#dev 2017-04-05

2017-04-05 UTC
[cleverdevil] and [mko] joined the channel
#
loqi.me created /Granola (+168) "prompted by [cleverdevil] and dfn added by [cleverdevil]" (view diff)
#
loqi.me created /Open_mHealth_Schemas (+205) "prompted by [cleverdevil] and dfn added by [cleverdevil]" (view diff)
#
cleverdevil.io edited /Granola (+63) (view diff)
#
cleverdevil.io edited /Health (+448) (view diff)
#
loqi.me created /HumanAPI (+158) "prompted by [cleverdevil] and dfn added by [cleverdevil]" (view diff)
#
cleverdevil.io edited /Health (+15) "/* Ideas */" (view diff)
[eddie] joined the channel
#
[eddie] https://files.slack.com/files-pri/T03QR2B2T-F4UJ1GB7C/screenshot_2017-04-04_21.11.04.png?pub_secret=b099084756&name=Screenshot 2017-04-04 21.11.04.png Yes!!! automated git commits to my repo when my files in my content section changes. The final step is just configuring the micropub endpoint to tweak the YAML frontend data to match my current structure.
#
aaronpk nice!
tantek joined the channel
#
www.boffosocko.com edited /Akismet (+110) "Plugin for Known" (view diff)
gRegorLove, tantek, cweiske and barpthewire joined the channel
#
petermolnar morning
#
Loqi morning!
#
petermolnar note to photo people: Amazon Rekognition is not too mature for photo tagging.
#
petermolnar and I have an interesting observation: none of the big image recognition service (Google Vision API, Amazon Rekognition, Flick; I have not yet tried Watson) takes the in-image GPS in account when trying to guess the contents. I partly understand this, because it's not what you exactly see on the picture, and, of course, it can be wrong or altered, but sometimes it would come extremely handy.
miklb, [kevinmarks], bonfire1 and singpolyma joined the channel
#
@klagefall @fabeblau Schon in Ordnung. Das sollte ein Test sein, ob die Webmentions via Twitter funktionieren. Siehe: http://klagefall.de/notizen/20170330 (twitter.com/_/status/849622687370678272)
[jeremycherfas] and KartikPrabhu joined the channel
#
ben_thatmustbeme trying to get mastodon's docker instances running, no bueno
#
ben_thatmustbeme i just want to fix their mf2
leg joined the channel
#
aaronpk nice
#
aaronpk if it's just rails it shouldn't be too hard to install natively without docker
#
aaronpk how should an event mark up RSVPs it received?
#
aaronpk are they comments with an rsvp property?
miklb joined the channel
#
aaronpk ohh, p-attendee
[kevinmarks] joined the channel
#
[kevinmarks] I pointed out the line to fix in mastodon a while back
#
[kevinmarks] But adding a test to stop it being broken again would be good
#
aaronpk alright, thoughts on this mf2 markup for upcoming.org? http://pin13.net/mf2/?id=20170405153619036
#
aaronpk here's a live URL http://upcoming.tunnlr.xyz/event/test-event-qr4tkf8q54
#
Loqi Test Event
#
aaronpk and parsed http://pin13.net/mf2/?url=http%3A%2F%2Fupcoming.tunnlr.xyz%2Fevent%2Ftest-event-qr4tkf8q54
tantek joined the channel
#
martymcguire[m] looks good from here! the images on the live example have src="undefined", not sure if that's in scope for what you're working on.
#
aaronpk That's just cause I didn't set up s3 keys
#
aaronpk oh hey tantek
#
aaronpk good timing. need your feedback on upcoming mf2
#
aaronpk http://upcoming.tunnlr.xyz/event/test-event-qr4tkf8q54
#
Loqi Test Event
#
aaronpk in particular: the "attendee" property is list of h-cards, this seems to match what's documented on h-event
#
tantek good morning
#
tantek seems like we are gaining critical mass for attendee
#
aaronpk I translated "recommend" to "like", since it seems to have a similar function, and also uses a star icon. however, there is very little information on the page, so i made the "like" property just h-cards!
#
Loqi *yawn*
#
tantek with real world examples
#
tantek (beyond the editor of the spec ;) )
#
aaronpk comments were straightforward, although they are not from elsewhere so i made them h-entry instead of h-cite, and their URLs have fragment IDs
#
tantek I wonder if the like should be an h-cite with an h-card or is that too much
#
aaronpk that would require me to add too much markup
#
aaronpk nice thing about comments is the fragment IDs parse through XRay properly! http://upcoming.tunnlr.xyz/event/test-event-qr4tkf8q54#comment-1
#
Loqi [Aaron Parecki] Here’s a comment
#
tantek no, h-cite is for something that's posted somewhere else and it's not
#
tantek whoa
#
tantek like + h-card is interesting / novel
loicm_, deathrow1 and gRegorLove joined the channel
#
ben_thatmustbeme can someone help me try to pick out any issues with mf2 on this https://ben.thatmustbe.me/static/test1.html
#
ben_thatmustbeme its a local dev i'm running it from so its not pretty
#
ben_thatmustbeme i think the mention is kinda funky
#
KartikPrabhu ben_thatmustbeme: which mention?
#
ben_thatmustbeme there is a mention of @bob
#
KartikPrabhu aah
#
KartikPrabhu ben_thatmustbeme: you want the u-url and p-nickname inside the h-card
#
KartikPrabhu otherwise they become the u-url and p-nickname of the parent with h-card inside
#
KartikPrabhu seE: http://pin13.net/mf2/?url=https%3A%2F%2Fben.thatmustbe.me%2Fstatic%2Ftest1.html
#
ben_thatmustbeme i think thats better now
#
KartikPrabhu yeah looks right
#
ben_thatmustbeme now to fix the individual page
#
ben_thatmustbeme err individual post page
#
ben_thatmustbeme https://mastodon.social/@mattl/1602791
#
Loqi [mattl] @LogicalDash Thanks :)
#
ben_thatmustbeme h-cites are not inside the h-entry
#
KartikPrabhu yeah the inreply-to won't work
#
ben_thatmustbeme ick, i can't quite figure out how to get the reply context to be a chain, right now i have it just replying to all of the previous feed
#
ben_thatmustbeme https://ben.thatmustbe.me/static/test2.html
#
Loqi [Ben Roberts] @bob weird that i can't reply from other websites
#
ben_thatmustbeme so it looks like everything in the reply-chain is being replied to directly :/
#
gregorlove.com edited /accessibility (+639) "update dfn to cover physical environments, add stub section with links" (view diff)
#
tantek ben_thatmustbeme: the other way to look at this is as a markup example that may need another way of indicating replying etc.
#
tantek one of the vague principles in microformats is to be flexible to fit the markup of the page
#
gregorlove.com edited /2017/Bellingham (+296) "/* Where */ Venue!" (view diff)
[kevinmarks] and [manton] joined the channel
#
tantek.com edited /IndieWebSummit (-11) "-r" (view diff)
#
ben_thatmustbeme tantek: trying not to change ordering of items, just classes
#
ben_thatmustbeme more likely to get the pr accepted
#
ben_thatmustbeme i think i have it worked out
#
ben_thatmustbeme everything outside of the direct context and reply now are just h-cite
#
tantek ben_thatmustbeme: exactly, not change order of markup
#
tantek ideally microformats should just add class names to existing markup and "just work"
#
tantek so when there's a use case that's not quite satisfied with that, we have to figure out alternatives
#
tantek e.g. like we have overtime with /authorship
#
ben_thatmustbeme yes, its the way it was written it was adding markup to ALL ancestors
#
aaronpk i've had to add a few elements to upcoming to make it work
#
ben_thatmustbeme and so i had to work out a way to target only the closest ancestor
#
tantek aaronpk, hopefully just wrapping spans?
#
aaronpk and a common pattern i'm finding that i struggle with is how author avatars are marked up
#
ben_thatmustbeme yeah, sometimes you need to add a wrapper here or there
#
aaronpk yeah mostly just wrapper classes. i'm trying to not change anything fundamental
#
tantek aaronpk: interesting, those are likely worth documenting
#
tantek in case we have to do an "avatar discovery" mini-spec similar to /authorship
#
aaronpk yeah the pattern is basically you want the photo hanging to the left of the comment, and sometimes the markup doesn't allow that
#
aaronpk like, there's no way to wrap both the photo and the name in an h-card without also encompassing some of the comment text
#
tantek like there is no div that contains both the photo and the comment?
#
aaronpk right
#
aaronpk well there is, but it also contains the whole comment
#
tantek aaronpk what does the alt of the photo have?
#
tantek if the alt has the name, you can get it from there
#
aaronpk but then you don't get the author URL
#
aaronpk or in this case the nickname also
#
tantek the photo is not linked to the author URL?
#
aaronpk correct
#
tantek that's. unusual
#
aaronpk i guess
#
aaronpk either way you're still losing the ability to express something
#
aaronpk here's how it looks https://media.aaronpk.com/Screen-Shot-2017-04-05-12-05-19.png
#
tantek um, nearly every social network links people's photos to their profile
#
tantek so I would guess it's an omission
#
tantek *accidental
#
aaronpk here's the markup https://media.aaronpk.com/Screen-Shot-2017-04-05-12-05-19.png
#
aaronpk oops https://media.aaronpk.com/Screen-Shot-2017-04-05-12-05-35.png
#
aaronpk with expanded h-card https://media.aaronpk.com/Screen-Shot-2017-04-05-12-06-29.png
#
aaronpk which contains p-nickname
#
aaronpk so there's no way with this markup to enclose the image URL, author URL, name and nickname in an element
#
aaronpk moving the author name and URL to the image solves everything except the nickname
#
tantek interesting. I feel like I may have run into this exact problem with my own notes markup
#
aaronpk i have with mine as well
#
ben_thatmustbeme someone want to review https://ben.thatmustbe.me/static/test2.html again
#
Loqi [Ben Roberts] @bob weird that i can't reply from other websites
#
tantek it's probably ok to not bother with "nickname" because it really is account name
#
ben_thatmustbeme make sure i'm not missing anything
#
tantek rather than actual nickname that people would call you in person
#
ben_thatmustbeme i didn't add that
#
ben_thatmustbeme but i can remove it
#
aaronpk should i not mark up the nickname then?
#
tantek yeah I wouldn't bother, unless it's an actual nickname
#
ben_thatmustbeme username != nickname
#
tantek as part of the URL it's already an implicit account name / username
#
tantek right
#
ben_thatmustbeme okay, updated with that fix
#
ben_thatmustbeme any others
#
ben_thatmustbeme ?
#
ben_thatmustbeme well its re-uploading that now
#
aaronpk what's the child h-card in the comment for?
[chrisaldrich] joined the channel
#
ben_thatmustbeme it was already there
#
aaronpk oh it's the mention
#
aaronpk in the text
#
aaronpk k
#
aaronpk there's a lot of children objects on the main object, what are they?
#
aaronpk also what's the link to this on mastodon so i can see how it looks with css?
[cleverdevil] joined the channel
#
gregorlove.com edited /2017/Bellingham (+587) "archive Initial Planning notes" (view diff)
#
gregorlove.com edited /Planning (-317) "Bellingham planned" (view diff)
#
tantek nice
#
tantek home page!
#
loqi.me created /O2 (+167) "prompted by tantek and dfn added by [cleverdevil]" (view diff)
#
gRegorLove Looks like it's already there
#
loqi.me edited /P2 (+9) "[cleverdevil] added "[[O2]]" to "See Also"" (view diff)
#
cleverdevil.io edited /O2 (+19) (view diff)
[keithjgrant] joined the channel
#
[keithjgrant] CSS Grid is going to make it tricky to wrap microformats around the right content in some instances
#
[keithjgrant] (that markup up above made me think about that)
#
tantek really?
#
[keithjgrant] Yeah. All grid items in a grid must be sibling elements
#
[keithjgrant] http://codepen.io/keithjgrant/pen/OpGGej?editors=1100
#
[keithjgrant] (at least until Subgrids get added to the spec, but that will be a while)
#
[keithjgrant] similar issue w/ flexbox too, in theory, but I think practical examples would be much more rare
[jeremycherfas] joined the channel
#
KartikPrabhu [keithjgrant]: until sub-grids show up one could use grid-items as div-containers with content in them
#
[keithjgrant] yeah, there are workarounds. But if you want to guarantee alignment of, say, avatars in a series of comments, there won't be a way to wrap an MF class around both the avatar and the comment that appear in each row of the grid
#
www.boffosocko.com edited /O2 (+112) "link and WordPress category; github version" (view diff)
#
[keithjgrant] avatars not the best example, since you can probably hard-code a set width on those. but some other piece of text that might be variant in length: domain, timestamp, etc
#
www.boffosocko.com edited /P2 (+85) "P2 resolved posts" (view diff)
#
KartikPrabhu [keithjgrant]: hmm yeah, I see the problem now
#
ben_thatmustbeme aaronpk: this is an local dev env
#
ben_thatmustbeme with css it looks exactly the same, i didn't change anything really
#
ben_thatmustbeme as to what all the extra children are,
#
ben_thatmustbeme this is just 5 direct replies, and this is the link to the 3rd one
#
ben_thatmustbeme so it has its context as the p-in-reply-to but also has just a cite of what that replies to
#
ben_thatmustbeme similarly comments on comments are not cites instead of it thinking that everyone directly replied to this
arush joined the channel
#
aaronpk alright, i'm happy withthe mf2 on upcoming event pages. now to move to venue pages.
#
tantek oh! have they restored venue pages? or are these just new venue pages?
#
aaronpk new venue pages
#
aaronpk i don't think the archive is part of this repo
#
aaronpk hm, already hit a snag
#
aaronpk the venue page links to the foursquare URL for the venue, but doesn't have a link to itself
#
aaronpk so adding u-url to the foursquare link makes it look like the canonical venue URL is foursquare
#
tantek aaronpk <data class="u-url" value=""></data>
#
aaronpk :sigh: k
#
tantek lol
#
tantek or if you really want to get all canonical
#
tantek <data class="u-url u-uid" value=""></data>
#
aaronpk this jade template thing is pretty funny
#
aaronpk it's kind of like i'm creating a data structure that gets turned into html that gets turned back into a data structure
#
tantek wish there was a way to build a relative URL that stripped any ? or #
#
tantek aaronpk, welcome to dataception
#
aaronpk k venue pages are done http://upcoming.tunnlr.xyz/venue/magnolias-corner-h6rd5ovt8a
#
Loqi Magnolia's Corner
#
aaronpk now user pages
#
ben_thatmustbeme aaronpk: does that make sense for the mastodon stuff?
#
aaronpk ben_thatmustbeme: it's just hard to see what that page is supposed to look like without CSS
#
aaronpk ben_thatmustbeme: like, which post on that page is the primary post? it's hard to tell with no css
[eddie] and KartikPrabhu joined the channel
#
aaronpk "weird that i can't reply from other websites" seems to be the primary post? and the things above that are reply context, and below are comments?
arush joined the channel
#
loqi.me created /global_timeline (+28) "prompted by tantek and dfn added by tantek" (view diff)
#
loqi.me edited /timeline (+22) "tantek added "[[public timeline]]" to "See Also"" (view diff)
#
aaronpk ben_thatmustbeme: I don't understand why the post with "hi there @broberts" is a child of the h-entry when it's the top post on the page
#
gregorlove.com edited /2017/Bellingham (+1360) "/* Participating */ travel and lodging" (view diff)
#
gRegorLove Here's a mastodon.social thread for reference: https://mastodon.social/@sarahjeong/1653278
#
Loqi [sarah jeong] @chr are likes really florps
arush and [kevinmarks] joined the channel
#
aaronpk oh another challenge is with date ranges
#
aaronpk there is no good solution here because the date range is generated by a library, so adding mf2 markup requires support by that library
#
aaronpk i added support specifically for this to my php library
#
aaronpk so i'm adding hidden <time> elements with start/end instead
#
aaronpk alright, i think i have everything done now
#
gregorlove.com edited /2017/Bellingham (+194) "/* Notes */" (view diff)
#
aaronpk ben_thatmustbeme: you got mastodon running locally right? can you try subscribing to @aaronpk@aaron.pk and tell me where it crashes?
#
tantek huh? can't you just subscribe to a URL?
#
aaronpk lol
#
aaronpk it uses webfinger to look up the atom feed for a user
#
tantek I used to be able to in statusnet
#
tantek so that should theoretically still work
#
aaronpk so it first looks at aaron.pk/.well-known/host-meta with the account provided
#
aaronpk to find the atom feed
#
aaronpk then there is some other stuff about salmon and signatures i can't figure out
#
aaronpk and that's when i run away screaming
#
tantek i.e. it would do hub+feed discovery on the URL and if it liked it it just subscribed
#
tantek yeah
#
tantek salmonprotocol--
#
Loqi salmonprotocol has -1 karma
#
aaronpk this does not appear to be very error tolerant
#
tantek waits for someone to deploy a Known instance called WellKnown
#
aaronpk er, i missed a step in the decription
#
tantek and then offer free accounts to all former "The Well" users
#
aaronpk first it looks at http://aaron.pk/.well-known/host-meta to find the webfinger endpoint
#
aaronpk then does a webfinger request
#
aaronpk boom https://github.com/upcoming/upcoming-www/pull/47
#
aaronpk diff: https://github.com/upcoming/upcoming-www/pull/47/files?w=1
#
www.boffosocko.com edited /Indieweb_for_Education (+82) "+Dan Scott" (view diff)
#
KartikPrabhu wait! it looks at a BIN file to find atom feed?
[kevinmarks] joined the channel
#
[kevinmarks] Does the pubsubhubbub hub have a way of telling the subscribers that the origin url has changed?
#
[kevinmarks] Markdown << https://css-tricks.com/little-stuff-markdown-always-forget-google/
#
loqi.me edited /markdown (+69) "[kevinmarks] added "https://css-tricks.com/little-stuff-markdown-always-forget-google/" to "See Also"" (view diff)
#
Loqi ok, I added "https://css-tricks.com/little-stuff-markdown-always-forget-google/" to the "See Also" section of /markdown
#
Loqi [Chris Coyier] Little Stuff about Markdown I Always Forget and have to Google
[chrisaldrich] joined the channel
#
tantek GWG: I just heard from Jeremy Zilar, he is looking at getting event space in NYC for indiewebcamp
Alexandar joined the channel
#
aaronpk KartikPrabhu: lol yeah that's webfinger
#
aaronpk well, not bin, but a weird xml format
#
KartikPrabhu aaronpk: oh ok FF told me it was a BIN file when I clicked the link
#
KartikPrabhu probably because it has no extension/mime-type?
#
aaronpk oh i didn't set the right content-type header on mine
#
KartikPrabhu yeah that is probably the cause of that
#
aaronpk great example of why host-meta is not a good solution
#
KartikPrabhu yup
#
aaronpk ok content-type header is set right now
#
KartikPrabhu yes, it does say "xrd-xml" now
#
tantek lol xrd run away
#
KartikPrabhu yeah didn't open it because why!?
#
tantek maybe feed it to kevinmarks unmung to see what it does
#
tantek maybe it will jsonify it ;)
#
KartikPrabhu in my world XRD = X-Ray DIffraction
#
KartikPrabhu unmung did that http://www.unmung.com/mf2?url=http%3A%2F%2Faaron.pk%2F.well-known%2Fhost-meta&html=&pretty=on
#
tantek hah
#
tantek so it has a rel to itself. ok
#
tantek.com edited /Planning (+80) "/* Planning */ Jeremy Zilar available to help co-organize IWC DC" (view diff)
#
aaronpk so the one thing I've learned from mastodon which may be useful to us is how the "remote follow" works
#
tantek huh
#
tantek probably worth writing it up
#
tantek though I detest that the UI is bleeding "remote" into the text
#
tantek why isn't it just "follow"? that is, as a user I should never see a "Remote Follow" button (which is what you see when you view someone's profile)
#
aaronpk ugh yeah
#
aaronpk the remote/local distinction is a weird artifact
#
aaronpk but, here's the cool part
#
aaronpk you're on someone's blog, and you see a "follow" button
#
aaronpk click, takes you to a new page on their site, which has a form input to "enter your URL"
#
aaronpk once you enter your URL, their server does a GET request to your site to find your "following" endpoint. either as an HTTP Link header or an html link rel.
#
aaronpk then it redirects you to your following endpoint with a parameter in the query string, url=http://url.to.the.blog.you.were.on
#
aaronpk now you're on your own interface, with your own authentication context, so you can log in however, or are probably cookied, and you can confirm the follow action, all which is under the control of your own software
#
tantek.com created /Template:jeremyzilar (+570) "create since I used it" (view diff)
#
tantek.com edited /Template:jeremyzilar (+10) "one more copy/pasta" (view diff)
#
tantek aaronpk, yes that flow makes sense. just s/remote follow/follow
#
tantek so does that require a new protocol and new discovery?
#
tantek IIRC OStatus didn't have anything for that
#
tantek anything *standard
#
tantek hmm, sounds like a web action with specific rel discovery underneath
#
tantek that is, you can replace all the above about "following" someone with "liking" something and it could work the same
#
aaronpk OStatus does!
#
aaronpk I can't find the docs on it ironically
#
aaronpk the old URL for it is http://ostatus.org/schema/1.0/subscribe
#
tantek sigh
#
aaronpk and for some reason people don't like to have web pages up at the URLs they use for these link rels
#
tantek like the actual rel value is a URL?
#
aaronpk yeah
#
gRegorLove aaronpk: Now that venue's confirmed, can we get 2017.indieweb.org/Bellingham set up, including indieweb RSVP and tito links? Let me know if we can push updates to that through github or something.
#
aaronpk from mastodon: <Link rel="http://schemas.google.com/g/2010#updates-from" type="application/atom+xml" href="https://mastodon.social/users/aaronpk.atom"/>
#
tantek and this is why using URLs for rel values is basically no better than a string, so you might as well just use a string and stick it in a registry :P
#
aaronpk <Link rel="http://ostatus.org/schema/1.0/subscribe" template="https://mastodon.social/authorize_follow?acct={uri}"/>
#
tantek I think this might be fundamental reason why RDF (which assumes URLs for all names of properties etc.) is a lot more fragile than advocates want to admit
#
gRegorLove Free tito registration, not the $5
#
tantek oh wait gRegorLove is actually having a productive conversation, my bad :)