#tantek.comcreated /Z (+637) "having just blogged about UTC and thus "Z" time (in use by indieweb servers and projects), this seemed good to capture" (view diff)
#LoqiZ is a suffix used on a time string to indicate that it is in UTC (Coordinated Universal Time), a common technique for the time setting on indieweb servers, and used by programmers to avoid problems dealing with timezones, especially in indieweb projects https://indiewebcamp.com/Z
#tantek.comcreated /alphabet (+3831) "make a generic page for alphabet which relegates the Inc to a see also, and draft a sample set of indieweb alphabet building blocks, let's see how well they survive community scrutiny" (view diff)
#tantekbut rather than try to figure out how to expand that to the whole alphabet I decided to keep it purely about hand selecting key things in 2014
#tantekthankfully a generic "indieweb alphabet" of building blocks can be made of anything from inception to the present so that was easier :)
#tantekI think I might stop by a bookstore to quickly browse a few children's alphabet based books to load their phraseologies into my brain for contextual writing.
#LoqiA reply (or comment) is a kind of post that is a text (typically, though photos are possible too) response to some other post, that makes little or no sense without reading or at least knowing the context of the source post https://indiewebcamp.com/reply
#tantekKartikPrabhu: I plan to post it to my blog soon (in the next couple of hours) as a timely static snapshot, but then of course I expect it will evolve on its own on the wiki, and will likely even encourage the broader indieweb community to come do so
#KartikPrabhutrue! good way to do it until we actually figure out /edit posts
#tantekalright, going home now and going to hack up some ::first-letter action to see if I can do something prettier for my blog. I'll be using /custom-post-style for that, naturally :)
#tanteklet me know if I've missed anything critical you think is worth mentioning!
#tantekcritical building blocks, projects, concepts, principles etc.
#fkoomanit is distributed in the sense that anyone who wants to create a group will be able to choose their instance and other users will for the purpose of this group use the same "silo" yes
#ben_thatmustbemeBut as I understand it hes a really big hack job. Started as wiki notification bit that was hacked to run an outside file in a different language
j12t, adactio, KartikPrabhu, frzn, LanceyWork, smcgregor, glennjones, friedcell, peacekeep3r, alexhartley, mlncn, LukasRos, nedorito, chreekat, fourtonfish and snarfed joined the channel
#voxpelliRelated to discussion a while back about of the fat pings of PubSubHubbub increasingly no longer really avoids an attack of bots
sanduhrs, LauraJ and friedcell joined the channel
#aaronpk!tell fkooman re: private webmention vs inbox, do you mean that you think the contents of a private message should be sent in the payload vs retrieved like webmention verification?
#aaronpkKevinMarks__: the webmention notification here is delayed now because of the buffering that's in place to cluster the notifications
KevinMarks_ and friedcell joined the channel
#aaronpk!tell fkooman re: https://www.tuxed.net/fkooman/blog/group_communication_platform.html how am I supposed to be able to get an access token to your site in order to send you a message? that seems backwards, since that means you have to be able to issue access tokens to arbitrary domains before you have received anything
#voxpelliThe only currently "controversial" thing about that one is the WebFinger/e-mail reliance – which the "From"-header relies on – with a different header that flow would work though
#voxpelliaaronpk: well, if you want to push content to someone, then that's out of scope of WebMentions, so one would need to call the alternative something – inbox could be one name
#aaronpki don't see how that's out of scope of webmention at all
#ben_thatmustbemewonders if a way to subscribe to webmentions would suffice
#voxpelliI think they can work together – WebMention tells me that I've new content and I then answer that they can send it to me at a certain place – a "ping/pong"-moment
#tantekindeed the entire "inbox" framing is a bad worldview in that it puts too much burden on the receiver, i.e. it prefers spam-type behaviors
#ben_thatmustbemesince, as I understand it, all current indieweb implementations of PuSH are only using thin pings
#voxpelliaaronpk: well, it would more or less be a one-off subscription PuSH for an individual piece of content – PuSH can subscribe to items as well as feeds
#aaronpkthat seems like an awkward overloading of PuSH, worse than this extension of webmention
#aaronpkbtw i'm only pushing back on the name of this, not on how it works
#tantekboth "inbox" and "fatping" are part of an obsolete email-centric worldview that has been shown to be so susceptible to spam that the only implementations that can deal with it are massive centralized silos like gmail, hotmail, yahoomail etc.
#tantekwe must reframe all problems to prefer maximum number of smaller independent implementations running on a millions of smaller sites, rather than problems where the solutions lend themselves to a centralized bias
#aaronpkvoxpelli: in your diagram, who is sending a message to who? alice to bob?
#voxpelliit's basically just telling Alice that the service that owns the callback that's being sent in is allowed to fetch it for Bob and Bob has verified that with a rel-delegate just like Micropub and WebMention uses rel:s to verify delegation today
snarfed joined the channel
#kylewmtantek: I don't understand your objection to fat pings, they are only delivered/trusted if you have subscribed to a feed
#aaronpkvoxpelli: i guess i don't see the advantage to adding this many intermediate steps
#voxpelliaaronpk: one doesn't have to generate any tokens / auth codes – that's one possible advantage
#aaronpkwhy is that different from alice just sending the message in a single POST request then?
#voxpelliit would have all of the advantages of PuSH in being able to subscribe/unsubscribe and it would prohibit spammers as the callback URL:s would be unique per subscription as ordinary in PuSH
#voxpelliso it's basically just an automated way of telling a trusted sender of a trusted callback
#Loqi[mention] Larry Halff, Bettina Warburg-Johnson, Jeff Rider, Beau Smith, Ryan Sarver, Ben Ward, Kitt Hodsden, Om Malik, Carla Borsoi, Brian Behlendorf, Nate Koechley, Lauren Breuning, Colleen Taylor, Pius Uzamere, Micah Snyder, Dharmishta Rood, Erin Jo Richey, Peter Hirshberg, Justin Ormont, Elizabeth Churchill, Faruk AteÅŸ, Sarah-Jane Morris, Jay Allen, Doc Searls, Benjamin Michael Goering, Pascale Diaine, Zibi Braniecki, Leah Culver, Laura Helen Winn, Dan Gailey, Joichi Ito, Tony Rai, Yoz Grahame, Amy MacKinnon, Justin David Kruger, Matt Mullenweg, Thomas Vander Wal, Jesse Vincent, Karen Nguyen, Tom Coates, Tim O'Reilly, Lillian Christina, Chris Messina, Janet DeHart, Bobby Fishkin, Jason Shellen, Jordan Harband, Simon Law, Megs ORorke, Lonnie Rae, Marie Williams, Rohit Khare, Matthew Levine, Cariwyl Hebert, Stephen Wyatt Bush, Nima Dilmaghani, David Baron, Jeremy Anderson, Kara Murphy, Elisa Jo Harkness, and Erin Stevenson O'Connor were invited to https://indiewebcamp.com/events/2015-08-12-homebrew-website-
#voxpelliaaronpk: it would subscribe to either a single update of the private message or continuous updates for the private message – depending on the consumers – and the sender would asap send the initial state as an "update"
#aaronpkkylewm: i seem to be getting a crapload of webmentions from your HWC page, but i'm not sure who is sending them. can you think of any reason you'd be sending a couple webmentions per second to the wiki?
Garbee and KevinMarks joined the channel
#aaronpkkylewm: they are coming from 192.241.192.83
#kylewmand I'm probably sending a salmention notification for each invite, even though I'm actually hiding the invite replies
#aaronpkah yeah i was wondering why i wasn't seeing those
#aaronpkvoxpelli: okay from what I can tell, all you've really done here is added some extra steps to get to the point of the message being sent to bob's webmention endpoint
#aaronpkthere's no authentication still, so it's just as vulnerable to spam as alice initially sending the message to bob's WM endpoint
#voxpelliaaronpk: to be fair – there are more steps in the current private messaging flow than there is in mine – not saying mine is better in any way, but the amount of steps seems like the wrong criticims?
#aaronpkno i'm saying your outline is exactly equivalent to alice just sending the message to bob's WM endpoint in the first place
#aaronpk(not comparing it to the wiki private messaging flow)
#voxpelliaaronpk: the authentication are the same as with all PuSH flows – unique callback URL:s?
#aaronpkthat isn't really authentication tho is it?
#voxpelliand the verification that it's actually Alice that sends them comes from the fact that the hub-discovery is made on Alice's page
#aaronpk(I'm pretty sure the From header is not needed as well)
#aaronpkthe important thing to note is that any POST request received cannot be trusted to be from a particular server, which is what all of this lookup is solving
#voxpelliit would either be the a header or a parameter in the body or query of the request – I guess Blaine wanted to make a generic delegation mechanism, hence rel-delegation and From:
#aaronpkit bugs me a little that there doesn't need to be a URL for the message, because that makes this way less of a webby protocol
#tommorristantek: I've been thinking more about checkin and studying how Twitter does it and may build a nice prototype of a "check-in editor" - that is, a simple web front-end that does check-in right
#aaronpkbut it's neat that there's no authentication required for this
#tommorrisTwitter's location is nice because it lets you specify whether you want to be specific or general
#tommorrisso, on the Twitter iPhone app, you tap the location button and it brings up "Tag location" and I can choose between "London, England", "England, United Kingdom" and "United Kingdom" and at the bottom there's a "Share precise location"
#tommorriswhich then sends the precise geo-coords. but if that's off, it just tags it as being in London
#tantektommorris: Swarm has neighborhoods and city checkins now too
#tantek.comedited /Tumblr (+623) "re-order how tos by how to use Tumblr for indieweb hosting, export, how to POSSE etc. to Tumblr as a logical temporal / order of implementation sequence, clarify post types to the 7 currently on Tumblr" (view diff)
#Loqifkooman: aaronpk left you a message 4 hours, 21 minutes ago: re: private webmention vs inbox, do you mean that you think the contents of a private message should be sent in the payload vs retrieved like webmention verification? http://indiewebcamp.com/irc/2015-08-11/line/1439307677643
#tanteksnarfed, please review https://indiewebcamp.com/Tumblr#How_to - I tried to order by most bang/relevance for the buck as it were for Tumblr users, then those looking to transition from Tumblr to more indepenent web hosting.
#fkoomanaaronpk, yeah, why not in the payload of the POST message. It is not always ideal, e.g. when sending "big" files, but needing to fetch stuff all around the planet (and keeping a copy) is also not ideal...
#fkoomanthe "spam" problem already exists on the indieweb with distributed indieauth
#fkoomanaaronpk, the "client", whichever app that is will obtain an access_token using OAuth, just like any other OAuth client
#aaronpkfkooman: the difference is in your proposal, i'm expected to dish out access tokens to anyone who asks. the way it's currently described, I only have to dish out an access token to people I send messages to
#fkoomanaaronpk, that was not regarding your comment, but something else from the log, the spam issue when having an "inbox". If your service supports distributed indieauth a spammer can put a lot of spam content on your service using their own indieauth compatible server to grant authentication
#fkoomanbut the person who asks needs to have their own (https) website listing their authorization server and methods for authenticating
#fkoomanso the restriction is the same as logging in to IWC wiki
#aaronpkfkooman: what you're describing can be separated into two totally different things. 1) you are describing a "commons", which like the wiki, multiple people can sign in to and add things to. 2) you are describing a notification mechanism
#aaronpksure, that's one possible spam prevention technique, and also doesn't rely on oauth and would work with even existing webmentions
#aaronpkin order of increasing prevention of spam: 1) accept arbitrary notifications (email, trackback), 2) accept arbitrary notifications where you can verify the sender (pingback, webmention) 3) subscribe-first notifications (PuSH)
#fkoomanbut how is private-messaging-brainstorming better than accepting arbitrary notifications?
#aaronpkI do like how all the work is pushed off to these delegated endpoints
KevinMarks joined the channel
#fkoomanaaronpk, and how does pm.alice get the message?
#aaronpkthat's probably where Alice wrote the message
#aaronpki think it's okay that it's unspecified here
#fkoomanaaronpk, yeah, i guess it could be an oauth client that puts a message in "outbox" or something, would also be great to confirm delivery etc by moving it to 'sent' or something :)
#aaronpkit's basically equivalent to asking "how does alice write a blog post"
#aaronpkthe answer is it varies wildly depending on the implementation
#aaronpk"hey there, I heard alice wants to send message 1234 to bob. Is this legit?" pm.alice replies "yes", *then* pm.bob generates a token and tells pm.alice to send the message
#fkoomanaaronpk, to be fair, my 'solution' also verifies before sending an access token
#aaronpki'm talking about *generating* an access token, not just sending it
#fkoomanaaronpk, it is generated at the time the authorization_code is exchanged for an access_token, so after the authentication step (and thus domain validation) has taken place
#aaronpkoh i might be confusing this with something else
#fkoomanjust like how OAuth works without client credentials
#aaronpkyes you know the domain validation has passed, but you still have to generate tokens regardless of whether there is a message there
#aaronpkwhat i'm saying is if i'm writing a thing that generates tokens, i don't want to be expected to generate tokens for everybody who comes along, i only want to generate tokens if i have an interest in them
#aaronpki found the conversational text in the diagram is useful because it guides you through the steps more, but having the details below is where the actual protocol lives
glennjones joined the channel
#gRegorLoveI came into it pretty blind so didn't spend much time trying to understand the diagram earlier. :)
#gRegorLoveI can understand it now that I've read the text version and understand it better, but the diagram still seems weird to me. Could just be that I'm not used to reading these type of diagrams.
#aaronpkyeah it does take a while to get used to them
#aaronpkoauth2 dropped the initial request token, so the equivalent to oauth1's 2-legged is the client credentials grant, which is literally just one post request