#social 2017-10-26
2017-10-26 UTC
rowan joined the channel
# 
cwebber horray it's still connected
# cwebber fixed that one
ppppp, sujin, xmpp-social and alanz joined the channel
# erincandescent cwebber: I'm not sure I understand the need for a CAS - surely you're signing the "child" object's ID, not the child itself?
rowan and tantek joined the channel
# cwebber erincandescent: depends, are you including all the child's data in the signed payload?
# cwebber you have the option of doing either
# cwebber
{"object": "https://example.org/foo"}# cwebber vs
# cwebber
{"object": {"id": "https://example.org/foo", "content": "ha ha"}}alanz joined the channel
# cwebber and those *mean* different things:
# cwebber did you Create some object at this uri?
# cwebber or did you Create a Note that said a specific message?
# erincandescent How do linked data signatures even work for this?
# cwebber erincandescent: they normalize to triples before signing
# cwebber I can explain more later but I have a call
timbl and rowan joined the channel
# puckipedia I should really migrate Kroeg to proper JSON-LD
# puckipedia if (objElement["@type"].Type == JTokenType.String) Console.WriteLine("BAD PUCK.");
# puckipedia ... what was this for
# puckipedia "to": "as:Public"
# puckipedia I mean yes
# puckipedia I should've seen this coming
# puckipedia .. okay, I think this primitive JSON-LD in-hacker-y works
# puckipedia okay I should've done this a bit earlier I guess. there's a whole lot of places that hard-code the ["type"]
# puckipedia lol. I just ran the whole code from start to end, tried to get a user object. result:
{"type": [], "bto": [], "bcc": []}# puckipedia aand I have a hacky JSON-LD added! This should also pave the way for triple-based storage
rowan joined the channel
# puckipedia Kroeg now also has each instance use its own @context - https://gist.github.com/puckipedia/918e47d859a8b969b84b0a84c088585d
# puckipedia any suggestions/notes?
# cwebber done with client stuff for the day, can get back to AP
# cwebber puckipedia: I have some code sketched out to show the debug information in the test suite but it's taking longer than expected... I'm going to switch to running the actual tests against your instance and try to make sure things are working right in general
# puckipedia right. I just rewrote Kroeg to use JSON-LD so everything is probably broken
timbl joined the channel
# cwebber oh :O
# cwebber well maybe I should work on adding hte debug stuff after all then
# puckipedia haha
# puckipedia this should also easily allow for triple storage which is fun
# cwebber puckipedia: are you going to CAS your quads too? ;)
# puckipedia maaaaybe
# cwebber oh
# cwebber the whole reason I was like "debugging is gonna be hard to set up" is
# cwebber because I am a moron
# cwebber I had most of it working
# cwebber and I defined new code to return a debug-apclient
# cwebber i put that code above where it returned
# cwebber and left it returning the normal apclient
# cwebber well no wonder it wasn't calling the debug version
# puckipedia me: halleluja, it works *two seconds later, after the client loaded the data* welp it broke again
rowan joined the channel
# cwebber nice
# cwebber okay I've got par tof the debug stuff working
# cwebber needs to be prettier :P
# puckipedia now I have the issue that it translates 'orderedItems' into 'items'
# puckipedia because orderedItems is just items, but as an @list
rowan_, rowan, rowan__ and rowan___ joined the channel
# puckipedia "to:talItems": 27
# puckipedia I don't know what happened here
# cwebber apo:calypse
# Gargron hmph
# Gargron i need to steam through and implement remote accounts and follow model so cobalt can actually distribute updates
# Gargron but i'm kinda meh on that, motivation is intermittent
# puckipedia so I think we've all missed a thing in the ActivityPub spec and now my code is hurting
# puckipedia https://www.w3.org/ns/activitystreams#Public compacts into as:Public
# puckipedia ... that *is* the Public collection, right
# puckipedia yep
# puckipedia now I'm trying to find a way to not have JSON-LD do that specific compaction
# puckipedia cwebber: ^
# puckipedia like. Kroeg won't have any issues with it. but everything else will probably choke on it
eugr joined the channel
# cwebber what's wrong with as:Public?
# cwebber oh
# cwebber I see.
# cwebber the spec doesn't point out that it'll compact to as:Public
# puckipedia and I'm not sure if anyone realised
# cwebber rhiaro: ping
# puckipedia oh huh. erincandescent actually did see this coming
# puckipedia <p>TODO: New URI? Can we get as:public in the schema or such?</p>
# erincandescent I'm slightly sad that Public ended up with a http(s) URI
# erincandescent I always felt it should have been an URN or something explicitly not dereferencable
# cwebber oop
# cwebber erincandescent: yeah that makes sense
# cwebber uhm
# cwebber well
# cwebber I'd say "it's a fragment id" but I'm not happy with that answer
# erincandescent You can dereference a thing and go hunting for a fragment
# cwebber tbh there's a major downside to both fragments as not-defererenceable things and fragments as embedded-in-something-else using the same URI structure
# cwebber yeah
# cwebber if there's one thing I wish I could change about the web it would be to have some way to set a bit declaring which of those is meant :P
# erincandescent Meanwhile something like urn:w3c:activitystreams:public would be very explicitly not dereferencable but also involve major standards organization wrangling at this point (becasue e.g. urn:w3c doesn't exist)
# cwebber erincandescent: you're right
# cwebber harumph
# puckipedia tbh I'd be okay with
{"Public": "as:Public"} as I don't think many other IDs in audience will be mangled# cwebber yeah I think it'd be "fine" if people knew to look for it
# puckipedia put it in the spec, and also make it the canonical URI to use? also then we need to somehow fix Mastodon because yeah that won't interoperate anymore
# cwebber :\
# cwebber well
# cwebber I mean technically the spec isn't wrong
# xmpp-social [ajordan] It's just underspecified
# cwebber it isn't informing people of this issue though
# cwebber and that is a problem
# xmpp-social [ajordan] I wonder what tantek's thoughts will be
# puckipedia .. should I make an issue on the github?
# xmpp-social [ajordan] Normally I'd assume he'd be against it because it'd break existing impls
# xmpp-social [ajordan] But that's only because the spec isn't clear enough and they're technically violating spec (kinda?)
# puckipedia if you assume the data that's sent has to be compacted JSON-LD, then yes
# puckipedia I guess Mastodon can slowly ease in a fix
# cwebber I think that's probably the best route, get other implementations to add support
# cwebber we aren't changing *spec* behavior, but we do need to bring other implmentations up to current behavior
# puckipedia https://github.com/w3c/activitypub/issues/264
rowan_, rowan and JanKusanagi joined the channel
# cwebber puckipedia: debug functionality added!
# cwebber but it's ugly as effff
# cwebber I need to clean it up
# puckipedia nice
# cwebber oh and I just broke it of course, lol
# cwebber fixed
# cwebber there
# cwebber now it's prettier
# cwebber puckipedia: https://test.activitypub.rocks/ give it a swing
# cwebber I should probably make it so you can click-to-expand them
# cwebber instead of displaying the whole thing
# cwebber but time is limited etc
# puckipedia I'm quickly adding a quirk into the JSON-LD compacter
# puckipedia quirk is needed for Mastodon compat :<
# cwebber ;<
# puckipedia until the PR I mentioned is merged
# puckipedia aaand making a post errors
# puckipedia forgot to update this to new ASTerm syntax
# cwebber puckipedia: well all the tests run though the one that's "inconclusive" looks like it's because my thing gets confused when it got the 403 again
# cwebber puckipedia: the debug info does sem helpful
# cwebber (running against your server)
# cwebber anyway, also pleasantly, it no longer times out
# cwebber (the websocket connection)
# cwebber so anyway
# puckipedia okay I think I got the quirk in the right place
# puckipedia cwebber: hm, I'll try in a bit
# cwebber hopefully that's helpful for you to test both your thing and to tell me which issues are my fault :)
# cwebber when you turn on debugging mode it gets verrry verbose. Again, I need to add click-to-expand
# cwebber but you should have useful data at least!
# cwebber does a bit of css styling to make it more readable
# puckipedia wow I think everything broke with my JSON-LD changes
# puckipedia cwebber: HEY WHERE'S YOUR @CONTEXT :<
# puckipedia well I need to probably be able to parse non-@context objects
# puckipedia but that's what everything exploded on on the new thing
# puckipedia cwebber: aha, the "An indecisive note" fails because there's no actor on the Create
# puckipedia same with the Follow, etc
# puckipedia cwebber: wow that http debugging is so useful
# puckipedia seriously. really helps!
# puckipedia going to bed. hopefully that helps, cwebber :P
# cwebber puckipedia: wowee is it not there
# cwebber yeah I guess it wasn't because implied context which is not good enough full ack :D
# cwebber puckipedia: I'll deal with it tomorrow.
# cwebber glad the http debugging is helpful
# cwebber puckipedia: and yeah I guess I should do "the most correct thing" on this :)
# cwebber rather than the minimally acceptable thing ;)
# cwebber puckipedia: thanks for all your help testing :)