#nightpoolI'm assuming there's a meeting today, but I won't be able to make it. Maybe the holiday weekend messed people's schedules up? (I know it did mine...)
#cwebber2ben_thatmustbeme: yeah I was intending to do a meeting, but if we don't want to do it this week that's ok
#sandroMy one topic right now would be in-page-social. Gargron said something about being inclined to do it in a mastodon-only way, which concerns me, of course.
#Loqi[Gargron] @akihikodaki We only have 3 options, I think:
1. "Remote follow" button, where you input your username@domain, the button needs to do a webfinger query to get the right URL template for a redirect to authorize the follow. Mind you there is no writ...
#cwebber2I think Evan and friends *had* something like this at some point....
#sandroIt's actually easy to read this is being a major deliverable that was completely ignore in the SocialWG charter.
#rhiarosandro: is that like what open social was aiming for/was doing (google gadgets stuff?) when swwg started?
#sandroI'm really not, sure, but I think it's possible.
#sandrotwitter still calls their code to do it widgets.js
#cwebber2sandro: relatedly, this might tie in with puckipedia's suggestion yesterday that we have an endpoint that tells you if you've shared/favorited something.
#nightpoolsandro: to clarify gargron's comment, I think he meant more that mastodon would be pioneering any implementation, since there's no spec in that area and (as far as I know) gnusocial and pA haven't signaled any interest in it.
#sandroThat is certainly related, yes. But there's a very, very tricky problem in how a page on my site can display whether you've shared it, or are following me, or whatever, when that information only lives on your site.
#nightpooli don't want to speak for him though, but it sounded like he was talking more about any *immediate* compatibility, rather then long-term interop.
#nightpoollike "there is no way we can do this today that will be compatible with existing code or use existing standards"
#sandroindieweb suggests using registerProtocolHandler for this, but there's no sign of that in mobile browsers, which makes it ... not ideal, I think.
#sandroif you can find any trace of how that plugin could possibly work, that'd be interesting. The only standards-like work I've ever seen in this space was webintents, but that died, and didn't quite do what's needed. https://www.w3.org/TR/web-intents/
#nightpoolben_thatmustbeme: because of conflicts with the existing protocol handling stuff apps can do, maybe?
#cwebber2visiting another person's federated site, I'd like to be able to easily have my browser coordinate and say, "hey, my user would like to like their post using their existing social site over here"
#sandroben_thatmustbeme, probably because the usability is .. hard. Even on desktop chrome, it requires the user notice weird little stuff in the title bar. On mobile, folks dont see the title bar so much. ?
#tsyesikaaaronpk: I mentioned in the github thread. We've experimented in the indeweb wiki and we use a button which requires one click but setting it up requires a lot.
#tsyesikaaaronpk: I also implemented the mastodon follow button where you type in your own site, it's an interesting approach but obviously not as good as the silos
#tsyesikaaaronpk: I'm pretty sure the only way to do this is have it part of the browser otherwise you're granting too much permission from the site your visiting. I don't think there is a way to do it securely other than using the browser or a user-agent doing the following(?)
#tsyesikacwebber2: my gut feeling is the browser has to be involved somehow. I'm worried about XSS attacks. For example if you had a like button, could you also make that handle a delete action too if you had granted the site both permissions. Does anyone have any thoughts on cross site trickery (more broad then just XSS)
#tsyesikasandro: I agree with what's been. Getting browsers to implement this stuff is going to be difficult. I think it'll be possible to do a polyfill.
#tsyesikasandro: polyfill looks as if it's in the browser but it's actually not. The site using it has to load a bit of JS that does some magic from a specific site
#tsyesikaaaronpk: when people put a subtome button on their site ... (something) it looks like a regular follow button. One solution around the browser is to get a reader support as it can add all the buttons at once and it also knows what you're doing (favouriting a post, etc.)
#tsyesikacwebber2: if we had a like-style button on the page. They'll probably be added by JS anyway (mastodon is react.js), we'll have to assume things like this will have to be robust enough for JS to add buttons like this and we have to think about JS being able to click buttons like this. Like google's click fraud
#tsyesikasandro: How I imagine polyfills it would use iframes. When I put a static facebook like button on my page it allows a user to like something without leaving the page and display how many friends have liked the page. It's bidirectional.
#tsyesika... the page can't mess with the double iframes. The browser treats them as separate origins, the iframe security is quite good.
#tsyesika... I implemented this double iframe thing. Is there interest in moving forward, we could rebrand Julian's code.
#tsyesika... someway we could get this functionality deployed across social websites
#tsyesikaaaronpk: I think we need to do some more exploring and experimentation. Our current set of options aren't ideal right now.
#tsyesikasandro: I'm glad you're saying that, the UX is too messy and there is no guarantee that it'd work on microsoft products or mobile and that's a large fraction of the web.
#nightpoolhave people mentioned multiple accounts yet? that's a usecase that protocol handlers can't provide, as far as I understand existing tech
#tsyesikacwebber2: There might be some theming miss match between sites
#tsyesikasandro: You could provide some theming but as a user I like the idea of seeing a tiny window into my own site so it should be themed like my site
#tsyesikaaaronpk: I agree. As a user I need to be able to trust that button to do something on my site so it should look like my site. The problem with button in someone's site it'll have problems with designers and such and it'll look wrong. It's why i prefer browser or reader support
#tsyesikaaaronpk: I have a bookmarklet that i can click and it lets me follow someone. It's integrated into the browser outside the thing i'm viewing
#tsyesikasandro: If it's styled in the polyfil, if it's put in the top and styled like my browser
#tsyesikaaaronpk: if it's intentionally taken outside the flow of the website. I could see myself using that. It'd also not conflict with the style of the website too
#tsyesikacwebber2: Would you be able to click the like or share or whatever on multiple things on a page with a drop down or would it only be one thing per page
#tsyesikaaaronpk: there is something in micropub (?) and it looks thorugh the href's on the page and shows things in yellow that you can do it on but one of the reasons i think readers are better because it can do things like that
#tsyesikacwebber2: nightpool was asking about multiple accounts
#aaronpks/something in micropub/a micropub chrome extension/
#tsyesikasandro: My answer is there will be a button for my currently chosen account and a drop down menu to switch to another account or add an account
#tsyesikasandro: what I'm thinking is on your page you put the follow button (with double iframes) and when i click on that (the button is displayed on my site).
#tsyesikanightpool: I was trying to say in IRC: One of the thing that happens on mastodon is that people have multiple accounts on multiple instances. I have two accounts on two different instances
#tsyesika... Looking at the protocol handler approach doesn't work well because you can only have just one of the instances buttons loaded
#tsyesikasandro: This could be another reason for the trusted domain iframe method so you can have the drop down talked about
#tsyesikasandro: you could have one site which proxies other sites, doesn't need to be a 3rd party
#tsyesikanightpool: I think it gets into the concern people from matrix or other federated networks. Matrix has a separate identity from your home server. It lets you have one identity but multiple servers but it's more complicated from a users point of view
#tsyesikacwebber2: I wanted to review as we have several proposals
#tsyesika... option 1) Use iframes to embed individual buttons (maybe browser support in the future)
#tsyesika... upsides: easy to look and reply to individual posts. downsides: theme mismatch (maybe a upside), it's more difficult to select specific users
#tsyesikanightpool: are we talking about register protocol handler or...?
#tsyesikacwebber2: higher level than that, more about the UX of how to embed this
#tsyesika... this is a bar that would appear on the top of the page, separate from the content. The advantages: it's separate from the content. disadvantages: it's harder to act on individual things on the page (multiple likes on a page or something)=
#tsyesikasandro: We probably need both of these. We probably want to act on specific things on the page and the page as a whole. I think if we get one to work, the other probably will come too
#tsyesikanightpool: when you mentioned option 1, we mentioned the difficulties of multiple accounts. I don't think that's a UX limitation, it could show a dropdown or a confirmation
#tsyesikacwebber2: I saw a demo at <somewhere> doing federated payments or something where you could one site and click pay and I wasn't sure how it worked but it was very impressive
#Loqitsyesika has 20 karma in this channel (22 overall)
#nightpoolin theory I like aaronpk's registerProtocolHandler -> intermediate domain -> rest of your accounts but it seems pretty hard to make it really work from a UX perspective.
#aaronpkto clarify: someone wrote up something they call "indie config", and at one point I figured it out and made screenshots for https://indieweb.org/indie-config but now I don't really remember how it works and couldn't help you set it up
#nightpoolsorry, someone brought up the idea of using a protocol handler to find a central domain, and then use that domain to embed different server-specific iframes, and I guess I didn't exactly catch who it was.
#aaronpkseparately: julien from superfeedr made https://www.subtome.com/ which does not use registerProtocolHandler and instead uses the domain to store things in localstorage
#tantekreads "what open social was aiming for/was doing" and wonders if there's any evidence (citations? URLs?) for what that might actually be beyond conjecture.
#tantekis really trying to make sense of telcon IRC and failing
#tantektl;dr: "We don't know how to make this work for a single account, so let's try to figure out how to make it work for multiple accounts." Really?
#nightpoolthe concern was that centering the discussion around protocol handlers won't work at all if you have multiple servers that you want to like statuses from.
#nightpoolthat's not architecture astronomy, that's how people use mastodon.
#tantekalright, I'm going to need to see screenshots on a wiki page documenting this to understand it
#tantekI have multiple accounts on Twitter, and Instagram, and they assume you're logged into one and make you switch to like things from another account.
#tanteksimilarly FB with personal account vs. pages, and you can choose at posting time which "user" to post as
#tantekthe Twitter, IG, FB examples of this I understand, and their UXes make reasonable sense
#nightpoolso the concern is that if we use web+social://like, or whatever, as a URL
#nightpoolthen only one server can handle that protocol
#aaronpkliek i mentioned on the call, it's a client issue
#aaronpkyou can make that handler be a thing that lets you choose an account
#aaronpkalso you can do it by logging out of your browser profile and logging in as a different one
#tanteknightpool, it seems pointless to theorize about the limitations of something that's not even widely implemented
#tanteki.e. you cannot definitively conclude "only one server can handle that protocol" until at least a few people have tried prototyping/implementing
#tantekthe other half is a site supporting registering of it's own indie-config handlers to then allow its user/author to click on the webaction buttons my permalinks so they can one-click respond using their own site
#tantekand then yes there is the third actor which is the browser which needs to support registerProtocolHandler
#tantekwe need a few folks to try implementing all three and seeing how they interoperate to properly reason about the possibilities
#tantekbefore voxpelli implemented indie-config and its libraries, everyone thought the whole idea was impossible
#tantekhence I reject your points about "only" or limitations until we have prototypes trying it out
#tantekaside: separately, if/when sandro is around, would be interested in understanding what he/you meant about Open Social and part of our charter we supposedly ignored / skipped / missed
#Loqi[@Lady_Ada_King] @luisbg @torgo Yesh this amazing!! Apparently somone ran a web server from a Sim card!!!!
#tantekcsarven, citing an incredible claim without citation for said claim from a tweet? Surprised to see you do that :) (nevermind the "someone ran ..." weaselwording ;) )
#csarvenI've found the idea itself interesting enough.
#csarvenHow do you find it different than your personal computer?
#tantekmy personal computer does not have implementations of insecure non-open protocols designed by telco cartels and oligopolies with a veneer of intl cooperation
#tantekhmm: "There is no major search engine that does not store our past searches or collect information on our activities" O Rly? DuckDuckGo ships in Firefox by default to hundreds of millions of users, that should be considered "major" enough
#tantekreally likes how fast duckduckgo returns results, and uses duckduckgo for his own personal site search too
#tantek"If we owned our own social graph, we could sign into a Facebook competitor — call it MyBook — and, through that network, instantly reroute all our Facebook friends’ messages to MyBook, as we reroute a phone call." <-- kind of like we do today with POSSE and backfeed, the latter of which does *precisely* "reroute all our Facebook friends’ messages to MyBook" where "MyBook" is MySite :D
#tantekThis is *precisely* why indieweb focused more on POSSE and backfeed than 'federation':
#tantek"If I can reach my Facebook friends through a different social network [via POSSE from your site to FB] and vice versa [via backfeed from FB], I am more likely to try new social networks [or just use my own personal social network, AKA personal website]."
#sandrotantek, to answer about the charter, read the last paragraph of the intro to https://github.com/swicg/general/issues/5. And my apologies, I accidentally used wiki markup instead of markdown on that text, so it was kinda unreadable. Fixed now.
#tanteksandro this? "a client-side API that lets developers embed and format third party information such as social status updates inside Web applications"
#tantekthough it is not an API, I believe jf2 is a building block for "a client-side API that lets developers embed and format third party information such as social status updates inside Web applications" and I should be able to demonstrate as such within a week
#tantekwhere I interpret "third party information such as social status updates" to include "third party RSVPs"
#tantekand "Web applications" to include personal websites like my site / blog
#sandroI agree the wording could mean a lot of other things, but still it seems odd to use the word "embed" the way the charter does. I don't really know what's meant by that if not iframe widgets.
#tantekI'll be more specific then, by "embed", *I* mean via iframe
#tantekI will admit to the word "widget" being meaningless to me and thus will drop it
#sandroAnyway, it's not important, and I mention it only in the hopes of shaking more relevant prior experience out of the bushes.
#sandroI use the word "widget" very precisely to refer to https://platform.twitter.com/widgets.js (which powers the things twitter lets you embed in static web pages), and other things that do that.
#tantekin the context of "iframe widget", no javascript is necessary so I'm confused by that reference
#tantekalso re: "This should allow Web application developers to embed and facilitate access to social communication on the Web." - I interpret "social communication on the Web" to include RSVPs from one site to an event on another site
#sandroI'm not sure about twitter's widgets, but Facebooks use iframes so that they can use the viewer's authenticated relationship with Facebook. That's what lets them say "Your friends Alice, Bob, and Charlize like this page". Thus, iframe widget.
#sandroBut your right, it either needs an iframe or javascript, not really both. It only needs both if you want some fancier things, like re-sizing the iframe and looking more like part of the browser.
#tantekeven the social graph context "Your friends" could be possible purely by iframe+cookies
#tantekto be clear, I'm not going to demonstrate getting the social graph context (yet)
#tantekbut I don't see that as an essential feature of the charter requirement you cited
#sandroI'm not saying any of this is a charter requirement. Only that "There is some text in the Social Web WG Charter that could be read as talking about this kind of functionality"
#sandroBut it could totally be read in other ways.
#sandroI have no idea what it's actually supposed to mean.
#tantekI'm saying it's something we have at least somewhat worked in the past year or so, or at least a piece of it, in the form of jf2
#sandroNo, as I said, I'm just trying to get some input from the folks who were working on this in the OpenSocial days.
#tantekAnd I think if we can demonstrate multiple realworld working impls (say 2+ clients/sites using at 1+ embedding service) of doing this kind of "social embedding" via jf2 we should reconsider putting it on REC track (which I think may involve republishing as FPWD).
#tantekTBH I don't remember seeing any realworld working impls of OpenSocial "social embedding" beyond demos and screenshots (e.g. at the Social Web Workshop in 2013) - though I admit my experience there is fairly low. KevinMarks (who did a lot more on Open Social) would be the one to ask.
#sandroPretty sure any new Rec Track work will not fly, as I read the tea leaves. But I wont actively stand in your way if you want to try, I guess.
#tantekI wouldn't bother except for the fact that I am agreeing with you that it appears we neglected "social embedding" which is explicitly in the charter, even if it is there with the example of OpenSocial (which was abandoned by everyone even its advocates)
#tantekThat would be my reason for pursuing it, a hole in our deliverables per the charter, and even then *only* after we have seriously proof of impls and some degree of interop (thus demonstrating we have a *chance* of rapidly going to CR, and then working through test suite / details)
#sandroThat would make sense except that our charter extension request should have said that, if we meant that.
#tantekWell you didn't discover the hole until now
#tantekso I think that's a reasonable extenuating circumstance
#tantekI'm really not worried about any political FOs TBH (if you're referring to what I think you're referring to), or rather, if that's what we get blocked on, that's a bridge we can worry about when (if) we get there