#indiewebcamp 2014-07-11

2014-07-11 UTC
paulcp joined the channel
#
Loqi [mention] https://brid-gy.appspot.com/like/twitter/t/487370638374363136/541 linked to http://indiewebcamp.com/2014#East_Photos (webmention)
#
Loqi [mention] https://brid-gy.appspot.com/like/twitter/t/487370638374363136/2149850018 linked to http://indiewebcamp.com/2014#East_Photos (webmention)
#
Loqi [mention] https://brid-gy.appspot.com/like/twitter/t/487370638374363136/18299065 linked to http://indiewebcamp.com/2014#East_Photos (webmention)
#
Loqi [mention] https://brid-gy.appspot.com/like/twitter/t/487370638374363136/19090770 linked to http://indiewebcamp.com/2014#East_Photos (webmention)
#
@mathpunk @frkbmb @skry: I know squat about security but @pjf's "Exobrain" project seems pretty #indieweb and is tying together services with 0mq. (twitter.com/_/status/487392973475549184)
snarfed and emmak joined the channel
#
mko POLL: For your IndieWeb homepage, do you prefer A) curated list of content (e.g. http://briancray.com/ ), B) entire social stream (e.g. http://aaronparecki.com/ ), C) personal statement / bio (e.g. http://edward.oconnor.cx/ ), or D) something more visual (e.g. http://samanthatoy.com/ -- just pretend it's indieweb enabled, I know it's not)?
#
KartikPrabhu mko: Currently I have a small welcome statemwnt and the latest 5 articles. But planning to change that, so the homepage is more of a visual "what's new" page
#
KartikPrabhu for ref: https://kartikprabhu.com/
#
mko Yep. I've got your site open in a tab already. lol
#
KartikPrabhu :)
#
mko You've got a nicely designed site. Not overly designed but still thoughtful about content presentation.
#
KartikPrabhu mko: thanks :)
#
KartikPrabhu !tell JonathanNeal: figured out the fragmention.js trouble from yesterday. It seems Closure Compiler does something funny so that the code looks only at the first word and ignores everything after the first + sign in fragmention. unminified js now on site, will play with uglify and see how that goes.
#
Loqi Ok, I'll tell them that when I see them next
#
KartikPrabhu mko: I like to have lots of whitespace and try to be mindful of loading huge images...
#
KartikPrabhu mko: also in case you haven't seen: http://indiewebcamp.com/homepage though I don't fully agree with everything there
#
rascul KartikPrabhu what don't you agree with on /homepage ?
mdik joined the channel
#
KartikPrabhu basically having your homepage as you. I like keeping that on my about page... and would like to have latest updates on homepage since I believe that is what most visitors would want (no actual data on this yet)
#
rascul ahh ok
#
KartikPrabhu so not really a disagreement... some diff. priorities i suppose
snarfed joined the channel
#
rascul well it does describe the stream for most recent posts and other activities for the homepage also
#
rascul most of the about stuff that's listed i include in my h-card on my homepage
#
mko Yeah. Basically, I agree with everything on /homepage
#
mko I just don't know how far to go with the stream vs the personal statement and simpler content.
#
KartikPrabhu rascul: yes... I want my full h-card on about. Basically I think of homepage as "latest stuff I posted" and about page as "this is me"
#
mko My full h-card is embedded on every page.
#
mko It's part of the footer.
#
KartikPrabhu mko: yeah that I up to your own design sensibilities. I feel that having design diversity is good :)
#
rascul i also have /about (although it's mostly empty right now) where i would potentially put more useful information about me so i don't clutter up / with it
#
mko Yeah. I do, too. Just trying to get a feel for what people might want as a "template"
#
KartikPrabhu !tell JonathanNeal: uglify seems to have worked correctly!
#
Loqi Ok, I'll tell them that when I see them next
#
rascul i agree that design diversity is good
#
KartikPrabhu mko: making a general template is kind of hard for that reason
#
rascul otherwise it would evolve into what has happened with facebook and twitter looking much the same
#
mko Basically, those are the 4 templates that I've found useful, and wanted to see if there was really any other kinds that people would be interested in.
#
KartikPrabhu rascul: yup! exactly
#
KartikPrabhu mko: feel free to document those on the /homepage on the wiki... such research is very much useful
#
rascul btw i'm documenting my ssl setup per bear's request after doing a little more cipher tweaking
#
bear rascul++
#
rascul might be useful for others who use nginx
#
Loqi rascul has 11 karma
#
rascul i'm up to 11! where did they all come from?!
#
mko Unicorns.
#
KartikPrabhu rascul: one must accept free internet points no matter from whence they come :)
#
bear net split in progress FYI
irdan, saurik_, donpdonp, arlen, iboxifoo, hadleybeeman, bigbluehat, CaptainCalliope, ngoldman, piney0, amblin, hugoroyd, tallpaul_, mattl, benward, voxpelli, jden and rknLA joined the channel
#
bear net split recovered
#
@cascadesf @t @chloeweil @indiewebcamp I'm sorry to hear this :( (twitter.com/_/status/487409557874802688)
#
@andigalpern @t @indiewebcamp I'm sorry to hear this sad news and hope you are doing okay. :( (twitter.com/_/status/487409977053573120)
#
Loqi [mention] https://brid-gy.appspot.com/comment/twitter/t/487370638374363136/487409977053573120 linked to http://indiewebcamp.com/2014#East_Photos (webmention)
KartikPrabhu joined the channel
#
KartikPrabhu that was a major net split
#
bear yea, about half went away from my view
#
KartikPrabhu I had 4 people! :P
#
bear wow - so it was more than one server restarted then
#
rascul.io created /User:Rascul.io/ssl (+3475) "describe my ssl setup" (view diff)
#
rascul i think that mostly covers it
#
bear nicely documented
#
rascul thanks
#
bear heck, IMO you should move those items over to /nginx
#
bear the comments from both would make it a killer page
#
rascul.io edited /User:Rascul.io/ssl (+145) "/* Basics */ add link to my ssllabs" (view diff)
#
rascul that's probably a good idea
#
rascul after the tweaking i did last night, i have full client compatability according to ssllabs
#
bear when I created the notes in /nginx I made them body items
#
bear having them as part of the config means the info copies with the config
#
bear my suggestion would be to keep the topic headings tho on /nginx
#
rascul the hsts stuff people might not necessarily want to include though, maybe i should note that
#
bear I think it goes as another section like HTTP and HTTPS is
#
bear "static site with HSTS"
#
bear so they can compare
#
bear that's why I put the SSL config items in the http block - to keep the "with SSL" block as small as possible
#
rascul i think i've got the host specific ssl stuff separated from the general stuff, and all the general stuff in http block
#
bear nods
#
rascul hrm why a rewrite for the static site with ssl termination?
#
bear just being pendantic about the possible changes
#
rascul and not a redirect?
#
bear when I wrote it I had forgotten about redirect
#
bear so +1 to changing it
#
bear (I was still deep in 1.4 land)
#
rascul ahh ok
#
bear now with 1.6 being available - let's definitely use all of the newer syntax
#
rascul i will add to /nginx tomorrow, i think i'm getting too sleepy for more wiki editing tonight
#
bear no worries - if I get out of this python stuff I may start it - but i'm also dog tired
#
rascul oh except for one minor edit
#
bear i'm just silly happy to have your config as a reference for my day job to be honest!
#
bear tweaking cert configs is a pain in the tuckus - so I'm always glad to find good sources
#
rascul.io edited /https (+27) "/* Level 5 security */ link to my ssl configuration" (view diff)
#
bear nice!
#
rascul if you get to /nginx before i do i'm happy with that, i probably won't get to it until tomorrow evening
#
rascul i did find many sources for nginx/ssl stuff, but much of them were outdated, and much of them were incomplete, i ended up going through a lot of the documentation anyway
#
rascul i think i came across one that you had written up bear
#
bear same here when I started that
#
bear yea, that blog post is going to get an update with this new info
#
rascul i think i didn't like your cipher set, probably it was older
#
bear yes, the one in the blog was almost 2 yrs old and pre-heartbleed IIRC
#
rascul i've got a set of strong ciphers that cover most clients then a few specific ones for the rest
#
bear nods
#
bear I like how you carved out java6 also
#
bear a *lot* of java6 out in the wild
#
rascul only java6 needs the ssl3, and i think that's because my dhparam is 4096 and i think java6 only supports up to 1024
#
KartikPrabhu Update: Marginalia is now active on all posts that accept responses! Fun… :D
#
bear KartikPrabhu++
#
Loqi KartikPrabhu has 43 karma
#
rascul or at least that's what ssllabs led me to believe
tabraldes, XgF, ellton, PMurphs, _axx, onewheelskyward, lmjabreu_, jancborchardt, wagle, dysfun, jonnybarnes, JonathanNeal, nemo-yiannis, fofr, SamB, ben_thatmustbeme, bret, edrex, realz, ozatomic, b0bg0d____, wolftune, danfowler, emmak, jacus, yakker, teknotus_, JasonO, protman, peat, catsup, rektide, lukebrooker, Nagahz and hober joined the channel
#
bear and the rest of the netsplit is back
#
Loqi JonathanNeal: KartikPrabhu left you a message 47 minutes ago: figured out the fragmention.js trouble from yesterday. It seems Closure Compiler does something funny so that the code looks only at the first word and ignores everything after the first + sign in fragmention. unminified js now on site, will play with uglify and see how that goes.
#
Loqi hober: KevinMarks left you a message 2 days, 1 hour ago: one name for the [Twitter for iOS] thing that twitter calls 'source' would be Activity Streams (and Atom's) 'generator' - so <span class="p-as-generator">Twitter for iOS</span>
#
Loqi JonathanNeal: KartikPrabhu left you a message 37 minutes ago: uglify seems to have worked correctly!
#
rascul bear are you aware that uwsgi now can run php natively?
#
rascul ie with embedded php and not cgi
#
rascul http://uwsgi-docs.readthedocs.org/en/latest/PHP.html
#
bear I knew that uwsgi was magical and had many powers - php I didn't know about tho
caseorganic joined the channel
#
rascul i'm curious to see benchmarks vs php-fpm to see how it compares but i'm not really interested in doing it myself heh
#
bear haha
#
bear yea
galfert, finchd, mlinksva, jtzl_ and rhiaro joined the channel
#
rascul it's nice though in the sense that you can just use uwsgi for everything
shepazu joined the channel
#
bear I find haproxy and varnish easier to configure than uwsgi
#
bear I may have to go find a book or a good tutorial on it
#
rascul uwsgi can be a beast to configure for big and complex setups
bitraten, hmans, Guerillero, aranasaurus, halorgium, memecake, kronda, Phae, walkah, aaronpk, tommorris, smcgregor, binbasti, mdik, muhh, dariusdunlap and the_merlin joined the channel
#
Loqi binbasti: tantek left you a message 1 week, 1 day ago: FYI I got the Moz App Manifest docs to link to the W3C Manifest for Web Apps spec. Making some progress.
#
rascul i've not worked with haproxy or varnish
#
Loqi binbasti: bnvk left you a message on 7/3 at 2:32am: cool, let's touch base next week then, my traveling & conf speaking whirlwind is just starting :P
#
bear more netsplit folks returning - up to 95 users - still about 20 or so missing
#
aaronpk Net split?
#
rascul irc servers become disconnected from each other
#
bear yea, huge multi server net split
#
rascul back before chanserv, it was a method of gaining control of channels heh
chrissaad, sdboyer, kylewm, Garbee, etymancer and Jeena joined the channel
#
bear yea, you would wait for one to happen and then join as many of them as you could directly until you got ops
snarfed joined the channel
#
KartikPrabhu bear: would be nice to have a python implementation of a an authorization endpoint for IndieAuth, so you can use your own website to log in instead of Github... just an idea ;)
#
bear KartikPrabhu - that was what I was working on last night
#
KartikPrabhu sweet! :D I'll definitely use it... once I have some log in system on my site :P
#
bear My flask helper app (for webmentions) now has a /login /auth and /success endpoint
#
bear all I need to do is make the auth code storage work
#
bear so right now you login and then it says "token valid" - but you can't do anything else :)
#
KartikPrabhu so you want to add token storing so you can sustain the log in state...?
#
bear yea, or learn how cookies work
#
bear so that the caller gets a token stored
#
kartikprabhu.com edited /fragmention (+139) "/* IndieWeb Examples */ added me :)" (view diff)
#
bear i'm working on an update to python-twitter right now - later tonight I hope to swing back to this and i'll push it to a repo so you can give feedback
#
KartikPrabhu oo python-twitter exists? hmm never played with that
#
bear it's a twitter wrapper in python
#
bear sighs
#
bear like I needed to explain it - the name does that already
#
KartikPrabhu wonders how diff. it is from the activity-streams unofficial stuff that bridgy uses
#
bear I have no idea
#
bear streaming support was just added
#
KartikPrabhu I see... will look at it for my Lift code... which is supposed to connect me to all this social-media hullabaloo...
#
bear the latest v2.0 release is just after a big refactoring of the library - so I welcome any bugs or feedback if you give it a try
#
KartikPrabhu sure thing... playing with Twitter API makes me tired. will look into it soon though
#
bear yea, it's a mess
#
bear documentation sucks sadly for python-twitter - not enough hours in the day
gRegor` joined the channel
#
KartikPrabhu not-enough-hours--
#
Loqi not-enough-hours has -1 karma
#
bear :)
scor, brianloveswords, ericholscher and wolftune joined the channel
#
Loqi [mention] https://brid-gy.appspot.com/like/twitter/t/487370638374363136/119575945 linked to http://indiewebcamp.com/2014#East_Photos (webmention)
indie-visitor joined the channel
#
Loqi Welcome, indie-visitor! Set your nickname by typing /nick yourname
fr0zen joined the channel
#
KartikPrabhu welcome fr0zen
#
fr0zen hi! :)
#
bear KartikPrabhu - https://github.com/bear/dainin/blob/master/indieweb_listener.py ignore the other python code, this is kind of my testing ground :)
#
KartikPrabhu bear: cool
#
bear it uses the new indieauth.py stuff I added to ronkyuu
#
bear wanders off to find food
#
KartikPrabhu fr0zen: how goes the website stuff?
#
fr0zen KartikPrabhu: I just found indiewebcamp. Digging to understand better. I was starting to move out of all these dangerous social networks and got very happy to find some more ppl thinking the same :)
#
fr0zen *sorry my not so good english, I'm Brazilian
#
KartikPrabhu fr0zen: not to worry... nice to to see indiewebcamp efforts getting some attention outside the US :)
#
KartikPrabhu fr0zen: do you already have a domain name or are you looking to start there?
#
bear heck, I'm supposed to be a native English speaker and I even get words wron
#
bear +g
#
fr0zen LOL :P
#
bear and yes, welcome to the IndieWeb!
#
fr0zen KartikPrabhu: I had in the past, but stopped use and the domain dropped. Now I'm looking to register a new one.
#
fr0zen \o/
#
KartikPrabhu neat! that would be the first step then :)
#
fr0zen Yeah! What you think about github pages? I was thinking about use it as my personal website (and domain).
#
fr0zen with jekyll
#
fr0zen (cheap $$ :))
#
KartikPrabhu fr0zen: isn't that just switching to another silo though? I would get my own domain name but use Github for hosting (if that is possible) That way if something goes wrong with github you still have your domain name and can move to somewhere else
#
KartikPrabhu more info on Github Pages: http://indiewebcamp.com/hosting#GitHub_Pages
#
fr0zen KartikPrabhu: awesome! So many things has already been thought. Let me keep reading though :)
#
KartikPrabhu fr0zen: yes! use the search on the wiki. People have been documenting a lot of stuff
#
fr0zen KartikPrabhu: I'll. Thanks!
#
fr0zen and then I can help with a pt_BR translation
#
KartikPrabhu also feel free to ask here. people will always reply even if later on
#
fr0zen :)
snarfed joined the channel
#
KartikPrabhu about.me seems to be getting quite popular...
addal joined the channel
#
jden fr0zen: i home many of my sites on gh pages, all on domains i own
#
jden eg my homepage, http://jden.us
#
jden major downside is that they don't support custom ssl hosting
#
fr0zen jden: nice! Following :)
#
KartikPrabhu jden: you should add your experience here: http://indiewebcamp.com/GitHub#Limitations
#
jden.us edited /GitHub (+180) "/* Limitations */ Add No SSL section" (view diff)
#
KevinMarks !tell hmans rel-me can be resolved in chains of mututal rel-me, so you link to a hub from the silos that allow only 1 link, then mutual-link the other sites together, but that can make the parsing trickier for consumers
#
Loqi Ok, I'll tell them that when I see them next
#
@benwerd Excited by the proposed #yxyy sessions. Creative writing! The future of storytelling! #indieweb! Hiking! I'm down for everything. (twitter.com/_/status/487474604370169856)
#
peat ... if anyone wants help with static hosting with Jekyll and Amazon's S3, I'm happy to help. It's cheap and durable (eg: easy to move to a different host). Still haven't figured out the plumbing on the more interactive indieweb tech (like webmentions), though.
lukebrooker, phil80, friedcell, eschnou, cweiske, chrissaad, jsilvestre, pauloppenheim, KevinMarks, glennjones, Sebastien-L, petermolnar, pfefferle and squeakytoy joined the channel
#
rascul peat this may be a start https://github.com/bcomnes/webmention.js
friedcell, dns53, adactio, lmjabreu_, jonnybarnes, pfefferle, phil80, scor, Sebastien-L, pbeaulieu and h1ro joined the channel
#
Loqi [mention] https://brid-gy.appspot.com/like/twitter/t/487370638374363136/5774462 linked to http://indiewebcamp.com/2014#East_Photos (webmention)
Victorium, EchoLynx, dns53, snarfed, brianloveswords and gRegor` joined the channel
#
@Wordius IndieWeb Twitter implementation http://wordius.com/?p=1104 (twitter.com/_/status/487605228363972608)
#
@Wordius A long way to go, but first #IndieWeb Twitter updates page is running on my site... http://wordius.com/?p=1104 (twitter.com/_/status/487605517783543810)
#
jonnybarnes anyone here use PHP and synidicate to twitter? do you use tmhOauth library?
#
jonnybarnes I can't upload photos when using thujohn's wrapper for Laravel
#
jonnybarnes https://github.com/thujohn/twitter-l4/issues/34
#
petermolnar yes, I do, w/ WordPress & Nextscripts
#
petermolnar but I believe nextscripts had altered it
pfefferle_ and mko joined the channel
#
JonathanNeal Good morning (PST)
#
gRegor` Ignoring DST out there? :)
#
hmans hi from CEST
#
Loqi hmans: KevinMarks left you a message 11 hours, 10 minutes ago: rel-me can be resolved in chains of mututal rel-me, so you link to a hub from the silos that allow only 1 link, then mutual-link the other sites together, but that can make the parsing trickier for consumers
#
hmans \o/
#
hmans I'm implementing Webmentions in #pants tonight.
#
gRegor` Awesome!
#
hmans That is, unless my ISP kills my line again. :/\/\/\/\/
#
hmans If you ever end up living in Germany, don't use Kabel Deutschland.
#
gRegor` Let us know when you want to test it out. We can send some wm.
#
hmans It works fine 95% of the time, but in those other 5% you're pretty much f'd.
eschnou joined the channel
#
hmans What's the official name? "Webmentions"? "Webmention"? "webmention"?
#
mko hmans: "webmention" is singular, while "webmentions" is plural. It's not a proper noun.
#
mko Note: This is just based off the documentation that I've been using at http://webmention.io/ to implement them myself.
#
hmans Just wondering... am I implementing "webmention"? "webmentions"? "the webmention protocl"? Any of these? None?
#
hmans Just thinking in terms of branding right now, it doesn't really matter much.
#
mko Oh. Uh, aaronpk's docs call it the "WebMention protocol" for the actual protocol.
#
hmans Also, what's the definitive source of the spec? https://github.com/converspace/webmention/blob/master/README.md?
#
gRegor` I think as a protocol, you could refer to it as "Webmention", capitalized. But I don't think the capitalization is too immportant.
#
mko Though the only place that I've seen it camelcased is on the README.
#
gRegor` We've moved away from the camel cased version
#
gRegor` "webmentions" plural is usually referring to the actual mentions that are sent / received.
#
gRegor` E.g. "I'm displaying received webmentions on my articles"
#
mko Yeah, hmans -- that's the current definitive spec.
#
mko And, agreed gRegor`
#
gRegor` mko: Where do you see camel-cased?
#
gRegor` On aaron's site
#
mko aaronpk's GitHub repo for webmention.io: https://github.com/aaronpk/webmention.io
#
mko Specifically throughout the README
#
gRegor` Ah, yeah that should be updated
#
gRegor` hmans: Yeah, shortcut to that spec is http://webmention.org. The wiki will have a lot more details of people's implementations, too. http://indiewebcamp.com/webmention
#
hmans \o/ thanks.
#
gRegor` Then you can add yourself to the implementations section of that page :)
#
hmans First things first... :)~
chrissaad, paulcp, j12t, wolftune and snarfed joined the channel
#
bret !tell peat see https://webmention.herokuapp.com/ instead of webmention.js. i did the later and it needs a fair ammount of work
#
Loqi Ok, I'll tell them that when I see them next
emmak and gRegor` joined the channel
#
hmans So, let's say I had a blog that serves posts as HTML and also as JSON. What would be the semantically correct way to link to the JSON version from the headers of the HTML version? link rel=alternate type=application/json?
#
snarfed sounds reasonable
#
bear +1
#
hmans \o/ thanks
#
snarfed if your html has mf2 (or mf1), users can also generate their own json with mf parsers, e.g. https://kylewm.com/api/mf2?url=https://snarfed.org/2014-06-09_tel-aviv-surfing-club
#
snarfed …but you already knew that
#
hmans I didn't.
#
snarfed details: http://microformats.org/wiki/microformats2#simple_microformats_2_examples
#
gRegor` DRY++
#
Loqi DRY has 1 karma
#
gRegor` Well that's surprising
#
gregorlove.com created /dry (+17) "redirect" (view diff)
wolftune and phil80 joined the channel
#
gregorlove.com edited /DRY (+22) "slight rephrasing" (view diff)
#
gregorlove.com edited /DRY (+1) "period" (view diff)
#
bear DRY++
#
Loqi DRY has 2 karma
#
bear gRegor`++
#
mko DRY++
#
Loqi DRY has 3 karma
#
bear :)
#
gRegor` Guys, don't repeat the karma. ;)
#
mko DRK--
#
Loqi dude
#
gRegor` Haha
#
bear I was just backfilling as I'm sure I would have given it at least *one* before
#
mko I would only ever give DRY one karma. Ever.
#
mko ;)
#
gRegor` mko: I like your "MO" logo/icon on your site.
snarfed joined the channel
#
gRegor` Nice show/hide detailed h-card, too.
#
mko Thanks. I went through about 80 different variations before I settled on that one.
#
mko The website actually had a ton more depth to it, but I removed all of the partially implemented stuff when I was starting to fundraise for my last startup. I decided to clean slate it this time around and change to Node (that one is in PHP right now), though I might keep the MO logo as part of the footer or something.
#
gRegor` My site is pretty minimalist and I having found a way to display h-cards in a way I like yet.
#
mko Yeah. h-cards are a cumbersome way to display a person from a design standpoint.
dariusdunlap joined the channel
#
mko Unless it's in a friend list or address book, h-cards just aren't a format we typically use. They're too information rich when you want to be wholly descriptive, and much of the information is pseudo-extraneous.
peat_ joined the channel
#
Loqi peat_: bret left you a message 46 minutes ago: see https://webmention.herokuapp.com/ instead of webmention.js. i did the later and it needs a fair ammount of work
jacus_, emmak_, protman_, teknotus, a, eschnou, rektide, JasonO, npdoty and KevinMarks joined the channel
#
KevinMarks mko: why are h-cards too rich? the minimal h-card is <a href="http://kevinmarks.com" class="h-card">Kevin Marks</a>
#
mko I was referring to practical usage of a "complete" h-card.
#
KevinMarks image+name+link is a useful subset
#
KevinMarks you don't need all the properties, we're not being DBAs here
#
KevinMarks no Schenectady problem
#
mko KevinMarks: h-cards don't have to be too rich. To provide a complete picture of an identity with one, i.e. a complete h-card that includes every property applicable to a person, is problematic from a design standpoint.
#
KevinMarks well, that ends up being a profile page
#
mko Right, except we don't have a good mechanism for forwarding h-card parsers from a minimal h-card to a canonical h-card. I've been thinking about writing up a proposal once I finish my updated site.
#
KevinMarks u-url u-uid?
#
mko That's a reasonable convention, but there's no parser out there that would interpret that right now as "if you want the full h-card, parse this url instead"
#
mko I was thinking more along the lines of how link rel="canonical" is used.
#
snarfed there's rel-author links. http://indiewebcamp.com/authorship does say to prefer the inline h-card to that, but it could be revised
#
KevinMarks http://microformats.org/wiki/representative-hcard-parsing is a pass at this, it may need updating
#
mko KevinMarks++
#
Loqi KevinMarks has 41 karma
#
KevinMarks not quite the same thing, but coudl extend the principle
#
mko KevinMarks: I had no idea that existed, and it didn't turn up in my first search into the subject.
#
snarfed right, that's more appropriate
#
KevinMarks it needs better linking then
#
KevinMarks longer discussion here: http://microformats.org/wiki/representative-hcard-brainstorming
tantek joined the channel
#
mko Yeah. That definitely is more in line with what I was thinking, though I was thinking like a u-canonical class in addition to the u-uid or u-url properties as I've always seen those two as having less specific purposes.
#
Loqi tantek: aaronpk left you a message on 7/10 at 12:16pm: I want to run some sort of indieweb event during http://pdx.devweek.org and am considering co-opting Homebrew Website Club to be more of a hack night or something. any thoughts on how that might work in conjunction with SF? or just not do joint that day?
#
KevinMarks also this is all microformats 1 syntax
#
mko Yeah. I noticed.
#
KevinMarks uid is potentially key - if used consistently across sites, combined with rel=me
#
tantek mko - I don't understand your points about h-card and presentation. h-card is added *after* you figure out the presentation you want. h-cards are presentation neutral.
#
mko tantek: Just because they're presentation neutral doesn't mean people don't consider their content structure heavily when determining what they are going to display in a meaningful manner.
dariusdunlap joined the channel
#
KevinMarks right, but that is DB thinking. https://plus.google.com/+KevinMarks/about is a big hCard, but it isn't the root
#
mko I'm not thinking about it from a "DB thinking" perspective. I'm talking about from a usability and experience standpoint. In order to have consistent presentation of one's identity (if that's what he or she wants), one needs the ability to have a consistently parsed identity, regardless of the parsing source.
#
mko In order to provide that currently, the only real option is to provide a "complete" h-card with all of the properties one would want in his or her identity's presentation everywhere that his or her h-card is embedded.
#
mko The concept of a canonical h-card, which seems to be what the representative-h-card could be if utilized properly, would fix that inconsistency in identity presentation.
#
mko Now, if an individual doesn't want consistency of identity presentation, they just omit the properties that indicate a canonical h-card, and parsers just use whatever is there.
#
mko tantek / KevinMarks: Does that clarify my interest a bit?
#
snarfed mko: you can always just hide some parts, so that ux/design is easier but parsing still gets everything. you already do that somewhat, right?
#
snarfed :trollface: (since invisible metadata is controversial around here)
#
mko Yeah. I already do that.
#
snarfed ducks
#
mko I always have the "invisible" metadata visible through some user interface convention, though.
#
mko It's not a criticism of h-card as a format. I love the flexibility and ability to provide varying identity presentation with as little or as much information as one would need or want for the purpose at hand. I just think it would be nice if the user had the option of better identity presentation to tools if they wanted it.
#
snarfed out of curiosity, is there a concrete problem you're trying to solve on your own site?
#
mko It's actually a problem that I'm thinking about with respect to IndieAuth.
#
mko I'm building a Passport authentication strategy for IndieAuth, and one of the concepts in Passport is that it builds a profile for the authenticating entity to use once the user has validated their identity. I was thinking about how we could leverage h-card parsing for building a consistent profile for the user.
#
mko A lot of people don't want all of their profile metadata on their homepage, but they don't want to link to their personal homepage's profile page on their social sites and third-party authenticating bodies.
#
mko "a lot of people" anecdotally, obviously.
#
snarfed …and you still ideally want the passport profile to find all of that info
#
snarfed do these people want that too? or do they just want to log in?
#
mko Therefore, if they use a minimal h-card on their homepage and validate their identity through IndieAuth using their homepage, they will be stuck with an IndieProfile (my working name for it) only containing the minimal profile information, which is a bad experience for both the authenticating entity and the user.
#
mko Typically, yes.
#
mko I've worked most of my career in social networking and API-based app design. When a user connects with Facebook, many of the users expect that app to be able to find out anything about them that exists on Facebook, even if they haven't explicitly already told the app that they could have permission to access that information.
#
snarfed sure. and in this case, you're saying they'd expect their indieprofile to find their stuff across silos and their web site, even if they haven't linked them to each other?
#
mko Not necessarily.
caseorganic joined the channel
#
snarfed i'd actually believe it
#
snarfed just trying to clarify
#
mko I think users logging in with IndieAuth to an app would at least expect their full h-card to be present and available for use to the app developer.
#
mko And some of the less technical users would probably expect the information from whatever silo IndieAuth used to validate the RelMe identity to also be available.
#
snarfed sure. you can build that now. i guess i still don't see where possible inconsistency comes from
#
snarfed a few concrete problem examples would probably help. (and probably better for a blog post etc than just in irc)
#
mko Sure. I didn't plan on this conversation exploding. lol
#
snarfed :P np
#
gRegor` snarfed: you're not alone with invisible metadata. I have some, too. :)
#
gRegor` quacks
#
snarfed gRegor`++
#
snarfed may it rot in peace
#
mko The problem then comes with parsing in other places (i.e. if you parse a blog post for the identities of the individuals associated with it, those identities might not be the same as if you parse the same users' IndieAuth URLs or their personal site profile pages).
#
tantek !tell aaronpk sounds good to try a variant on 2014-07-30 - go for it and update the HWC event page for that day accordingly with your thoughts and let's make it happen!
#
Loqi Ok, I'll tell them that when I see them next
#
tantek snarfed++ for asking about concrete problem to solve
#
Loqi snarfed has 33 karma
bnvk joined the channel
#
bnvk tantek: yo, where in the world are you?
#
tantek mko re: exploding, I think you're overthinking the problem.
#
tantek bnvk - on a highway in the mojave desert
#
mko tantek: Maybe, but I managed to run into the problem pretty fast when I did a quick parsing test on a few different domains with h-cards on them. It all comes back to entity resolution and access to consistent information.
#
tantek mko - I see a lot of "people don't want" and "might not be" in your problem description which is less conducive to problem solving than say, "people currently do this on the web (insert real world URL)" and "people sometimes do this (insert another URL)"
#
bnvk tantek: ah, are you going that yxyy or xyxx or something?
#
tantek mko - so the problem is the lack of documentation of such "problem pretty fast" with citations of real world URLs
#
tantek no-one else can reproduce your analysis / reasoning
#
mko tantek: Fair enough. Once I finish with what's first on my plate, I'll write up the full scope of what spurred the interest in the concept.
#
tantek don't need "full scope"!
#
tantek that's the point
#
tantek would be happy to start with just *one* simpler user-flow that you're trying to solve
#
tantek s/simpler/simple
#
Loqi tantek meant to say: would be happy to start with just *one* simple user-flow that you're trying to solve
#
tantek BTW I will also point out that the reasoning against the negative approach (use of reasoning by "people don't want" and "might not be") ends up creating very overly complex protocols and formats
#
bnvk snarfed: you around? You up for getting drinks in SoMa / Mission area this eve?
#
tantek bnvk - en route to #yxyy
#
snarfed bnvk: sure! i have plans later but i could do something around 6pm or so?
#
bnvk tantek: nice, have fun / say hi to anyone I know ;)
#
tantek mko - to be fair - I admit to also seeing anecdotal data of different use of contact / other info about people on different representations of themselves.
#
bnvk snarfed: that could work, i'm at Facebook campus till 4 or 5 ish and not sure how long it will take me to get up to city
#
tantek and before I forget
#
tantek mko++ for "talking about from a usability and experience standpoint"
#
Loqi mko has 2 karma
#
bnvk snarfed: also going to a hackathon at the EFF tomorrow / could get coffee in the morning or something
#
tantek mko - back in the Technorati days ryanking and I did a lot of work on consolidating hCards of people across different sources into a superset of information - so it is definitely possible.
#
KevinMarks mko is part of this that the Passport model has the User ID cards idea, which look like business cards so have an implict expectation of address, phone number etc?
#
tantek we just need to look at today's real world examples, and documented user-flows, and figure out what details we need to make things work
#
tantek what is Passport?
#
snarfed bnvk: sure! that works too. or, my house is in glen park, on your way to soma/mission, you're welcome to stop by and say hi
caseorganic joined the channel
#
mko tantek: Passport is a tool for authentication and basic profile creation.
#
tantek bnvk - sorry to miss you! how long are you in SF?
#
KevinMarks right. name/photo/url is the common subset that is used everywhere
#
tantek mko - absent a larger use-case (i.e. a general social network like FB), I'm not sure a separate "tool for authentication and basic profile creation" makes sense
#
KevinMarks mko not this? http://support.microsoft.com/kb/277759
#
tantek mko - then again - would love to see what you mean by that on your own site
#
snarfed KevinMarks: heh that's what i thought too
#
tantek mko - sounds a bit like InfoCards (which was a bust last time I checked)
#
mko KevinMarks: And the idea of the "IndieProfile" was an extension beyond Passport itself. I just discovered the potential need for it while developing for Passport.
#
mko Passport = http://passportjs.org
#
tantek mko, URL for "Passport"?
#
mko It's just like OmniAuth or EveryAuth
#
KevinMarks aha
#
tantek aha now we're getting somewhere - I thought "Passport" was some abstract thing mko was building new!
#
KevinMarks is there an acronym for node that is like tmtowtdi?
#
tantek mko - could you start indiewebcamp.com/Passport with at least a stub pointer to that? and your one line summary above about potential applicability to the indieweb?
#
mko Anyway, sorry for the distraction. I wasn't trying to have this become a full-fledged roundtable. lol. I was really just trying to mention that it'd be nice if there was a way to point h-card parsers to a canonical h-card to use if we didn't want to include all of the markup in a specific place. You know, invisible metadata and all that jazz.
#
KevinMarks this is the mydomain.com vs mydomain.com/about split that KartikPrabhu has been talking about too
#
@Wordius ‘Dumb’ tweets on my WP site not working the way I’d like, because figuring out how to fetch ‘tweet_id’ so far beyond me. #IndieWeb (twitter.com/_/status/487670719635935232)
#
KevinMarks thing is, people refer to you by your domain externally
#
snarfed KevinMarks: re node acronym, atwood's law is close
#
mko Yeah. His website was one of the examples that I found this problem on.
#
mko All pages but his about page get a minimal h-card. His about page gets a really information-rich view of him.
JonPincus joined the channel
#
KevinMarks snarfed it's more like "there's a framework for that, oh and another 12"
#
JonPincus greetings all
#
snarfed KevinMarks: ah. then probably the xkcd on standards :P
#
JonPincus mko, i was just looking at the log and noticed you mentioning a passport indieauth module ... i had been thinknig about working on something similar
#
bnvk tantek: I'm here until monday- I should have pinged you a few days ago, but was jet lagged / in the belly of the beast at this conf!
#
tantek mko - informally we've used u-uid for that in the past
#
KevinMarks or the xkcd on from future import antigravity
#
KevinMarks npm install great_old-ones
#
kylewm !standards
#
Loqi http://xkcd.com/927/
#
KevinMarks node.js sometimes feels like the Laundry Files
#
JonPincus kind of like that
#
KevinMarks "I just typed npm install, why are tentacled monsters coming out of my computer from an alternative reality?"
#
snarfed mko: agreed on that problem. (i should have said that to start. :P) i do the same thing. i only want a picture of me on https://snarfed.org/about , not on the front page or articles, but i want the picture available to h-card consumers. i do that right now w/invisible metadata, which isn't ideal
#
Loqi gives snarfed the picture available to h
#
mowens.com created /Passport (+239) "Created page with "http://passportjs.org - An authentication library for Node.js Michael Owens is working on an IndieAuth authentication strategy for Passport to enable faster implementation of In..."" (view diff)
#
tantek snarfed what about the icon you want when people save snarfed.org to their home page?
#
tantek or any of your posts?
#
KevinMarks snarfed we were talking about using the favicon for that
#
KevinMarks or whatever the new name is
#
tantek KevinMarks: how about just icon? or shortcut icon? favicon is an IE anachronism at this point.
#
tantek what is an icon?
#
Loqi An icon in the context of the indieweb typically refers to a home page icon for an indieweb site http://indiewebcamp.com/icon
#
KevinMarks apple-touch-icon
#
snarfed tantek, KevinMarks: confused. that works now as a favicon, but h-card consumers (e.g. webmention recipients) don't use it. not sure what you were asking…
#
tantek snarfed - it *could* be part of your h-card
#
snarfed tantek: still confused
#
tantek that's what KartikPrabhu KevinMarks and I have been discussing in #microformats
#
snarfed i...agree?
#
snarfed ah ok
#
tantek snarfed, let me summarize in code
#
KevinMarks make it part of your hcard by moving the h-card to your <html>
#
tantek perhaps that's clearer :P
#
snarfed aha. sure…? i'm markup agnostic, i'll put it wherever
#
KevinMarks <link rel="shortcut icon" class="u-photo" href="km.jpg" />
#
tantek <html class="h-card"><title class="p-name">Ryan Barrett</title><link class="p-logo" rel="shortcut icon" href="avatar.jpg" /> ...
#
tantek s/p-logo/u-logo
#
Loqi tantek meant to say: <html class="h-card"><title class="p-name">Ryan Barrett</title><link class="u-logo" rel="shortcut icon" href="avatar.jpg" /> ...
#
tantek snarfed you're already putting it somewhere - that's the point
#
KevinMarks logo?
#
tantek you shouldn't need to put it somewhere else and violate DRY
#
KevinMarks photo surely
#
Loqi agreed.
#
mko I'm not against doing it that way either, as you know, DRY.
#
GWG I'm reviewing this conversation. I was thinking about this issue
#
snarfed tantek: got it. sgtm! honestly i'd rather the favicon not be my picture either, but not a big itch
#
tantek kevinmarks see http://microformats.org/wiki/h-card#Properties
#
KevinMarks tantek's point was that this is not invisible metadata if the photo is shown in the browser bar
#
tantek it's like a smaller version of u-photo
#
tantek and carries more of the semantic of something you want used as an icon
#
tantek snarfed - so the point is you *do* already put your logo on your homepage
#
tantek snarfed - you could distinguish too if you wanted, with u-logo being the "not your picture but your icon for your website" and u-photo for the more typical "photo of you"
#
snarfed tantek: yeah, i understand your point. i don't quite concede it UX wise, since favicon is pretty ghettoized and not on the home page…but it is visible, yes
#
GWG When did this conversation start?
#
KevinMarks it means company rather than person - it's under organisationa properties in vcard
#
tantek KevinMarks: I didn't even say anything about visibility in this thread
#
tantek GWG what is this?
#
KevinMarks but you could put u-logo u-photo
#
tantek KevinMarks: but no one uses it like that
#
GWG tantek: The whole h-card conversation.
#
mko GWG: around 10:34 PDT
#
snarfed lol. *i'll* say visibility. personally, my use case is, i *do* want my picture to be invisible metadata, ie available to code but not visible
eschnou joined the channel
#
GWG I'm reading down about the minimal h-card vs the complete h-card
#
tantek GWG - that conversation started in 2004-09 at FooCamp
#
mko lol
#
mko I think he meant this specific instance. ;)
#
mko Smartass.
#
GWG tantek: As mko said...the one today
#
GWG tantek: I can't see from a design standpoint having an extensive h-card displayed up front, but I could see having an extended version available, which appears to have been the discussion.
#
snarfed GWG, totally. mko, my invisible picture use case might actually back you up. if we revised http://microformats.org/wiki/representative-hcard-parsing to optionally prefer following a link to the complete h-card on another page, i could put a visible picture there and only there
#
tantek :)
#
mko snarfed++
#
GWG snarfed: Still catching up on the discussion
#
Loqi snarfed has 34 karma
#
tantek snarfed - your icon is not invisible. it is semi-visible.
#
tantek it shows up sometimes in browser UI, and it shows up when the user does add to home screen on their mobile device
#
snarfed tantek: yes, but again, ideally it wouldn't be visible. ideally i want the only "profile" picture on my site to be on https://snarfed.org/about
#
Loqi gives snarfed the only "profile" picture on my site to be on https
#
GWG snarfed: I keep wanting to build an h-card widget for Wordpress that will pull from the user profile.
#
gRegor` snarfed, mko: You can add me as an example of invisible photo, too. I have not set up an h-card on /about yet, but likely will. I'm already showing *a* photo there, but not the one I would like to use in my h-card. Easy to add another there, though.
#
mko Huzzah for support.
#
gRegor` also uses a "logo" for my site icon and would prefer it stay that way vs. my photo.
#
tantek snarfed - you don't want a /icon for when people bookmark your site? is that what you're saying?
#
tantek snarfed - no need to feature bloat "representative hCard"
#
tantek it's an algorithm meant to apply to a single page
#
tantek for information about that page
#
KevinMarks fair gregor - that was my logo/photo distinction - for tantek+me we use our faces as site "logos"
#
tantek your use-case is more like representative hCard for a *site*
#
snarfed tantek: sure, i don't care how/where we implement this. and no, i'm fine with the bookmark icon being my picture. i'd just rather not have my home page *or* favicon have my picture
#
KevinMarks also, using the same jpg for the hcard photo and the icon saves an http fetch on mobile
#
snarfed (just a personal design preference)
paulcp joined the channel
#
tantek snarfed - favicon is used for bookmark icon sometimes so not sure how to reconcile your different prefs there
#
mko I have to run, but I'll get around to organizing thoughts and examples hopefully this weekend or so.
#
snarfed mko++
#
snarfed tantek: sure, but that's an implementation detail
#
Loqi mko has 3 karma
#
gRegor` snarfed: Interesting, so one icon for the browser, but a separate apple-touch-icon (or equiv) when saving your site to a mobile screen?
#
tantek thank mko - appreciate it!
#
tantek snarfed - it is? I thought I was asking about user experience - not implementation.
#
@rknLA RT @t: shocked & numb from losing @chloeweil. saw her 11 days ago @IndieWebCamp http://indiewebcamp.com/2014#East_Photos http://indiewebcamp.com/images/9/90/indiewebcamp_nyc2014.JPG (ttk.me t4X12) (twitter.com/_/status/487676768006844416)
#
GWG tantek: What happened to chloeweil?
#
tantek GWG see #indiechat
#
rknLA :(
#
Loqi [mention] https://brid-gy.appspot.com/repost/twitter/t/487370638374363136/487676768006844416 linked to http://indiewebcamp.com/2014#East_Photos (webmention)
#
snarfed tantek, gRegor`: yeah, complicated. i'll try to rephrase. when people bookmark my site, i don't care if that icon is my picture. if users want it, that's fine with me. i just ideally don't want my favicon (ie the one you see when just opening the site in a browser) or home page or posts to have my picture.
#
tantek snarfed, the "or home page or posts to have my picture" I understand and I think is a reasonable user/author preference
#
tantek snarfed, to make the browser show something different than what is bookmarked - I'm not sure that is possible.
#
tantek I mean like in the browser address bar
#
snarfed tantek: sure. i'm thinking ideal use case.
#
tantek or rather the current spec for "shortcut icon" specifies it's for *both* bookmarking and browser UI / addressbar
#
tantek so I'm not sure how to separate those two to get what you want
#
snarfed yup. i was squarely in new-invention-land. sure. obviously there are lots of complicated implementation details. i don't care about the code and markup, i'd use whatever if it was possible.
#
tantek for now you could make it only available to iOS bookmarking/homepage by only using the iOS proprietary apple-touch-icon rel value
#
tantek but then android / firefoxos users would be left out
#
snarfed well, "left out" is debatable. they'd still have an icon that represents me. "favicon should be your face" is a subjective ideal, right?
#
snarfed definitely not universal
#
snarfed anyway. i'm only pushing hard so that i can describe it as a clear, concrete personal itch use case, since we like those :P
#
snarfed i'll find a wiki page to add it to
#
tantek.com edited /Passport (+129) "stub, dfn, see also" (view diff)
#
tantek snarfed, perhaps /icon#Brainstorming ?
#
snarfed sure
#
tantek snarfed - agreed about subjective, perhaps common practice, rather than "ideal" (even if some of us have the opinion of it being ideal)
#
tantek what is Passport?
#
Loqi Passport (capitalized) is an authentication library for Node.js http://indiewebcamp.com/Passport
#
tantek snarfed, you're not the only one who doesn't (want to) have their face as their icon, and it is useful to document the user/author/publisher preference accordingly, especially with multiple examples, so we can design for it.
#
snarfed on it
#
tantek but flipping that around to a positive, what *do* you want as your icon?
#
tantek would prefer to document positive alternatives rather than rejection of.
#
tantek aside: GWG - more here: http://microformats.org/wiki/hcard-history (if you're curious :) )
#
snarfed tantek: heh. i'm all for that flipping! but this might be the rare case where it's a dead end. it'd be any arbitrary non-representative picture, e.g. https://snarfed.org/favicon.png
#
bret lol what is that?
#
snarfed bret: not sure, i think it's something called a snarf from the transformers cartoon?
#
bret haha
#
bret cool
#
kylewm jesus, it's from Thundercats
#
tantek snarfed - great! cartoon icon example - I've seen others do that too
#
kylewm much disappoint
#
snarfed.org edited /antipatterns (+315) "/* invisible metadata */ use case" (view diff)
#
tantek snarfed - another similar example - combining cartoon and face: https://twitter.com/meyerweb/profile_image
#
tantek snarfed, I think there's a (slight but important) distinction between *invisible* metadata, and *linked* metadata to where it *is* visible.
#
tantek I think the latter is what you/we want for this use-case
#
tantek thus we should still unconditionally recommend avoiding invisible metadata
#
snarfed tantek: sure. the former what mko was pushing for
#
tantek and perhaps I should document that in the antipattern as an alternative
#
tantek he was? ok I must have missed that. I thought he just wanted *a* way to find the data (hence linking would work)
#
snarfed sure, yes
#
tantek can't believe he typed the phrase "linked metadata". paging tommorris.
#
snarfed i still have plenty of other invisible metadata that i'm fine with giving h-card consumers but hide everywhere UX wise, so i'm still pro, but i understand i'm the mnority
#
tantek snarfed - well I'd like to work with you on each those "other invisible" use-cases to see if we can find alternatives that suit you
bnvk joined the channel
#
snarfed tantek: understood. pretty low priority though :P
#
tantek snarfed, fair. :P
#
GWG tantek: So, how do you indicate there is more information elsewhere?
#
snarfed GWG: short answer, we don't know yet, but it might be worth figuring out an answer
KartikPrabhu joined the channel
#
snarfed.org edited /h-card (+750) "Issues" (view diff)
#
GWG snarfed: I'll be watching that one
#
tommorris Heh, tantek, there is still time for you to find that old time SemWeb religion.
#
snarfed.org edited /antipatterns (+32) "/* invisible metadata */" (view diff)
#
tantek GWG - short answer, typically with rel values are used for discovery of information on other pages related to the current page
#
snarfed GWG, gRegor`, mko: feel free to add to http://indiewebcamp.com/h-card#Issues and http://indiewebcamp.com/antipatterns#invisible_metadata
#
tommorris So, an example of invisiblish metadata: GPG public key.
#
KartikPrabhu mmm lots of Icon/logo talks
#
tommorris Nobody really cares about the bits but being able to retrieve it is useful
#
gRegor` tommorris: But that isn't really intended for human eyes, right?
#
tommorris Also maybe EXIF data although that tends to be quite interesting and useful
#
tantek tommorris: parsing HTML rel values at a page level is about as close to linked data I think I'm going to bother with. And LD is too new to be considered "old time". It's more like Latter Day. ;)
#
gRegor` Hey, KartikPrabhu. To be safe, I think I'm just going to update the venue to Panera for 7/16. I'd rather we have space and it be a bit quieter.
#
KartikPrabhu gregor`: want to add teh Panera location for next week
#
KartikPrabhu ha!
#
gRegor` Jinx
#
tantek tommorris: funny, I see snarfed's GPG public key quite visibly, by following the *visible* link from his about page
#
gRegor` :)
#
tommorris Like, it is quite useful to be able to see the aperture and focal length and exposure etc.
#
gRegor` I'll update and ping Denton and Mari to make sure they see it.
#
snarfed tantek: yup. i usually regret that though. i only added it recently, and i'm inclined to drop it again
#
tantek snarfed - would appreciate if you could help with the semi-visible vs. invisible distinction, especially when the use-cases you're documenting are all for *semi-visible* data, not invisible.
#
KartikPrabhu gregor`: neat!
petermolnar joined the channel
#
tantek help *spread* the semi-visible distinction
#
KartikPrabhu tantek: where was that 'official protocol' to conduct a HWC... now that Chicago might have more than 2
#
gRegor` It's on /Homebrew Website Club
#
tantek more than 2 what?
#
snarfed tantek: public key and geo region were actually from other people i'd talked to about this. i can say that if you want? or should i only post firsthand use cases?
#
tantek start with firsthand use cases
#
tantek because those have URLs :)
#
KartikPrabhu tantek: err people :P
#
tommorris tantek: the right religious analogy may be some combo of Kabbalah and the Millerites (now the Adventists).
#
tantek "other people I'd talked to" is more meh, because that can be hearsay / wishful thinking / people saying what they think they want but not actually knowing etc.
#
tommorris This is why I have a degree from a theological college. ;)
#
david.shanske.com edited /h-card (+73) "/* Issues */" (view diff)
#
tantek the whole steve jobsism of don't bother asking users what product they want - it's not their job to know etc.
#
KartikPrabhu tantek: nvm gregor` showed
#
GWG Okay. There is my small contribution
#
snarfed.org edited /h-card (-39) "/* Issues */" (view diff)
#
gregorlove.com edited /events/2014-07-16-homebrew-website-club (-50) "/* Where */ Chicago new location for this week." (view diff)
#
tantek there was some brainstorming about a rel=contact at some point to point from a home page to a more thorough "contact" or "about" page
#
gregorlove.com edited /events/2014-07-16-homebrew-website-club (-1) "/* Where */" (view diff)
#
GWG tantek: What are the downsides to that?
#
kartikprabhu.com edited /h-card (+35) "/* Issues */ added me" (view diff)
#
tantek GWG - downsides to what?
#
tantek GWG - you *can* link to a more complete h-card - just use a hyperlink to do so.
#
tantek people do it all the time with /about pages
#
tantek that is, with linking from their home or header to /about
#
GWG Having a link to an expanded h-card?
#
KartikPrabhu GWG: yes... I do that too... my expanded h-card is on about#me and it is linked to from everywhere there is a smmaller H-card
#
KartikPrabhu I think the issue it to do it programmatically or something
#
snarfed.org edited /h-card (+83) "/* Issues */ people" (view diff)
#
GWG Okay. So...if there is no downside...what is the upside?
#
KartikPrabhu for instance can we update the authorship/representative-h-card algorithm to optionally fetch the full h-card
#
GWG KartikPrabhu: But then you have to say where it is
#
KartikPrabhu yes... the problem is marking up the link to the full h-card. not linking to the h-catd
#
GWG rel="alternate" wouldn't work, right?
#
KartikPrabhu rel is scoped to the entire page
#
bear hmm, if the main page has the full contact info then the /about could (should?) link to it and add on descriptive addendum that is, by it's nature, out of scope to h-card
#
tommorris So, having data about the same thing spread across multiple pages seems like a limitation to the microform and approach
#
Loqi [mention] http://gregorlove.com/notes/2014/07/11/1/ linked to http://indiewebcamp.com/events/2014-07-16-homebrew-website-club (webmention)
#
bear I would say it's a violation of DRY
#
KartikPrabhu not necessarily... there can be one complete h-card. I'd really hate to have my full bio on very article and note
#
bear if we are working from the basis that your domain is your identity - then / should contain your h-card
#
tantek KartikPrabhu: re: " the issue it to do it programmatically or something" - not good enough, and the wrong approach. What's the use-case for which a programmatic solution is needed?
#
tommorris In RDFa, you can say things about the same resource in multiple documents
grantmacken joined the channel
#
bear and each article would have a link via the h-author to that h-card
#
tantek tommorris - you can say that - but no one actually implements any use-cases in that way
#
tantek classic example of data-mindset first rather than ux-first
#
KartikPrabhu tantek: if I am making an "address-book" or if I am replying to something and using the authorship algo, and my site uses more info than is on the smaller h-card on the post-page.
#
tantek KartikPrabhu: even "update the authorship/representative-h-card algorithm to optionally fetch the full h-card" is too abstract - the full h-card *why* ? that's still just data.
#
bear wants to continue but needs to get this fuzzy head of his sheared
#
bear looking forward to reading scrollback
#
tantek tommoris - you can aggregate information by uid
#
KartikPrabhu tantek: one case I ran into was parsing webmention from adactio... he had no photo. but maybe he doesn't put photo on every post but has it on hus about page
#
tantek tommorris - also rather than saying "seems like a limitation" better to ask how do I "(insert use-case)"
#
KartikPrabhu in fact if we had a way of doing this, I'd remove my avatar from every article and just link to the about page
#
mko KartikPrabhu++
#
Loqi KartikPrabhu has 44 karma
#
snarfed KartikPrabhu++
#
Loqi KartikPrabhu has 45 karma
#
KartikPrabhu people reading stuff don't need to see my avatar on every post
#
tantek KartikPrabhu: we do - rel=author
#
KartikPrabhu yes, but if I send a webmention to someone how do they find my avatar to display on their site?
#
KartikPrabhu the authorship does not cover it and neither does representative h-card
#
tantek they use existing /authorship which says to follow rel=author
#
KartikPrabhu wait it does..
#
tantek sigh.
#
KartikPrabhu tantek: sorry :P
#
KartikPrabhu errr it does! blerg
#
snarfed yeah. mko (etc) want the equivalent for home page h-cards
#
snarfed more specifically, they want that when the home page also has a partial h-card
#
KartikPrabhu aah then the equivalent situation is making "address book" or "people collection"
#
snarfed i guess you could just not mark up that partial info as an h-card, and include a rel=me link. doesn't seem ideal, but would work w/http://microformats.org/wiki/representative-hcard-parsing
#
KartikPrabhu I want to make a collection of people I follow, I put their domain name in my site UI, now the site has to go an fetch their info including possibly stuff not in their homepage
#
KartikPrabhu tantek: use case ^ ?
#
tantek KartikPrabhu: why are you making a collection of people you follow?
#
tantek what happens then?
#
KartikPrabhu tantek: reader stuff maybe (I get updates to their posts and all)... wasn't kylewm doing something like that
#
KartikPrabhu also the auto-linking in notes to people
#
KartikPrabhu insert their avatar as a little face
#
KartikPrabhu I forget what name was decided for this...
#
kylewm KartikPrabhu: http://indiewebcamp.com/Red_Wind#Address_Book
#
KartikPrabhu aah yes that^
#
KartikPrabhu kylewm: what does you "fetch profile" do?
#
KartikPrabhu your*
#
KartikPrabhu just homepage h-card?
#
kylewm yes
#
snarfed.org edited /h-card (+415) "straw man solution for partial plus complete h-cards" (view diff)
#
kylewm you can play with it if you want
#
kylewm https://kylewm.com/admin/addressbook
#
KartikPrabhu ooh
#
KartikPrabhu why do you have 2 entries of yourself :P
#
gRegor` It's pretty cool that we have someone coming to HWC for the first time this week, and they're already posting on their site and POSSEing to twitter.
#
kylewm just noticed that too
#
kylewm one is http, one is https
#
KartikPrabhu kylewm: does it do rel-me lookup for Twitter and FB?
#
KartikPrabhu gregor`: Denton?
#
gRegor` Yeah
#
kylewm I think so :)
#
KartikPrabhu ok so here is the use-case. I don't link to my silo profiles from my homepage. I have them on my about-page... and kylewm's "profile fetcher" does not pick them up. Is there a way for it to do that kylewm?
#
rascul use <link>
#
snarfed KartikPrabhu: http://indiewebcamp.com/h-card#Brainstorming ?
#
KartikPrabhu rascul <link> what?
#
rascul http://indiewebcamp.com/How_to_set_up_web_sign-in_on_your_own_domain
#
gRegor` Hmm. I think there needs to be *something* to indicate https:\/\/kartikprabhu.com\/about#me has more information about you. Something beyond just the u-url mf2
#
rascul there's an example there, put the link stuff in head
#
rascul then they're not visible but stuff can find them
#
gRegor` I think, based on skimming, the uid can be used for that?
#
KartikPrabhu gregor`: yes that is my thinking
#
KartikPrabhu rascul: I don't want to put all my info in my homepage <head>
#
snarfed gRegor`: rel-me, right? http://indiewebcamp.com/h-card#Brainstorming
#
rascul oh wait i'm thinking something else
#
rascul you can ignore me
#
gRegor` snarfed: perhaps? I got quite confused months ago in a discussion (with kbs, I think) about using rel-me for that sort of purpose. :)
#
mko gRegor`: Currently, the convention is not established as such, though. It *could* be used that way. I was suggesting that we either contrive a new u- property (i.e. u-canonical similar to rel="canonical") or state as convention that u-uid is to be used in that way.
#
gRegor` I haven't revisited it in a while
#
gRegor` Oh hey, wiki. :)
#
gRegor` catchs up on /h-card
#
mko Basically, I brought the whole issue to light because I was looking into a way to create a consistent IndieProfile that apps and websites could access through parsing any h-card that refers to me, but the best I can do with that is have my "complete" IndieProfile on my homepage, which is a design problem.
#
mko :-)
#
gRegor` learns to spell
#
KartikPrabhu mko: u-uid seems like a good idea... iirc we thought of that for partial-feeds vs full-feeds http://indiewebcamp.com/h-feed#partial_feeds
#
gRegor` Aye, mko. I read most of the conversation from this morning. Have paid less attention the last hour or two though
#
mko Yep. I like using u-uid that way. It just seems like it might get potentially overloaded.
#
mko i.e. some vendors may be using that to store their user identifiers (I haven't research usage of u-uid yet)
#
mko Hence why I didn't really pursue that line of discussion yet.
#
@activefree RT @benwerd: Excited by the proposed #yxyy sessions. Creative writing! The future of storytelling! #indieweb! Hiking! I'm down for everythi… (twitter.com/_/status/487694638292008960)
#
snarfed.org edited /h-card (+266) "/* Brainstorming */ canonical rel-me" (view diff)
#
gregorlove.com edited /h-card (+218) "/* Issues */" (view diff)
#
gRegor` Do the mf2-parsers handle following rel-author links and retrieving the h-card? I was thinking they didn't yet.
bnvk joined the channel
#
gRegor` Or that rel-author was mostly legacy
#
KartikPrabhu gregor`: mf parsers should not be following links all around the place
#
gRegor` Good point
#
KartikPrabhu this is to be implemented by some code on your end with help from a mf parser
#
mko https://github.com/glennjones/microformat-node finds rel=author, but doesn't follow it.
#
gRegor` Yeah, which is what I think I've done with my wm support.
#
mko I've done a recursive mf-node follow on mine. If it finds no h-card but finds a rel=author, it'll run again on the rel=author URL.
#
mko and so on until it finds a h-card
#
gRegor` prepares an infinite loop for mko
#
mko Shush, you. I already found two infinite loops in my testing. lol
#
KartikPrabhu ha!
#
gRegor` :)
#
gRegor` Mine isn't that sophisticated. I think it will just look at the first rel-author if no h-card is found in the h-entry
#
mko It was trivial to just have it run itself over again with a new URL, so that seemed like a reasonable error handling case.
#
gRegor` But this has me thinking maybe I should just ditch the invisible h-card author on my posts and use rel-author to my /about page
#
mko I wonder how many parsers are actually seeking out the rel=author and parsing it though.
#
gRegor` That was my concern
#
mko I mean, it'd be easy enough to test. http://www.google.com/webmasters/tools/richsnippets should be the most important one to test
#
mko It covers rdfa, hcard, rel=author, and shows you what the snippet results in.
tilgovi joined the channel
#
mko I do think it's kind of annoying though that they coopt the rel=author feature for Google+ profiles.
#
gRegor` I'm more interested in indieweb / webmention support than Google's authorship stuff, but if I can get them both working together, sure.
#
snarfed feel free to update http://indiewebcamp.com/h-card#Brainstorming with any conclusions. i'm (morbidly) curious now
#
gRegor` I just want to make sure if I wm a note to indieweb folks, most/all of them can pull in my info from the rel-author
#
mko Oh, I just don't like that they coopted a pre-existing rel value to provide better visibility to people using their silo than the open pre-existing value.
#
gRegor` So I guess that's a good poll: Does your webmention parsing follow rel=author to find an h-card if none exists in the h-entry?
#
mko It will.
#
hmans Making #pants webmention compatible has been easier than expected. Woohoo! https://trello.com/c/r0dppbnO/161-webmention-support
#
hmans Granted, I've perverted the protocol a little bit, as #pants isn't actually keeping track of actual mentions so far, at least not in the way the protocol intended. But one thing after another...
tantek joined the channel
#
gRegor` "What to do with the webmention" is really a secondary topic up to you. As long as you're sending source and target parameters, you're good :)
paulcp joined the channel
#
gRegor` Nice work, hmans
#
gRegor` hmans++
#
Loqi hmans has 1 karma
#
hmans Yeah, #pants is sending just as intended, but it's doing something very specific when receiving
#
hmans it'll fetch :source, do a bit of discovery for its JSON representation (see http://hmans.io/idl994 vs. http://hmans.io/idl994.json), fetch that, load it into its local database, put it into its user's network timeline, send notifications... ah, fun stuff
#
gRegor` Does hmans.io receive webmentions now, then?
#
hmans Haven't deployed it yet.
#
hmans Putting finishing touches on it
#
hmans Once again, receiving webmentions from non-pants systems is going to be a bit disappointing
#
hmans but I'll close that gap eventually, too.
#
hmans I need to look into the indieweb thoughts on likes and reposts, too
#
tommorris wonders what qualifies a a pants system
#
hmans tommorris, pants is a project of mine.
#
tommorris Ah okay
#
hmans It currently powers hmans.io and hmans.de and ~15 other sites off the same instance, but it's fully distributed, happily shifting data back and forth. It's essentially blogging + social network
#
tommorris apologises for drunk IRCing
#
tommorris Awesome!
#
hmans *checks your timezone*
#
gRegor` hmans: Your json looks very close to what an mf2 parser would return, so might not be too difficult to bridge data from external sites (if they're marked up with mf2, of course)
#
hmans Yeah, I need to look into that.
#
hmans I'm kind of torn... I'd like to just go mf2 in the actual HTML (which I assume is the way to go, right?), but I also don't want people building custom themes to break the connectivity.
#
tommorris hmans: London :)
#
hmans tommorris, in that case you are excused.
#
gRegor` E.g. http://pin13.net/mf2/?url=http%3A%2F%2Fgregorlove.com%2F2014%2F06%2F1183%2F I don't have a uid, but could easily add one.
#
gRegor` Scroll down to "The Dream of the '90s"
#
gRegor` published vs published_at, updated vs. edited_at
#
hmans I'll take a very close look at that soon.
#
gRegor` No rush :)
#
hmans The JSON I've been serving has so far mostly been an internal protocol used by pants to pull data off other nodes, I haven't really focused on getting it in line with what standards and ideas are out there
#
hmans Pants is very indieweb, I think... if anyone feels like giving it a go, I'm happy to set you up: http://hmans.io/lrn569
#
hmans It's very early (I've been at it for ~2 weeks), but it's working fine so far.
#
hmans Hm, I think for the time being I'm going to just dump received webmentions into an array in my post records and display that.
#
hmans Then at least I don't have to have a bad conscience.
#
hmans (Depending on the links. :b)
#
hmans Is there going to be another IndieWebCamp in Brighton alongside dConstruct this year?
#
voxpelli hmans: looks like it: http://indiewebcamp.com/2014/UK
#
hmans Oh, didn't see it earlier. I'm useless.
#
hmans I probably can't come, wife is expecting a child around that time
bnvk joined the channel
#
KevinMarks the defn of UID in vcard is To specify a value that represents a globally unique identifier corresponding to the entity associated with the vCard.
#
KevinMarks so using that for "this is my main site" is a good idea. the challenge with all this rel-me identity consolidation is that your idea of what is you may change over time
#
KevinMarks so different legacy pages my have different answers
#
KevinMarks if you follow my rel-me graph you get a bit of a mess
#
voxpelli have some thoughts on playing around with rel-me graphs on my webmention endpoint eventually, see if I can get any useful data by hooking it up to my rel-me spider code
j12t and chrissaad joined the channel
#
hmans I'm happy to report that #pants is now not only webmention-compatible, but uses it as its core protocol.
#
hmans And I apologize for writing "#pants" every now and then, I know IRC doesn't have hashtags, but it's kind of becoming the project name. hashpants!
#
@hmans And just like that, #pants embraces Webmention as its core protocol. Time to go to the next @indiewebcamp! (twitter.com/_/status/487718274147561472)
#
hmans \o/
#
gRegor` Awesome!
#
hmans And it's working beautifully, with a fallback to the old protocol for #pants nodes that haven't upgraded yet. (the old protocol was essentially webmention, but with different parameter names and hardcoded endpoing URLs instead of discovery.)
#
hmans Time to implement rel=me so I can add it to the wiki... :P~
scor joined the channel
#
gRegor` About to send you a webmention, hmans
#
gRegor` crosses fingers
#
hmans Go for it. \o/
bnvk joined the channel
#
hmans It won't appear anywhere yet, but I'm already storing incoming webmentions in the database, so I should see it when it's there
#
gRegor` http://hmans.io/ping gives a 404
#
hmans did you POST to it?
#
gRegor` I sent via indiewebify.me and it appears it didn't work. Let me try manually and see what I get.
#
hmans It's been working fine for me, but I've been POSTing to it.
#
gRegor` I need to set up automatic webmention sending on my notes. It's only on aritcles currently.
#
gRegor` Just a sec.
#
hmans "Successfully sent 26 webmentions/pingbacks from http://hmans.io"
#
hmans (on indiewebify.me)
#
gRegor` Looks like it was accepted using curl. You should have one, maybe two, from http://gregorlove.com/notes/2014/07/11/2/
#
gRegor` Bah. Wrong URL
#
hmans Yup, I'm seeing one, with target hmans.io/dun302
#
gRegor` I put in-reply-to who you were replying to :)
#
hmans Yeah, I was wondering
#
hmans But hey, this is awesome
#
gRegor` Fixed. Take 3. :)
#
hmans I assume http://gregorlove.com/ is properly mf2'd, so I can play around with that a little?
#
gRegor` Er, well I don't know about properly. :) Seems I'm always learning something new to fix about it. But yes, generally should work.
#
gRegor` aaronparecki.com and tantek.com are also good, established sites.
#
hmans Just enough for me to get started will be fine. \o/
#
hmans Thanks
#
gRegor` Most regular /irc-people sites should have pretty good mf2
#
hmans Thing is, I will *probably* keep using #pants' own JSON thing for the time being, but it'd be great to be able to import non-#pants posts via mf2
#
gRegor` Sure, it wouldn't be difficult for you to map the mf2 json fields to yours.
#
gRegor` If no uid is supplied, you could presume the URL as the uid
#
bnvk just heard about Chloe, kinda crushed…
#
hmans Yeah... there's a couple of semantic assumptions #pants is making about post data right now that I need to check
#
hmans and re the GUID, it's basically just the URL without a scheme.
#
voxpelli hmans: nice with webmentions! if you want to test your parser against a number of different possible mentioners, then you can run it through https://github.com/voxpelli/node-webmention-testpinger :)
#
hmans Cool, thanks voxpelli
#
gRegor` bnvk: Yeah, super sad news :(
caseorganic joined the channel
#
bnvk gRegor`: yah. I just… fuck.
#
hmans I'm sorry. :(
#
bnvk thanks
caseorganic and paulcp joined the channel
#
hmans Regarding IndieAuth, I guess if I have two sites A and B, and B is syndicating posts from A, it'd be a bad idea to allow those posts to contain rel="me" links, right?
#
gRegor` Usually B would have a rel-me pointing back to your domain. E.g. that's what the URL field on Twitter does.
#
hmans Nah... uh...
#
hmans When you're logged into #pants, you have a timeline of posts posted by your friends.
#
hmans They're being aggregated from their sites and displayed on yours.
#
gRegor` Ah
#
hmans It's not accessible to the outside world, so I guess it's not a big problem, but theoretically a rel="me" link embedded in someone else's post could take over your site if I'm not careful
#
gRegor` Correct. The rel=me is scoped to the page, as far as I know.
#
hmans (I'm rendering Markdown to HTML, but I'm also allowing embedded HTML, which I'm running through a fairly strict whitelister.)
#
gRegor` Interesting scenario
#
KartikPrabhu you should always scrub out the rel=whatever from external content
#
hmans Yeah
#
Loqi I agree
#
hmans I think I'm already doing it, but I'm going to double-check.
#
gRegor` KartikPrabhu: Perhaps another checkmention test to add :)
wolftune joined the channel
#
KartikPrabhu checkmention is mostly doing cross-site type security but yeah this would be a good check for any external embedded content
#
KartikPrabhu you don't want spurious rels in your page
#
hmans Oh god, I'm not filtering rel yet :(
#
hmans Please don't hack my site.
#
hmans ;-)
#
gRegor` Haha
#
gRegor` Yeah, I think it could be easy to focus on fixing XSS and overlook what appears to be "harmless" HTML
#
gRegor` I don't think many indieweb people are showing the HTML of webmentions anyway. And if so, they definitely should be whitelisting just a few HTML elements
#
aaronpk mko: I wrote an omniauth gem for indieauth already https://github.com/aaronpk/omniauth-indieauth
#
Loqi aaronpk: tantek left you a message 4 hours, 8 minutes ago: sounds good to try a variant on 2014-07-30 - go for it and update the HWC event page for that day accordingly with your thoughts and let's make it happen!
#
aaronpk It looks for a h-card on the URL and provides the name and profile photo for the user if set
#
aaronpk If available
#
aaronpk It would be nice to look for a more full h-card linked from he home page since it seems like that's a thing people are doing
#
mko Yeah. I saw your OmniAuth gem. It was one of the references I used while working on my Passport-IndieAuth implementation (which I just finished a rough version of and got working on my prototype).
#
mko It was quite easy to follow.
#
mko /Passport just uses a more complete user profile, which was how I started thinking about this whole crazy canonical/extended h-card that I've been thinking about as "IndieProfile"
#
mko It uses the Portable Contacts contact schema: http://passportjs.org/guide/profile/
#
mko Not my favorite schema, but it's a standard in the Passport community.
#
Loqi [mention] https://brid-gy.appspot.com/like/twitter/t/487370638374363136/17958179 linked to http://indiewebcamp.com/2014#East_Photos (webmention)
chrissaad and pauloppenheim joined the channel
#
voxpelli mko: seems like https://github.com/voxpelli/relspider could interest you
#
mko voxpelli: Interesting. Do you have a hosted version running anywhere?
#
mko The demo link has an error page.
#
voxpelli mko: Yeah, I haven't taken time to migrate to a new Neo4j addon on Heroku after the last one shut down
#
mko Ah, gotcha.
#
voxpelli but to get a good user experience I think one really needs to pre-crawl and relations and have those cached already
chrissaad joined the channel
#
mko I hate working with Heroku instances. I know they're really actually very nice, but I don't mind the devops of a real server.
#
mko I definitely like the looks of this.
#
mko voxpelli++
#
Loqi voxpelli has 3 karma
#
voxpelli yeah, I'm just happy I don't have to remember to update my Linux as well :P
#
voxpelli mko: the dependencies are probably a bit out of date and such, has been focusing on my webmentions endpoint lately, but would love to give it some more focus of people would want to use it
#
mko Honestly, precrawling and caching all of the relations between potential identities that a person can be referenced by seems a bit overkill. It really intrigues me, but I don't see it as a practical approach to consistent identity access.
#
mko Obviously, the most consistent identity access is to only have one URL that represents you and use that *everywhere* but since that's clearly not covering every use case, I still feel like having some sort of standard access mechanism to a canonical h-card is valuable.
#
voxpelli depends on the use case, but you might have to jump a few steps anyway
#
hmans For IndieAuth, do *all* linked services need to point back at my domain?
#
voxpelli hmans: yes
#
hmans Ow :/
#
mko hmans: Only the ones you want to log into using IndieAuth.
#
hmans Oh, will it give me a choice?
#
mko hmans: i.e. if you're only going to use Twitter and GitHub to log in, you can set your App.net to whatever you want.
#
hmans and still link it?
#
mko Yeah. Go to indieauth.com and test it out. You'll see the flow.
#
hmans alright
#
hmans cheers.
#
hmans I'm facing the following problems:
#
hmans I'm German, so I'm deeply schizophrenic
#
voxpelli That's implementation specific though – I would prefer if it could infer a link from my site to eg. Twitter by going -> me -> me about -> G+ -> Twitter – and rel-me allows that
#
hmans Because I have a "German identity" (hmans.de) and an "international identity" (hmans.io)
#
hmans While at the same time I only have one Twitter account, one Github account etc.
#
hmans This complicates things a bit for me.
#
mko Right. Makes sense.
#
mko http://f.cl.ly/items/2n2a2E0y0t1301061y2Y/Screen%20Shot%202014-07-11%20at%204.07.20%20PM.png
#
voxpelli hmans: I have a similar setup, which I wrote about at http://indiewebcamp.com/User:Kodfabrik.se – kind of separating a professional and personal profile
#
mko My github account name changed to /mko recently, which is why it's giving me the error.
#
mko But that's the same kind of error you'd see if you didn't link back to yourself.
#
hmans voxpelli, hey, I remember you! We talked when you were still with Flattr.
#
hmans I think I was bugging you with API related questions back when I was building sloblog.io
#
voxpelli hmans: very likely :) (that's btw where the ideas to relspider came from)
chrissaad joined the channel
#
hmans Oh wait, couldn't hmans.io "vouch" for hmans.de simply by including a rel="me" link? Would IndieAuth grok that?
#
mko Not sure how far IndieAuth's scan goes. aaronpk would be the right person to ask on that one.
#
mko Or maybe rel=alternate
#
voxpelli hmans: that would work from a rel-me perspective, but I don't know if IndieAuth supports it
#
voxpelli one have to stop following rel-me at some point because it could pretty jump forward infinitly
#
KartikPrabhu I don't think IndieAuth goes more that one step in the rel=me chain
#
hmans :|
#
hmans Oh well, it's a stark.
#
hmans start.
#
KartikPrabhu it goes from yoursite -> Github and then verifies Github > yoursite and allows you to log in
#
hmans It's not a deal-breaker or anything.
#
hmans Yeah, I know
#
hmans my problem is that I have two domains I want to use IndieAuth with, but only one Github/Twitter/etc.
#
hmans It's an edge case, admittedly.
#
KartikPrabhu having it do anything more than 1 link in the rel-me chain opens it up to "when to stop following the chain" issues
#
mko You could set one to GitHub and one to Twitter.
#
voxpelli Having IndieAuth follow at least two steps would solve much
#
mko Pretty easy to handle 2 domains to IndieAuth. Granted, you probably want a little bit of silo redundancy.
#
KartikPrabhu why no three?
#
kylewm harder to come up with a use-case for three
#
voxpelli KartikPrabhu: no proven use case for three? But yeah, I would prefer if it actually followed all steps
#
KartikPrabhu that is not scalabel at all... anywya upto arronpk
#
KartikPrabhu kylewm: what if I have 3 websites I want to use IndieAuth with
#
voxpelli KartikPrabhu: You would need something like Googles former Social Graph API
#
kylewm hmans: you can also use a PGP key or a TOTP app on your phone
#
kylewm those are not tied to a particular silo
#
hmans Oh? #pgp
#
kylewm KartikPrabhu: just have k-way rel-mes?
#
KartikPrabhu voxpelli: yeah too complicated. Which is my point. IndieAuth right now is nice and sinple
#
voxpelli KartikPrabhu: but a bit arbitrary in how it follows relationships
#
KartikPrabhu voxpelli: arbitrary how?
#
mko voxpelli: Not really. It only looks for mutual rel=me.
#
KartikPrabhu yeah... it is the simplest thing to do
#
mko It doesn't "follow" rel=me chains. It's just looking for a mutual relationship.
#
voxpelli following just a single step away is because of ease of implementation as far as I understand
#
voxpelli not because site's has shown that rel-me relations never chain
#
hmans I clearly need to read up on Indieauth.
#
hmans Does it need to my authorization on Twitter because it needs to get my Twitter profile URL via its API?
snarfed joined the channel
#
voxpelli hmans: I think Indieauth is an implemenentation of RelMeAuth: http://indiewebcamp.com/RelMeAuth
#
hmans I'll do some more reading.
#
voxpelli hmans: it needs it as a proof that you are the owner if the account
tilgovi joined the channel
#
hmans Yeah, makes sense.
#
snarfed mko: just saw the Passport/Poco connection you mentioned. cool! i've done some work with that myself: https://github.com/snarfed/portablecontacts-unofficial
snarfed joined the channel
#
peat hmans: Indieauth does RelMeAuth for several providers, but is also kind of a playground for non OAuth methods as well, like the GPG signing trick: https://indieauth.com/gpg
#
hmans That is great.
caseorganic and snarfed joined the channel
#
Loqi [mention] https://brid-gy.appspot.com/like/twitter/t/487370638374363136/9244262 linked to http://indiewebcamp.com/2014#East_Photos (webmention)
#
gRegor` Re: rel-me chaining, "Three shall be the number thou shall chain. Four shall thou not chain... Five is right out."
#
mko Has anyone ever imported their entire Facebook and Twitter histories into their IndieWeb enabled site? I downloaded my entire tweet history and entire Facebook account, but the Facebook data is not at all tagged or labeled in a way that would be easy to import. Twitter gave full JSON objects of every tweet from their database. Facebook does this pretty print bullshit that destroys the usefulness of the export.
#
KartikPrabhu mko: not surprised at FB being useless. I tried downloading my twitter but for some reason it doesn't download anything on FF.
#
kylewm mko, yes twitter, no facebook
#
kylewm sounds like i'm glad i didn't try
#
mko kylewm: did you happen to open-source your import scripts? :-)
#
kylewm mko: looks like no, sorry :( that was early days of my site before i got organized :P
#
mko Here's an example of a "Photo" post from the Facebook archive.
#
mko https://gist.github.com/mko/7944a21887576ad3edd7
#
mko And yes, it is not tidy HTML at all.
#
mko That's exactly how it was delivered
#
KartikPrabhu blergh
#
KartikPrabhu does not like div-soup
#
mko It's not even good div-soup.
#
KartikPrabhu there is no good div-soup
#
mko It's like it was intentionally done in a way that couldn't have been scraped.
#
KartikPrabhu mko: i suspect that is true
#
KartikPrabhu I've stopped putting things to FB entirely now
#
mko I can't really do that. The majority of my non-immediate social network is Facebook-focused right now.
#
KartikPrabhu that is unfortunate
#
mko It was Twitter for a bit but the pendulum swung back to FB.
#
snarfed mko: i have a dormant project that imports fb, twitter, g+, and instagram. all posts, w/metadata, comments, etc. it's rough but works. http://www.freedom.io/
#
mko snarfed: I always knew I liked you.
#
snarfed aww :P
#
snarfed emphasis on rough though!
#
mko I don't mind rough.
#
mko If it's better than that div nightmare for Facebook, I'm happy to have them liberated.
#
mko lol, well, Dropbox (the only one of those options that I actually have reasonably set up) doesn't work.
#
snarfed very. generally, don't bother scraping the silos if at all possible. the APIs are all way better
#
snarfed doh!
#
snarfed heh yeah, dropbox is by far the least tested/supported
#
mko Oh, I wasn't trying to scrape the silos. I was trying to convert the downloaded archive into a usable format.
#
snarfed right, i see
#
mko Since now Facebook lets you export your entire account.
#
mko Including metadata on every ad you've ever clicked on, every security alert you've had, every wall post, and more.
#
snarfed mko: you expect to write some code to import them in a custom way? if so, check out https://github.com/snarfed/activitystreams-unofficial . has both python lib and REST API for all four
#
hmans regarding IndieAuth, if it authorizes me on Twitter in order to prove that I'm the owner of the account, why does it need my Twitter bio to be configured to link to my indieauth domain? It doesn't seem there is a technical requirement for it.
#
mko It just exports it in this really crap format.
#
snarfed hmans: that proves that you also own the domain
#
snarfed mko: yeah. the format is aimed at people, not code
#
hmans snarfed, but if I own the Twitter account, I can set it to anything
#
hmans so it doesn't really prove anything.
#
mko hmans: It won't validate the Twitter account to even attempt a login without the mutual rel=me
#
mko It's kind of like two-factor authentication.
#
mko RelMeAuth provides the first factor (i.e. like a password). Then verifying your identity via Twitter provides the second factor (i.e. like receiving a text message on your phone or whatnot).
#
snarfed hmans: right. anyone can point at your twitter account with a rel-me. anyone can also point their twitter profile at your domain. only you can do both, which proves you own both.
#
hmans But what attack vector is it supposed to prevent?