suggests "unnecessarily complicated"?
#tantekexcept that makes it clear you don't understand what the p- is
#ZegnatI think csarven’s first statement is the real point here: it is just as easy for an implementer to look up the meaning behind “p-name” as it is to find it behind “foaf:name”. The only real pro for mf2 is in the parsing through prefixes, even if the parser doesn’t know what “name” is, it knows it will be text
#tantekit's just a parsing instruction, not present in the parsed JSON result
#Zegnat(I just spent 20 seconds writing what tantek just did in 2)
#tantekfor consuming code, there are only 5 prefixes h- p- u- dt- e-
#tantekso it makes sense for both publishers and developers to use mf2 - less work, simpler
#csarvenThere is an EAV in place. That's the generalized form of mf2 and RDF(a).
#aaronpkfor parsers, there are only 5 prefixes. for consuming code, there are none, because anything using the parsed result doesn't even see the prefixes
#tantekfor anyone that has legacy systems / investments in "triples" in their backend, yes they can do conversions
#tantekaaronpk is right - once you've run a parser, you don't deal with any prefixes at all
#csarvenIf we are going to bring our calculators out... well, RDF has one way to parse ;) spo. Crawl the graph..
#csarvenNot to forget that, it is practically the same way one would query that graph!
#csarvenIf we look at microformats/Microdata/RDF(a) history, I think we can see how each evolved due to the other.
#tantekcertainly agreed - been saying that for quite some time
#tantekcsarven: and microformats2 is the current latest version that takes into account all the lessons from the previous efforts (microformats/Microdata/RDF(a) history)
#csarvenRoughly: RDF in HTML -> RDFa -> microformats -> Microdata -> RDFa Lite -> mf2 .. ?
#tanteksemantic class names evolved out of modern web designer practices once they started splitting presentation into CSS, and out of HTML, and started using semantic HTML
#aaronpkwonders how much longer that plus.google.com URL will exist ;)
eschnou, csarven, KartikPrabhu and TallTed joined the channel
#tantekapparently MySpace circa 2012-06-15 supported hCard, according to the "Inspect Element" screenshot in this video, which shows a quite readable class="vcard" on a MySpace profile/friends page: https://youtu.be/iApvUMgk5Mo?t=2m47s
#tantekKevinMarks: do you have an example from the "wordpress corpus" that uses rel="bookmark" rel="tag" rel="author" ?
#tantekKevinMarks: also rel="author" was never part of hAtom, but if you find examples of hAtom in the wild that seem to depend on it - please provide their URL(s) so we can make an real-world-based back-compat decision
#tantek!tell kylewm,KevinMarks does mf2py do backcompat parsing of hAtom's class=hentry?
#Loqikylewm: tantek left you a message 3 minutes ago: does mf2py do backcompat parsing of hAtom's class=hentry?
#tantekedited /h-entry (+491) "/* Parser Compatibility */ move rel=tag to parser compat (from proposed), and add rel=author to proposed per input from KevinMarks, awaiting citations to real w" (view diff)
#tantekKevinMarks: since you brought it up again, I upgraded rel=tag from proposed backcompat to part of the backcompat spec ^^^ - obviously it may require more parsing code due to the special treatment of "last segment of the URL"
#Loqitantek meant to say: KevinMarks: since you brought it up again, I upgraded rel=tag from proposed backcompat to part of the backcompat spec ^^^ - obviously it may require more parsing code due to the special treatment of "last path segment of the URL"
#kylewmyeah, bookmark, tag, and author. we added hEntry scoped rel=bookmark backcompat parsing to mf2py
#kylewm@pwcc requested it for bridgy to support better original-post-discovery on wordpress blogs
#tantekupon inspection, looks like rel=author is on the same link as class="url fn n" inside a span with class="author vcard" - thus no need to look at rel=author (existing author vcard backcompat handles it)
#tantekKevinMarks - what's your preferred way to handle (and thus specify) "/?tag=boat" ?
#tantekedited /h-entry (+354) "/* Parser Compatibility */ add notes to rel=author proposal, note examples so far don't need it, class="author vcard" enough for WP default themes" (view diff)
#tantekedited /h-entry (+642) "/* Parser Compatibility */ proposed: time.entry-date[datetime] in the absence of "published", evidence, WP default themes 2011-2014." (view diff)
#tantekKevinMarks, following your musing thought, there's a concrete proposal with citation for "entry-date" ^^^ for backcompat with WP themes 2011-2014.
#tantekI'll leave it to you and kylewm to vote with your code, implementation or lack there of :)
#tantekedited /h-entry (-63) "/* Proposed Additions */ move u-featured to a solid proposed addition, since there are 2+ indieweb sites with multiple posts using it in the wild" (view diff)
ben_thatmustbeme, Left_Turn, KevinMarks_ and benborges joined the channel