#social 2018-08-09

2018-08-09 UTC
#
nightpool[m] federation is not difficult to code.
#
nightpool[m] applications are difficult to code.
#
nightpool[m] I added dead-simple federation for a class project in less then two days.
#
nightpool[m] the "problem" of federation is the exact same problem as any other development, figuring out what UX you want, and then figuring out how to build it.
#
nightpool[m] people just have less experience with federation UX so they have less hooks with which to get started.
#
kaniini that's literally what litepub intends to provide, but okay
#
nightpool[m] litepub intends to provide a UX?
#
kaniini it intends to provide the hooks
#
nightpool[m] the "hooks" are the existing UX patterns people are familiar with
#
nightpool[m] if you have a schema of how a "instagram-like" website is supposed to work, and you're a programmer, you generally know where you'd start building one.
#
nightpool[m] most people who start building federation projects do not have a schema for how a federated website is supposed to work
#
kaniini the eventual goal is to provide those resources too
#
kaniini if we are to crush the corporate social media machine
#
kaniini we need like 50 mastodons
#
nightpool[m] we have thousands
#
kaniini i don't mean instances, i mean software
#
kaniini we need to gather the fundamental details and make them available to aspiring designers like Gargron
#
kaniini so that the Gargrons of the world can actually build these things
#
kaniini (and make them actually look good)
#
nightpool[m] nearly all ircd's are a fork of the original
#
nightpool[m] i get why people are concerned about a monoculture, but i don't think the bottleneck is "more standardization initiatives"
#
nightpool[m] frankly, we need another spec group like we need a hole in our head.
cwebber2 joined the channel
#
cwebber2 https://test.activitypub.rocks/ is back up
#
pantherse cwebber2, I've been meaning to catch you and continue or conversation about reviving activipy's dev activity.
#
aaronpk Who said we need another spec group?
#
aaronpk I absolutely agree we need to make it easier for people like Gargron to make new projects that interop with the rest of them
#
aaronpk also lol my autocorrect wanted to change interop to Interpol
#
pantherse Given what I've read so far, it seems to me someone is having problems figuring out how to get their input to a form that the language can process effectively. At the risk of sounding like an pompous jerk; but, that "issue" has been around since the days of C.
ahihi2, nightpool, xmpp-social, vasilakisfil, vasilakisfil_ and fr33domlover joined the channel
#
tuxether[m] Where do I post a remote follow? The Actor's outbox, the Target's outbox, or both?
tantek, ajordan and timbl joined the channel
#
cwebber2 tuxether[m]: to the actor's inbox
#
cwebber2 set the type to be Follow, the object to be the actor you want to follow
#
cwebber2 then they will have to either Accept or Reject that request
#
saranix tuxether[m]: c2s -> sender's outbox -> sender's server<->target server -> target inbox
#
saranix depending on whether your web browser (js) has activitypub smarts or if the server is acting as both client and server (usually is)
#
cwebber2 ah yeah
#
cwebber2 sorry I meant the target actor's inbox
#
cwebber2 not *your* actor
#
tuxether[m] thanks! it seems to work now. I'm thinking my application be able to join the fediverse really soon 🙂
#
cwebber2 if it's c2s, you indeed post to the outbox
#
cwebber2 if it's s2s, you post to the person you want to follow's inbox
#
cwebber2 (er, in c2s post to *your* outbox)
#
kaniini nightpool[m] we are just documenting a specific profile of activitypub that implementors can make use of right now.
#
kaniini if cwebber2 is open to purging @context from activitypub 1.1, then that profile could be useful as a starting point for activitypub 1.1
#
kaniini if not, it exists independently. no harm, no foul.
#
cwebber2 we're not purging json-ld from activitystreams, but we will maintain the current state where if using the base of activitystreams you don't need to use @context in activitypub because the AS2 context is implied
#
cwebber2 that is a pretty baked-in decision at this point
#
kaniini as i said, i'm not against presenting extensions in a way that json-ld parsers can make use of them specially.
#
kaniini litepub intends to present extensions in a JSON-LD compatible way, so they would behave as if there were a @context
#
kaniini at least, based on my read of the JSON-LD specification, URI-namespaced keys are equivalent
#
kaniini this allows for pure JSON implementations to extend the messages without needing JSON-LD
#
kaniini which solves the problem
#
cwebber2 kaniini: you mean where you use the full expanded URI?
#
cwebber2 for the property?
#
kaniini yes
#
cwebber2 you can certainly do that
#
kaniini then, there we go :)
#
cwebber2 I'm not sure implementors will like doing it, but it's valid
#
kaniini cwebber2 i have half a dozen top-level fediverse projects already agreeing to doing precisely that
#
kaniini so, what i would appreciate from W3C side
#
kaniini is a commitment to that same pattern in activitypub
#
kaniini in lieu of @context for extensions
#
cwebber2 I'm not sure what the trouble is... if you already have a working json-ld library, you can just compact incoming messages to your local context and then never think about json-ld again
#
kaniini the problem is that you cannot implement json-ld in statically compiled languages
#
kaniini json-ld is great if you are using javascript or a scheme derivative
#
kaniini json-ld is hell if you are using haskell, go, etc.
#
cwebber2 kaniini: I don't see how you can have an easier time with the extension system you proposed
#
kaniini simple
#
cwebber2 when you compact locally as I described
#
cwebber2 you'd end up with what you said
#
cwebber2 where the URIs not in your local context are expanded
#
cwebber2 and from that point, you're in the same boat
#
kaniini well, if you can update activitypub to say to do that
#
kaniini then we can agree to disagree
#
kaniini on the rest of json-ld
#
csarven Who is using Haskell/miranda... with ActivityPub/JSON-LD right now? I'm geunuienly curious.
#
kaniini and i feel this would be the most productive outcome :)
#
kaniini see
#
kaniini csarven Who is using Haskell/miranda... with ActivityPub/JSON-LD right now? I'm geunuienly curious.
#
kaniini this right here, is getting back into the danger zone
#
cwebber2 I'm fine with adding non-normative errata explaining how to compact to a local context
#
cwebber2 kaniini: here's the thing though: the reason haskell and etc have trouble with open world systems like json-ld is that you can have an unlimited number of other fields
#
kaniini nobody is using Haskell/miranda with it because it is too much of a pain
#
cwebber2 but that's the case also with any open world system, including when you have expanded URIs
#
kaniini cwebber2 yes, what we are saying, as implementors
#
cwebber2 because the records aren't set up to have them
#
tuxether[m] there are people in this chatroom using statically typed languages including haskell. json-ld isn't impossible, just hard
#
kaniini is that we do not wish to consider it an open world system
#
cwebber2 kaniini: then don't have extensions at all, right?
#
kaniini no, rigidity and extensibility are compatible
#
kaniini but you have to understand something very important
#
cwebber2 could you explain how to solve the restricted record scenario
#
kaniini every activitypub message
#
kaniini has side effects
#
cwebber2 and how it's different once you have an expanded message
#
kaniini i give up
#
cwebber2 that's true but a different issue
#
cwebber2 these are two different issues, happy to discuss them separately:
#
cwebber2 side effects of activities
#
cwebber2 and mapping extensions to statically typed languages
#
cwebber2 if you have extensions, it's open world
#
kaniini it is very simple
#
kaniini you just don't process
#
kaniini extensions you don't care about
#
cwebber2 you can do that already
#
kaniini yes
#
kaniini you can
#
kaniini what we are saying is
#
kaniini we can just process
#
kaniini https://pleroma.social/foo
#
kaniini properties
#
kaniini by hand
#
kaniini without json-ld
#
cwebber2 ok, that's a different statement though then about the record restriction of statically typed languages
#
kaniini if "https://pleroma.social/foo" in data["object"], do: [whatever]
#
kaniini but the thing is, all of this @context stuff
#
kaniini is not usable
#
cwebber2 sure, you have to compact to your local context. there is that step
#
cwebber2 again, it'll look like what you described once you do that
#
kaniini yes, and what i am saying is
#
cwebber2 so I think what you're saying is
#
kaniini activitypub servers must compact when sending on the wire
#
kaniini which solves @context
#
cwebber2 sure, they must compact when something comes in
#
cwebber2 as long as they're using extensions
#
kaniini i'm talking about *sending*
#
cwebber2 we already recommend sending in compacted form
#
kaniini they should also compact when *receiving*, but in litepub, we can just be like "oh, your extensions were sent wrong, sorry"
#
cwebber2 the problem is whether or not the *extensions* are the same on your end
#
cwebber2 and that's why you compact on incoming
#
kaniini that's the entire point of using the URIs
#
kaniini on the wire
#
cwebber2 kaniini: I think what you mean isn't compacting
#
cwebber2 you mean expand the extension properties
#
cwebber2 you've got it backwards
#
csarven !
#
kaniini okay, whichever
#
kaniini the point is
#
kaniini {"https://pleroma.social/foo": "bar"}
#
kaniini do it that way
#
kaniini no @context
#
kaniini geeze
#
cwebber2 kaniini: if you want to do that, that's fine, and it's even valid json-ld
#
kaniini yes, i know it is
#
kaniini that's why we plan to specify the extension method that way
#
cwebber2 I think we aren't going to be mandating it, but if implementations all migrate towards doing that as best practice, then that can become noted as the established best practice
#
kaniini what i am asking is for W3C to have consensus with us on doing it that way on the wire
#
kaniini when AP 1.1 comes out
#
cwebber2 kaniini: if you get all implementations to move towards this, then when AP 1.1 comes out it's a likely event that we could have that as implementation advice in the spec that this is established best practice
#
kaniini cwebber2++
#
Loqi cwebber2 has 19 karma over the last year
#
cwebber2 I'm doubtful it will happen but that's something to test towards if you want to see that outcome
#
cwebber2 we can't say that it's mandatory, because we can't break backwards compatibility with the 1.0 spec like that either
#
kaniini it's already agreed to by multiple projects :)
#
cwebber2 but I agree it's a valid way to do json-ld
#
kaniini and will be part of litepub
#
cwebber2 well, then see if it becomes the consensus in running code
#
kaniini litepub is not your enemy, it's really your friend
#
kaniini the goal is to dial in the parts of the spec that were underspecified using realworld projects
#
cwebber2 I don't have any problem with exploring a subset of activitypub, as long as everyone collaborates politely
#
cwebber2 kaniini: the main reason I was frustrated the other day was all the negative language was making this chatroom really stressful for myself and I know at least a few others
#
cwebber2 so can we avoid that when we work together?
#
cwebber2 if we can do that, then awesome
#
kaniini JSON-LD makes me angry
#
kaniini :D
#
cwebber2 well, we have a code of conduct
#
cwebber2 we have to treat each other with respect
#
cwebber2 we don't have to agree
#
cwebber2 and it's okay for tech stuff to make us frustrated
#
cwebber2 but let's do so cooperatively and in the best spirits we can
#
kaniini yes, we can collaborate in a polite way
#
cwebber2 great
#
kaniini and in fact, you're welcome to drop by #litepub at freenode
#
kaniini but litepub uses a different consensus model, it is meant to be biased towards implementors
#
cwebber2 rough consensus and running code, I get it
#
cwebber2 kaniini: I'd encourage you to not be totally dismissive of the consensus model we have here though before trying to participate in it. We also prioritize running code here, but also some other processes
#
cwebber2 it ain't perfect
#
cwebber2 but I think it's not as bad as I've seen some messages make it out to be
#
cwebber2 and this is in general, despite some of the complaints I've seen, not a "corporate-controlled" group
#
kaniini i said the W3C in general is corporate :p
#
cwebber2 in fact this has been quite unusual, and the working group was even at some risk at some point, in that we had very little traditional corporate backing
#
cwebber2 well and I don't disagree there
#
cwebber2 my commentary on w3c membership structure being problematic is well known
#
cwebber2 anyway
#
cwebber2 happy we got to a more friendly place kaniini, hope it continues
#
kaniini sorry, it is just every time i try to implement JSON-LD in elixir, it winds up being obnoxious :P
#
saranix IMGO this is the only w3c community group that *actually works with the community* and not just some privileged ivory tower subset of it
#
kaniini yes
#
saranix IMHO
#
kaniini anyway, you might be surprised. there's a few people on #litepub who are giving AP a new chance because we intend for the spec to be rigid.
#
kaniini including diaspora
#
cwebber2 great!
#
cwebber2 happy to hear it
#
pantherse cwebber2, ping
#
tuxether[m] If I create a document via C2S and cc a remote actor, is it the server's job to trigger S2S?
#
saranix yep
vasilakisfil_ and vasilakisfil joined the channel
#
kaniini cwebber2 i mean, obviously, litepub is not that
#
kaniini cwebber2 but the problem with these particular script kiddies
#
kaniini cwebber2 is that they will just keep doing it again and again
#
cwebber2 kaniini: so keep kickbanning them
#
kaniini i mean
#
kaniini id id ban them
#
kaniini you didn't let me finish :)
#
cwebber2 kaniini: ok, I'll rejoin, but something that could help
#
cwebber2 put up a code of conduct
#
cwebber2 at least make it absolutely clear that from the litepub maintainers' point of view
#
cwebber2 that stuff is not ok
#
cwebber2 put it in the /topic even
#
cwebber2 Contributor Covenant Code of Conduct is a good one
#
cwebber2 https://www.contributor-covenant.org/
#
kaniini i agree
#
kaniini i just have fatigue dealing with this fucker
#
kaniini ever since i posted in fediverse that kloeri was not a pedophile, he has been chewing me up
#
kaniini i figured antissh would make him leave me alone, i guess not :p
#
cwebber2 :(
timbl joined the channel
#
kaniini cwebber2 :D
#
cwebber2 gotta go for a while. I have guests coming over soon.
#
cwebber2 later
timbl, jondashkyle_, taravancil_, dwhly_, bitbear_, mattl, melody, Loqi_, ma1uta, schmarty[m], cdchapman, Gargron, er1n and ahihi joined the channel