#social 2018-08-10

2018-08-10 UTC
cdchapman, tantek, ajordan, xmpp-social, vasilakisfil, martijnvdven and cwebber2 joined the channel
#
kaniini_ cwebber2 dust has mostly settled
saranix joined the channel
#
cwebber2 kaniini_: woo
#
tuxether[m] a lot of new people are learning about AP thanks to the spam kaniini is facing :D
#
cwebber2 well... I wish they could learn about it a different way ;P
#
cwebber2 but that's good I guess
#
cwebber2 I'm sure kaniini_ wishes that too ;P
#
csarven Maybe freenode will even run on AP
#
cwebber2 lol
#
csarven I'm pretty sure we'll be at the drawing board in about 10-20 years discussing how to make social stuff work on the Web.. that's better than IRC and email.
#
kaniini_ cwebber2 actually i came up with a pretty clever idea
#
kaniini_ i am going to write an irc bot
#
kaniini_ that also has an AP inbox
#
kaniini_ so to ensure the meetings do not get crapflooded
#
kaniini_ we can lock down the channel
#
kaniini_ and if somebody needs to be let in, they can message the bot in fediverse with a UUID they request from the bot in IRC
#
cwebber2 kaniini_: an interesting idea... are you talking about for on #litepub or for in this channel?
#
kaniini_ whereever needed
#
cwebber2 got it
#
cwebber2 well I think it would be an interesting thing to see experimented with
#
cwebber2 give it a shot!
#
kaniini_ the problem is that IRC is essentially an unauthenticated platform
#
kaniini_ but if we can force people to prove identity through using fediverse
#
kaniini_ then we can trust they aren't a bot ;)
#
cwebber2 kaniini_: arguably XMPP MUC is much better than IRC in that regard as well
#
cwebber2 XMPP MUC is such a nice protocol. Too bad all the clients that use XMPP aren't also nicely designed to look nice with multi-user-chat environments
#
cwebber2 speaking of meetings, I guess we didn't have one this week so we should have one next week
#
cwebber2 I know everyone's sick of the AS2 extension mechanism conversation... I think evanp has ideas on how to wrap that up
#
kaniini_ XMPP MUC has a lot of design flaws
#
cwebber2 kaniini_: I haven't actually implemented it, just read the spec some large number of years ago
#
kaniini_ for example, it sends the same XMPP message to each individual user
#
kaniini_ there's no topology knowledge like matrix or AP sharedInbox
#
kaniini_ inefficient
#
cwebber2 https://github.com/piprate/json-gold json-ld lib for Go
#
Loqi [piprate] json-gold: A JSON-LD processor for Go
ahihi joined the channel
#
cwebber2 no Linked Data Signatures support yet, but apparently is coming https://github.com/piprate/json-gold/issues/3#issuecomment-411461864
#
Loqi [kazarena] @BigBlueHat we have support for LDS (2015 flavour) internally, but haven't exposed it in a public library. For 2 reasons: 1) we were waiting for the new standard to do it and also add support for the LDS test suite 2) we use a content-addressable...
#
tuxether[m] cwebber2: that is cool! but you lose the benefits of static typing when using that library.
#
donpdonp someone got a link of some json-ld examples showing it in original form, then 'expanded' form?
#
donpdonp json-gold can expand/compact/flatten/frame. /me hmms
#
aaronpk https://json-ld.org/playground/
#
donpdonp woah. thx.
#
donpdonp wow. there's more to json-ld than meets the eye.
#
donpdonp https://donp.org/i/9Kup.png json-ld graph of a mastodon Create/Note post.
#
pantherse cwebber2, you got a minute to talk about activipy?
ajordan joined the channel
#
cwebber2 pantherse: I'm too busy to have a deep conversation, but what's up
#
pantherse cwebber2: I was going to ask you if you got my email that I send on msg a few weeks back. I was thinking it might be easier for you to email me your thoughts for activipy going forward.
#
pantherse One thing I just noticed is activipy doesn't include the "@context". Although all discussions made it clear that it's not a requirement, I find the concept useful especially for implementing protocols that extends activitypub.
#
pantherse My thinking is one can use activipy as a jumping off point to implement the new/inheriting protocol.
#
cwebber2 pantherse: hah, yes that's true. part of the reason activipy isn't as good as it could be is I was responding to a challenge of one of the SocialWG members saying "see how far you can get without doing anything with json-ld"
#
cwebber2 and the answer is, I could do all the core things, but I regretted not baking it in to be more json-ld like earlier
#
cwebber2 the adjusted goal of activipy as I found it moving forward was to be a generalized system for working with linked data objects while feeling python-like
#
cwebber2 but... then I stopped working on it :P
#
cwebber2 pantherse: I didn't see your email, I'll look for it
#
cwebber2 you're right that I should braindump my thoughts on "where to from here for activipy"
#
pantherse I have a fork of the project in github that I can start doing work at. Just to be sure, activipy current works with python 2.x and 3.x?
timbl joined the channel
#
cwebber2 pantherse: I forget if it works with 2.x, it works with 3.x
#
cwebber2 I don't think I tried with 2.x
#
cwebber2 btw there's been a lot of misinformation about json-ld and its inclusion in ActivityPub. I wanted to clear some of it up: https://octodon.social/web/statuses/100527636732474823
#
tuxether[m] is there any guidance on whether or how long to cache the public key retrieved from someone's profile?
#
csarven cwebber2: Awesome. Thanks for that write-up!
#
fr33domlover1 tuxether[m], check web payments / security vocaulary docs if any exists? wherever that public key stuff came from?