#social 2018-08-23

2018-08-23 UTC
ben_thatmustbeme, ben_thatmust, KevinMarks, timbl, ajordan, xmpp-social, kaniini, vasilakisfil, KevinMarks_, fr33domlover, sknebel and dansup joined the channel
#
up201705417
you might find GS tech report on ActivityPub plugin useful as well
#
up201705417
s/useful/helpful
#
nightpool[m]
including the unix philosophy is kind of lo
#
nightpool[m]
*kind of lol
#
nightpool[m]
i don't think there are many people who would stand behind "making full use of ASCII repertoire to minimise keystrokes" in this day and age
#
nightpool[m]
i think you might want to make it clear that gnusocial defines an actor as having a URI like "https://myinstance.net/nickname"—obviously activitypub puts no such restriction on actor uris
#
cwebber2
pushed some implementation reports to activitypub.rocks
#
cwebber2
pleroma, go-fed, microblog.pub
#
cwebber2
took me forever because I suck, but they're up
#
up201705417
actually I might want to fix that because GS's Actor URI do not have that format (anymore) *embarrassed* (and yes, I should emphasise that)
#
nightpool[m]
up201705417: this is a pretty small note, but "create two big guys" feels pretty informal for the tone of the rest of the report.
#
nightpool[m]
also it's not particularly idiomatic english.
#
up201705417
nightpool: In the very beginning it was stated to be an informal report but it might be too portuguese, will change that as well, just a sec (and thx by the revision)
KevinMarks joined the channel
#
up201705417
All right, updated :)
#
nightpool[m]
looks good!
#
fiatjaf
do I have to implement http signatures?
#
nightpool[m]
fiatjaf: http signatures are not a required part of activitypub, however, most implementations do currently require them
#
fiatjaf
is there a spec for what most implementations currently do?
#
up201705417
I think pixelfed supports the absence of signatures with a fallback to outbox verification... (dansup can you confirm this?)
#
up201705417
fiatjaf: that would actually be quite nice xD
#
nightpool[m]
http signatures avoid a "round-trip" problem, where sending a message requires a subsequent GET request to authenticate the activity
#
dansup
up201705417: nah, that was just an idea.
#
dansup
would be a DoS vector
#
nightpool[m]
^ that's also a concern with the fallback approach
#
nightpool[m]
fiatjaf: the page I linked above documents what standards people are using for authentication.
fr33domlover joined the channel
#
nightpool[m]
https://activitypub.rocks/implementation-report/ should (ideally) document what individual servers require
kaniini and fr33domlover1 joined the channel
#
cwebber2
nightpool[m]: well and the implementation report page might not even be what people are looking for
#
cwebber2
since we kind of want to have a "shop around for activitypub servers/clients" page
#
cwebber2
and the implementation report provides interesting info to standards people, but not to users
#
nightpool[m]
with respect to individual server federation requirements?
#
cwebber2
and features
#
cwebber2
why might you want pixelfed vs mastodon vs pleroma vs etc
#
cwebber2
implementation report page throws a giant table at you of spec details
#
nightpool[m]
the question was "is there a survey of what most implementations currently [require for auth&auth]"
#
nightpool[m]
that doesn't feel like a very user-level question
#
cwebber2
ah yes I suppose
#
cwebber2
sorry to derail
#
cwebber2
my mind is on this topic today so I just kind of spit it out loud
#
nightpool[m]
that's fine
#
nightpool[m]
i think a user-facing page would be cool but there are a lot of hurdles like "how does this application specifically process activitystreams data" that would be pretty hard to communicate to users.
#
nightpool[m]
Like, how do you reasonably describe the position of Peertube in the network where different servers have totally different ways of handling Video objects?
KevinMarks, rhiaro, csarven and timbl joined the channel