#aaronpkthe other reason for returning it there is multi-user sites like Known return the user's profile URL who actually logged in, when the URL they may have typed at the beginning was just the root domain
#aaronpkokay yeah the request to get the token is the first time the client knows the final "me" URL
#cweiskebut why is the URL allowed to change between authorization and token fetching?
#aaronpkin order to support multi-user sites, or aliasing equivalent URLs such as "http://aaronparecki.com" and "https://aaronparecki.com/"
#aaronpkthat second one is common, where the user enters "http://aaronparecki.com" or even just "aaronparecki.com" but the auth server returns the full https URL
#aaronpkright but at the time of the redirect, that code (and any other parameter in the redirect) can't be trusted by the client yet
#aaronpkand in this flow, the client never verifies anything against the auth server, since it just exchanges the code for an access token at the token endpoint
#cweiskethen the client never needs to fetch a token
#aaronpkin that case, the client still has to verify the auth code (at the auth server) so it wouldn't be needed there either as long as the client is using "state"
#aaronpkoh yeah that reminds me of the other thing i was supposed to review about "state"
#aaronpkI think it's not required to send the state in either the token request or the auth code verification request
#aaronpksince it's really a mechanism for protecting the client against CSRF (and the client can use it as a session key if it's not using a cookie for that)
#cweiskeor is this a leftover from the "Self-Encoded Tokens" example?
#aaronpki think that's a leftover from self encoded tokens
#aaronpkthe micropub endpoint needs to know 1) whether the token is valid (the token endpoint should error out if it's not), and 2) the "me" URL, and client_id and scope
#aaronpkthose are the values that my micropub endpoint actually uses when deciding how to process the micropub request
#aaronpke.g. the micropub endpoint can reject the request if the scope isn't valid, and i have some rules that say "this app can only create posts in this category" so that uses the client_id,
#cweiskeI can't just say "micropub cli client 1.2.3"?
#aaronpkthe client_id URL is where the client can look up the "registered" redirect URIs plus fetch application info like name and app icon to use in the auth screen
#cweiskeah I remeber the redirect URI verification
#aaronpki think the redirect URI stuff is only mentioned in brainstorming right now, but the app info is used by indieauth.com and my p3k apps
#cweiskemy auth endpoint does not do redirect verification
#aaronpki think that would be a reasonable request for Quill, to basically handle login redirects better
#cweiskeso blogs could have a "add comment" button, which links to my comment hosting silo, preconfigured with the reply-to-url
#aaronpkyou want to provide an "in-reply-to" URL along with the login request so that after they log in with Quill it takes them to the screen with the reply URL already populated
#cweiskeand the silo would then redirect to quill, passing the "me" url, so that quill automatically redirects to login
#aaronpkyour p-name was the same as p-content so it disregarded it as a name
#tantekmy point is that bookmarks are special treated by indienews, thus they should be by /this-week as well!
#aaronpkyes i'm saying it is a bug where the permalink is not included when there is no name
#tantekthat is, the "name" of a bookmark post should be taken from the thing being bookmarked and similarly its URL, and then optionally show "bookmarked by" or something for the outer h-entry author info etc.
#aaronpkwhere "no name" means "name == content" because that's the heuristic it has to use
#tantekI'm saying that's inconsistent with the heuristic that indienews already uses
#tantekwhere it uses the name of the thing being bookmarked, not the bookmark post
#aaronpkthe thing being bookmarked doesn't always have a parsable name
#aaronpkso it uses the name that is on the site that bookmarks it
#tantekso it needs an explicit h-cite on the u-bookmark-of?
#tantekwe need to help him make an IWC happen near him! get another 3 person / implementation SWAT0 demo happening
#tantekThe long list of "Changed Wiki Pages" is making me wonder if there would be some way of giving each of them unique-ish icons to indicate what kind of page they were
#tantekso they might have icons in addition to whatever grouping
#tantekthen we have building block pages (which I'm torn about, because OTOH, user features like "album", OTOH, plumbing building blocks like protocols and formats like "Webmention")
#tantekmaybe we have "technology" building blocks (a nicer name for plumbing), and "feature" building blocks? (just thinking out loud here)
#aaronpkthat's an interesting use case for categories
#aaronpki use categories on a couple things, i think only indieauth and p3k pages
#tantekI remember being vaguely against it, but now having come upon this use-case, I wanted to at least present / document the evidence possibly for it
#tantekit feels like the kind of thing that MediaWiki Categories were meant for
#tantekI've also been wanting a way to distinguish building blocks, in order to prioritize feature-centric thinking (e.g. what IndieMark does)
#aaronpkwithout having to maintain a separate list of things with that category
#tantekrather than plumbing/protocol centric building blocks (webmention, microformats, h-* etc.)
#gRegorLoveI don't know if I advocated them, but when I first came into IWC I was used to them from Wikipedia so thought it odd we weren't using them much. Havne't seen a whole lot of use for them for IWC yet, though you can do some cool things with listing links to pages in a category.
#gRegorLoveI vaguely recall trying to do something in the last six months with them, but it required newer MW
#aaronpki use them on my recipe wiki to tag recipes based on a few categories
#tantekgRegorLove: I think we didn't use them because we didn't have well defined uses for them
#tantekif there is no incentivizing use-case, then categories are not maintained, and then those "neat listing links to pages in a category" pages become horribly inaccurate / misleading representations of what's on the wiki
#tantekaaronpk, that's not been the case in practice
#tanteke.g. lists of indieweb implementations of stuff = one edit on the appropriate "indieweb examples" section
#tanteksometimes with *one* edit you can add *multiple things / people* !
#tantekwhereas if you depend on Category: then you MUST do one edit per thing / person
#aaronpkif I add a new p3k project, then I have to create the page for it and also add it to /p3k. with categories, i can just add the category to the project page itself and /Category:p3k is updated automatically
#aaronpkthings like indieweb examples are different though, because there's additional information *about* that thing that's effectively added to the tag
#tantekright, that's the one kind of edit that tht helps with
#gRegorLoveThinking out loud: if we got in the practice of adding sub-pages under our User page for implementations, then the category and additional information would be in one page.
KevinMarks joined the channel
#gRegorLoveNot as obvious as going to the page and editing #IndieWeb_Examples though
#aaronpkhuh i'm not sure what that would look like
#tantek2010-09-30 Six Apart shutting down Vox on September 30, 2010 (going read-only on 2010-09-15), irony: Six Apart encouraged Pownce users to switch to Vox
#tantekPownce.com shutting down, switch to Vox.com. Vox.com shutting down, switch to ... ?