#social 2018-07-20
2018-07-20 UTC
tantek, kaniini, xmpp-social, vasilakisfil, timbl, JanKusanagi, timbl_, jankusanagi_, uw_fluxus, cwebber2 and saranix joined the channel
# 
pantherse I have a few questions regarding the development state/intention for activipy
# JasonRobinson[m] pantherse: I'm under the impression that it will not be developed further, or if it will there will be a rewrite, I seem to remember the author cwebber2 (who is not here now) saying that, unfortunately
# pantherse Does anyone have any suggestion for an alternative Python library? I've been using it and it's currently serving my purpose. I'm also tossing the idea of possibly forking it as an alternative Python library.
# donpdonp pantherse: what is activipy missing?
# pantherse The immediate item I noticed is it expects "@id" and "@type" only; however, ActivityStream indicates it can be one or the other.
# JasonRobinson[m] a fork and tweaks do
# JasonRobinson[m] ... doesn't sound like a bad idea
# JasonRobinson[m] if it is already usable to you
# pantherse I'm also thinking keeping it in sync for future updates to ActivityPub.
# pantherse And yes, it's proven very useful.
# pantherse On another note, would it be safe to say that mastodon is the de-facto reference implementation of ActivityPub at this point? I ask because as I've been lurking in the channel I notice it comes up a lot.
# JasonRobinson[m] funkwhale is python on the backend: https://code.eliotberriot.com/funkwhale/funkwhale/tree/develop/api
# JasonRobinson[m] but their AP code is not separate
# JasonRobinson[m] pantherse: no, it just has the most users :) so many people want to make sure they federate with mastodon. I wouldn't say by any meter that it is the de-facto reference. I don't think there is one, since AP is a bit broad anyway
# JasonRobinson[m] pantherse: there is also this Python library: https://github.com/tsileo/little-boxes
# JasonRobinson[m] haven't tried it but it looks pretty good
# pantherse JasonRobinson[m], thanks for the info
# JasonRobinson[m] aaronpk: mastodon is microblogging, so I doubt it. There are no threads, each post is very individual AFAIK (maybe someone can correct if wrong). More of a question on platforms where replies are not individual items without the post. There that could depend on implementation. Though, the receiver should really respect the targeting of the post. If the reply is not targeted public, then it should not be shown publicly. Of course there is no way
# JasonRobinson[m] to enforce this as a sender :)
# JasonRobinson[m] (no thread = I mean "not threaded" in the sense of like Facebook, DIaspora, where replies are only a part of the post, not individually available)
# JasonRobinson[m] aaronpk: but they are also individual toots - mastodon just visualizes them also as threads, just like twitter
# JasonRobinson[m] it would be odd if the original target has any meaning in terms of visibility, but of course it's a possibility, that was my point I guess :)
# JasonRobinson[m] but for example on diaspora, replies don't even have visibility - they automatically inherit it from parent
# JasonRobinson[m] which makes sense, since the replies don't live individually
# JasonRobinson[m] on socialhome, I went for a monster middle - replies can be different visibility than parent even though structure is the same as diaspora/facebook :E
# JasonRobinson[m] sorry for rambling, I read the "or anything else", didn't think this was entirely related to mastodon. But speaking about pure ActivityPub, if the sender targets it not public, it sounds like a bug if the receiver shows it publicly. But of course, no idea what mastodon does
# JasonRobinson[m] sorry, by target I meant "to, cc, bcc" - targeting
# pantherse I have a curiosity regarding this topic. To make the reply non-public does this mean that the value for the target will _not_ include https://www.w3.org/ns/activitystreams#Public?
# pantherse Seems to me like it's up to the "server" how it interprets "Activities addressed to this special URI shall be accessible to all users, without authentication".
# pantherse Actually, that's incorrect. It seems like, it comes down to whether the receiving server decides that all replies inherits the visibility of the parent note.
# JasonRobinson[m] exactly. and as the reply itself has a "to, cc, bcc" recipient in ActivityPub - it sounds like a bug to me if some server would show those replies that are not with "public" in one of those fields to users that are not explicitly in "to, cc or bcc"
# nightpool[m] aaronpk: sorry i wasn't around, no idea why you received such a confusing answer haha
# nightpool[m] obviously, mastodon doesn't show non-public replies on the thread page for that status
# nightpool[m] mastodon has four levels of visibility—public, unlisted, private, and DM. none of these have substantially different behavior for replies vs non-replies
# nightpool[m] public is to: as:Public. unlisted is cc: as:Public. private is to/cc the author's followers collection and DM is "everything else"
# JasonRobinson[m] Well, this is hardly "Mastodon support channel", not everyone knows how it works 😁
# nightpool[m] Sure, sure.
cwebber2 and widowclip joined the channel
# widowclip I was reading the Hacker News comments for https://news.ycombinator.com/item?id=17577372 (Article on PeerTube's successful crowdfunding campaign, congrats by the way!) and something that struck me is that people talk a lot about centralization vs. decentralization but not about federation (much to my frustration.) In my mind there is a crucial dist
# widowclip inction to be made here, as if tomorrow Facebook, Twitter, and YouTube all implemented ActivityPub they would be federated, centralized services. Perhaps it would be useful to the promotion of the standard, and hence further adoption, to highlight the benefits of federation, how they differ from but can complement decentralized/distributed models a
# widowclip nd how federation plays with existing models. Not sure if anyone in here has some thoughts on that subject.
# widowclip Oh neat, I forget that happens.
# JasonRobinson[m] I think federation implies decentralization kind of automatically, but not the other way around. Unfortunately, when people hear "decentralised", mostly it feels they expect "blockchain" 😬
# widowclip That contradicts something I said in my last message. How does federated imply decentralized?
# dansup nightpool[m]: How does mastodon treat a GNU/Social instance that suddenly has AP support?
# puckipedia dansup: it migrates over
# dansup puckipedia: Ah okay, perfect. :)
# dansup up201705417 ^^
# JasonRobinson[m] widowclip, if its federated, isn't the information then decentralised?
# widowclip There is a single point of truth, the server your registered with
# widowclip To my understanding when you send it data it tells the servers your followers have registered with, but it doesnt have to send them a copy, it could send them a link
# widowclip If for example PeerTube was closed source there would be no way to make instances, and hence it would be centralized. Because all your data lives with one legal entity, managed according the their Privacy Policy and pertinent regulation.
# widowclip Like I originally said, Twitter could implement the spec and then Twitter and Mastodon accounts could follow each other. But in doing so they don't need to forfeit their data, just expose an API and fulfill the obligation of the spec in how they handle some of that data (ie sending it to followers on foreign servers)
# widowclip Federation can certainly be decentralized, but it could also be a walled garden, with a very friendly gatekeeper, and generous visitor policy.
# widowclip (again to my understanding)
# JasonRobinson[m] ActivityPub and The Fediverse. Do you think it can be the future of social networks?
# JasonRobinson[m] > There is a single point of truth, the server your registered with
# JasonRobinson[m] Well, this is true for all federated services, for example Mastodon, Diaspora, etc. The identity lives on one server. They are no more decentralized than say Facebook, if federation doesn't bring decentralization :)
# puckipedia this isn't quite true
# puckipedia hubzilla uses a decentral identity mechanism
# JasonRobinson[m] yeah, hubzilla is a bit different of the bunch
# JasonRobinson[m] but, it's unique
# puckipedia but, as said, it's not true for *all* federated services :P
tantek joined the channel
# nightpool[m] dansup: yeah there's a lot of migration logic, because mastodon itself migrated from ostatus to activitypub
# nightpool[m] i don't think there's anything there that would be specific to mastodon though
# nightpool[m] widowclip: i think it's a mistake to say that federation can be a walled garden
# nightpool[m] federated systems are vulnerable to EEE policies, but that's a different thing.