#microformats 2016-02-29

2016-02-29 UTC
fuzzyhorns joined the channel
#
tantek edited /microformats2-parsing-issues (+342) "/* exclude style elements before parsing */ drop both style and script when parsing" (view diff)
#
tantek greetings mf2 parser developers, I've made a proposal to resolve the style (and script) tag issue, please review and provide feedback: http://microformats.org/wiki/microformats2-parsing-issues#exclude_style_elements_before_parsing
#
tantek tommorris, kylewm, aaronpk, KevinMarks, et al
#
aaronpk http://microformats.org/wiki/index.php?title=microformats2-parsing&curid=8366&diff=65418&oldid=65356 also solves the camelCase issue on that page, right?
#
aaronpk edited /microformats2-parsing-issues (+152) "/* exclude style elements before parsing */ +1" (view diff)
fuzzyhorns joined the channel
#
tantek ah, yes it does!
#
aaronpk I know I said I wasn't going to write actual code for the php mf2 parser, but I just broke that rule
#
aaronpk but I have it removing script tags now, I think.
#
aaronpk oh darn nevermind, I need to make it search recursively :(
#
aaronpk ok yeah this is hard
#
aaronpk I think I've got it
#
aaronpk woo hoo all test are passing https://travis-ci.org/indieweb/php-mf2/builds/112496040
#
aaronpk https://github.com/indieweb/php-mf2/pull/83
fuzzyhorns joined the channel
#
@ProjectPeachUK We've #played with #microformats. Love the #idea of #marking up our #business data to #machines as well as #humans! #biznoticeUK #cpp (twitter.com/_/status/704130279326289920)
davidmead, fuzzyhorns, Zegnat, netweb, dogada, tantek, adactio and Garbee joined the channel
#
@bittopper Microformats: Empowering Your Markup for Web 2.0 by John Allsopp - https://www.bittopper.com/item/82995851decb8bce65f154869d379ba4a3d79/ (twitter.com/_/status/704295722489815042)
fuzzyhorns, nitot, tantek, TallTed, misa_ and JohnBeales joined the channel
#
tantek edited /microformats2-parsing-brainstorming (+1177) "/* Parse language information */ additionally id, and consider html-lang parsed property name" (view diff)
#
misa__ hello channel, just found this line on the microformats wiki: microdata was explicitly dropped by the W3C (and therefore not part of W3C HTML5) due to a lack of interest by anyone to edit the spec and keep it up to date.
#
kylewm edited /microformats2-parsing-issues (+21) "/* exclude style elements before parsing */ +1" (view diff)
#
tantek misa__: welcome, and yes that's a summary result from the W3C HTML Working Group discussions on microdata
#
misa__ I understand that, and just a note here, I am quite unexperienced regarding these concepts so I would appreciate some guidance from more experienced folks, just this seems a bit outdated as microdata blog seems to be quite a busy place...
#
tantek misa__: it's up to date. there has been no further work on microdata at W3C so it's still just as dead/dropped from a W3C standards perspective
#
tantek people work on all kinds of things outside W3C
#
tantek which is fine too, it just means those things are not W3C standards
#
tantek misa__: if you're interested in the latest work on microdata-like added markup, take a look at microformats2: http://microformats.org/wiki/microformats2
#
@kykyorg Вдогонку и видео дня: Лео с Оскаром в руках рассказывает киноэлите про глобальное потепление: http://kyky.org/microformats/2016-02-29/video (twitter.com/_/status/704346910224678913)
#
tantek microformats2 is an even simpler replacement for microdata
#
@kykyorg И видео дня: Лео с Оскаром в руках рассказывает киноэлите про глобальное потепление: http://kyky.org/microformats/2016-02-29/video (twitter.com/_/status/704347134519349248)
#
misa__ yes I can see that, just deciding with what to get going is somehow overwhelming, then there is schema.org... could you care to share your thoughts in a brief on that?
#
tantek misa__: schema.org is a Google run effort, with some contributions from Microsoft and Yandex. It's not an open standard.
#
tantek I sympathize with the sense of being overwhelmed
#
tantek hence why a lot of us have worked on simplifying things with microformats2
KartikPrabhu joined the channel
#
misa__ so as far as indieweb is concerned this is not something i should care about?
#
misa__ (google's shema.org)
#
tantek correct, it's been pretty much completely ignored
#
tantek there has been some use of OGP / Twitter Cards as a fallback for some use-cases, but schema is largely ignored for being overly complex and unnecessary
#
tantek welcome KartikPrabhu !
#
KartikPrabhu yo!
#
KartikPrabhu was just checking logs
#
KartikPrabhu misa__: you should decide what to use depending on what you'd like to use if for
#
KartikPrabhu Schema.org has not been very useful for indieweb things mainly due to its complexity of parsing and publishing
#
misa__ indieweb seems very right for me, I can comprehend most of its concepts, just this whole thing with microdata/microformats and the rest got me overwhelmed...
#
tantek edited /microformats2-parsing-brainstorming (+141) "/* Parse language information */ first instance of id attributes only as a way to de-dup / uniqueify id attrs at parse time" (view diff)
#
tantek misa__: microdata is ignorable for indieweb. no one is actively using it.
#
KartikPrabhu misa__: I would suggest starting with microformats as they are the simplest ones to publish
#
tantek there may be a few folks publishing a few microdata things experimentally, but it's never gotten any traction in the peer to peer independent web
#
tantek misa__: glad to hear indieweb seems very right for you! come on by to #indiewebcamp and say hello to discuss indieweb concepts :)
#
KartikPrabhu schema.org is so horribly over-thought that it is ignorable too
gRegorLove joined the channel
#
tantek welcome gRegorLove !
#
gRegorLove Howdy
#
ben_thatmustbeme aaronpk: moving that over here
#
aaronpk example URL: http://xray.p3k.io/parse?url=http%3A%2F%2F2015.aaronparecki.com%2Freplies%2F2015%2F10%2F07%2F1%2Famp
#
KartikPrabhu amp;dr ;)
#
aaronpk so notice how I made the value of in-reply-to just the URL, even though it's actually an h-cite on my site
#
aaronpk and then moved the actual h-cite content outside the main entry
#
aaronpk my goal with XRay (and jf2) is to reduce the number of exceptions you have to deal with when consuming a page
#
ben_thatmustbeme hmmm, interesting
tantek joined the channel
#
ben_thatmustbeme so if you had just an embedded object, would it also be just in "refs"
#
aaronpk so rather than if(is object) {...} else if(is url) { ... } it's just always a URL, and if you want you can check if there's extra data about the URL
#
aaronpk not sure about that case yet
#
ben_thatmustbeme or is refs specific to thinks like in-reply-to like-of, etc
#
aaronpk i'm taking the opposite approach we originally took with jf2
#
aaronpk i'm explicitly adding things to the output when there's a reason to, rather than trying to map a complete mf2 document to this output
#
tantek aaronpk, that's how mf2 JSON was built
#
tantek "explicitly adding things to the output when there's a reason to"
#
tantek hence the evolution of how the parsed rel values made it in there
#
aaronpk this is one level above that though
#
tantek so it will be interesting to see if you come to similar/different conclusions
#
aaronpk at the author level
#
tantek so was mf2 JSON
#
aaronpk e.g. someone can put an h-card anywhere on the page, and it will end up who knows where in the mf2 JSON
#
tantek started at the HTML author level
#
ben_thatmustbeme i like the idea of pulling it all out, almost like the refs: section could be completely ignored since you have to fetch the content anyway
#
aaronpk i'm only interested in that h-card if it has explicit meaning that I can consume
#
tantek "put an h-card anywhere on the page" - then the h-card likely has different meanings, so it makes sense for it to show up different places in the mf2 JSON
#
ben_thatmustbeme aaronpk: it looks like you are doing [] specifically for some values but not for others
#
aaronpk actually tantek.com is a great example. There's an h-card as the last child object of the top-level Tantek h-card
#
tantek aaronpk - that's not author-centric (as you claimed originally), that's *consuming* centric ("i'm only interested ... if it has explicit meaning that I can consume")
#
aaronpk I have no idea what that means, so it's not going to show up in the XRay output
#
aaronpk I didn't say author centric, I said author-level
#
tantek but you're not doing author-level either, you're doing consuming-level
#
aaronpk what was your intention of marking up that Rebecca Daniels h-card?
#
tantek it's a reference to a person
#
aaronpk (as a child h-card of your top-level one)
#
tantek as a publisher, it makes sense to markup all your content that's meaningful with established microformats
#
ben_thatmustbeme can we get back on to the point we were discussing?
#
aaronpk there's no other references to it though, so it has no context. For example if there was some other object on the page with a u-url of rebeccadanielsphoto.com then I might know what it's for
#
tantek right, no other context, and that's ok
#
aaronpk and at that point it would show up in the "refs" list
#
tantek all you know is, this is a person that is referenced on this page
#
tantek that's it
fuzzyhorns joined the channel
#
tantek so e.g. if you have a tool that shows you a list of people on a page, you can display them
#
tantek (there are such tools like Operator FF add-on)
#
tantek and that's useful because you can do things like keep a history of where people were mentioned
#
tantek (e.g. in the browser)
#
tantek histories of people mentioned are useful for things like search, auto-complete etc.
#
ben_thatmustbeme *sigh*
#
tantek plenty of applications for even minimal context like that
#
tantek just maybe not your specific application today
#
ben_thatmustbeme perhaps JF2 is evolving to a more specific use case of social rather than just general microformats
#
tantek point is, if it doesn't make sense to your application, you can just ignore it
#
ben_thatmustbeme microformats has a JSON representation already
#
aaronpk ben_thatmustbeme: yeah that's kind of what I'm thinking
#
tantek ben_thatmustbeme: that's how it starts, but as you add more use-cases, you'll likely end up making something very similar
#
aaronpk the problem is when a property can be either an array or a string, then both cases end up needing to code exceptions for
#
tantek e.g. every use-case I listed above for random h-cards on a page *IS* social
#
aaronpk in mf2 json everything is an array, so most of the time you're doing [0] to get the first. but when jf2 makes a property a string if there's only one value, then you have to do a bunch of checks
#
tantek aaronpk: precisely why that design decision was made for mf2 json
#
ben_thatmustbeme what if we just make this one rule aaronpk, as soon as you hit somthing that has a specific URL outside the domain context (non-authoritative content) we move it over to refs. Basically we could do that as the only change to the MF2 JSON and see what we get
#
tantek so consuming code wouldn't have to do "bunch of checks" (or at least fewer)
#
aaronpk i'm not saying that's bad, just what it is
#
tantek right
#
tantek which I'm happy to see the alternative being explored
#
aaronpk so with the XRay output, I made it vocabulary-aware, so that it's easier to consume when you know what your'e consuming
#
aaronpk e.g. "this is an h-entry. if there is a published date, it will always be a string. if it's a reply, you can find all the URLs it's in reply to in the array 'in-reply-to'"
#
ben_thatmustbeme so, one of the biggest complaints i keep hearing is the need to check if something is an array, single item, or object
#
aaronpk also the value of "in-reply-to" will never be an object with this, since if it was an object in the mf2 JSON, that object gets moved down to refs and the URL of the object is the value in the array
#
ben_thatmustbeme shouldn't author: be a single array item then? couldn't you have multi-author posts?
#
tantek where do you hear these complaints ben_thatmustbeme ?
#
aaronpk some of them are from me
#
tantek hah
#
ben_thatmustbeme lol
#
aaronpk but i've heard that from others as well
#
ben_thatmustbeme i have heard others, i do not have citations right now, will try to keep them noted down from now on
#
ben_thatmustbeme and i do rather agree with them, it is sort of annoying
#
aaronpk it's very annoying. annoying enough that i'm encapsulating all this logic into XRay so I don't have to do it again
#
aaronpk I need this for: readers, showing reply context, showing comments/reactions
#
ben_thatmustbeme again, you are assuming only one author ever?
#
aaronpk well so far i haven't seen any multi-author posts
#
aaronpk and even if there was one, 99.9% of all posts i encounter are single author
#
ben_thatmustbeme notices you aren't processing comments either
#
ben_thatmustbeme is that just haven't gotten there yet?
#
aaronpk no not yet. like I said I am only adding things when I want to consume them
#
ben_thatmustbeme comments could likely just get reduced to a list of urls too
#
ben_thatmustbeme unless they comment directly on the site
#
ben_thatmustbeme i know some allow that
#
aaronpk most likely I'm going to make the "comment" property a list of URLs, and the actual comment objects will live in the "refs" below
#
ben_thatmustbeme exactly
#
ben_thatmustbeme just for comments that don't have a URL, what do you do?
#
aaronpk if there's no URL for a comment (including no fragment URL) then I'm just going to drop it, since nothing will be able to do anything with that comment anyway
#
ben_thatmustbeme i don't think thats true, salmention would still work with a comment that doesn't have a url
#
aaronpk stick a fragment URL on the inline comment and then it's useful again
#
aaronpk ben_thatmustbeme: in practice, any consuming code trying to handle something that doesn't have a URL isn't going to end up with good results
#
ben_thatmustbeme good point
#
ben_thatmustbeme i feel like i'm responding just a moment too early to you
#
ben_thatmustbeme :P
#
aaronpk combine that with tantek's earlier suggestion of XRay returning the object inside the fragment identifier and then fragment comments act just like comments with their own URL
#
aaronpk https://github.com/aaronpk/XRay/issues/8
#
ben_thatmustbeme indeed, i'd love for the mf2 parser to be able to do that directly actually
#
aaronpk i guess that's a totally fine job for the mf2 parser
#
tantek yeah!
#
ben_thatmustbeme still not totally sold on all the items that have been dropped, (location, shortlink, name)
#
aaronpk name hasn't been dropped
#
aaronpk but it's only there when it's actually a name
#
ben_thatmustbeme ahhh
#
aaronpk e.g. it gets removed if it is the same (or a subset) of the content
#
ben_thatmustbeme or a subset? that seems... wrong. As most people will reference the title of a post in the content
#
aaronpk er, prefix
#
ben_thatmustbeme okay
#
aaronpk it's what's described on comments-presentation
#
aaronpk now all of a sudden "name" is useful again
#
aaronpk i'll add location soon
#
aaronpk basically every property on http://microformats.org/wiki/h-entry should show up if present
#
ben_thatmustbeme all seems to make sense, looks like uid and logo aren't carried over, but those aren't really needed / prefer url over uid and logo is just photo again
#
aaronpk there is no "logo" on http://microformats.org/wiki/h-entry anyway
#
ben_thatmustbeme indeed, so there is never any x- prefixes or anything parsed
#
aaronpk yeah
#
ben_thatmustbeme trying to think of a good argument for shortlink, i feel like it is needed as an authoriative alternate url
#
aaronpk btw i'm not sure this is actually the best step for jf2, which is why i've just been building this as an API, but this is how I want to consume pages
#
ben_thatmustbeme which is different from other redirects
#
ben_thatmustbeme i know i made a bunch of optimizations with that
#
ben_thatmustbeme some no, some yes
#
ben_thatmustbeme i actually really like the refs: idea
#
ben_thatmustbeme anything non-authoritative becomes SUPER easy to just throw away if you don't want to look at it
#
aaronpk yeah, it's more like you have two options to find out about a URL that's in the in-reply-to or whatever
#
aaronpk you can check the refs property, or go fetch the URL yourself
#
ben_thatmustbeme indeed
#
ben_thatmustbeme i may look at just applying that to a straight mf2 json output to see what it looks like
#
ben_thatmustbeme keeping the "always an array" idea, and just cleaning up all that stuff
#
aaronpk interesting idea
#
aaronpk i think the rule would be if the object has a url property, replace the object with that URL and move the object to the refs array
#
aaronpk btw gRegorLove do you have a sec to review my PR to the php parser? https://github.com/indieweb/php-mf2/pull/83
#
gRegorLove Sure
#
gRegorLove Looks good at a glance, without testing. I think the innerText method should remove the script and style, but I'm not aware of any problems explicitly removing them first, either.
#
aaronpk i think innerText does
#
aaronpk that's what's used for the "value" property
#
gRegorLove Oh, you're stripping it from the 'html' value?
#
aaronpk but the html property is built up with the calls to $node->C14N() which does not remove them
#
aaronpk yes http://microformats.org/wiki/microformats2-parsing-issues#exclude_style_elements_before_parsing
#
gRegorLove Ok
#
gRegorLove Heh. Forgot my own bug report :)
#
aaronpk oh ha
Calli, fuzzyhorns, Left_Turn, KartikPrabhu and Garbee joined the channel
#
tantek edited /Template:MicroFormatCopyrightStatement (+53) "or already have submitted, updating purely for temporal prose accuracy" (view diff)
#
tantek edited /rel-tag (+85) "note rel-tag incorporated into HTML5" (view diff)
tantek, fuzzyhorns, uf-wiki-visitor and mkaply joined the channel
#
mkaply Has anyone else used microformats-shiv? I'm not seeing the results I expect and I'm trying to figure out what I'm doing wrong.
#
mkaply I'm using it against this page.
#
mkaply http://microformats.org/wiki/hcard-examples
#
mkaply Getting no h-cards
#
mkaply https://pastebin.mozilla.org/8861689
#
aaronpk you might need to try with actual HTML, not just wiki text
#
aaronpk also that page looks like it only contains microformats1 objects, not sure if the microformats-shiv library parses those or only microformats2
#
mkaply It's supposed to parse 1. It's been so long since I've touched this stuff. I guess it's time to bring Operator back from the dead.
#
aaronpk well if you're literally parsing the wiki URL, you probably won't find anything there, since it's all escaped HTML
#
mkaply aaronpk: It's parsing the DOM directly, so it should be finding the nodes. Must be something else going on
#
aaronpk that's what i'm saying though, the HTML on that page is stuff like class="vcard">
#
aaronpk it doesn't actually have any microformats in it
fuzzyhorns joined the channel
#
mkaply I tried tantek's page too. same result. i must be doing something wrong. I'll keep looking
KartikPrabhu, MeanderingCode, fuzzyhorns and tantek joined the channel
#
tantek mkaply: not sure. maybe ping mixedpuppy? I know he had tests working.
#
mkaply tantek; i figured it out. I was passing a string as filters. I opened a bug against shiv to handle that.
#
tantek were you able to get it to work without a filter?
fuzzyhorns and Chordachi joined the channel
#
mkaply Yes. It's working now. Oddly if I add the filters: ["h-card"], it hangs the browser. But if I don't specify a filter, I get the h-card
#
tantek that *is* odd. another bug?
#
mkaply I'm debugging now. It's strange because it does work in our tests.
fuzzyhorns and KartikPrabhu joined the channel