#social 2018-08-08

2018-08-08 UTC
kaniini joined the channel
#
cwebber2
<kaniini_> why should i believe that i will be treated any differently than evanp or mike macgirvin
#
cwebber2
evanp was co-chair of the socialwg
#
cwebber2
and most of activitypub's design (as was ostatus') was his
#
cwebber2
also evanp became an advocate and defender of not having webfinger in activitypub fwiw.
kaniini, ajordan and kaniini_ joined the channel
#
tuxether[m]
out of curiosity, what would a replacement for json-ld be?
#
tuxether[m]
or was the point that extending AS2 via JSON-LD isn't necessary at all?
#
dansup
yes, I dont think its necessary for S2S
#
dansup
pixelfed will not do anything with LD, its just plain json for mastodon & pleroma support.
#
dansup
I see the value of LDS for C2S but that should be optional
#
dansup
ActivityPub is a great protocol, and I appreciate the work everyone has put in. I think we can come to a consensus that encourages a more simple protocol for new implementations
vasilakisfil, kaniini, xmpp-social and vasilakisfil_ joined the channel
#
JasonRobinson[m]
> <kaniini_> why should i believe that i will be treated any differently than evanp or mike macgirvin
#
JasonRobinson[m]
not this stuff again, sigh...
#
JasonRobinson[m]
litepub??
#
saranix
yeah that was a waste of a few hours
#
JasonRobinson[m]
> ActivityPub is a great protocol,
#
JasonRobinson[m]
AP is more of a framework than a protocol, for good and bad
#
JasonRobinson[m]
would be great if someone made a clear spec on top of AP that works for specific use cases and is clearly documented including signing, delivery etc. then people could expand it to whatever they need but could easily create apps that interop
#
kaniini
that is precisely what litepub intends to be
#
JasonRobinson[m]
kaniini: is it something that has been started or is it just an idea?
#
JasonRobinson[m]
is interested
#
kaniini
yes, it is started. we are having meeting this saturday to set specific timelines and goals for MVP.
#
kaniini
litepub is not meant to be competitive to activitypub, but instead to bring more rigidity to the s2s side
#
JasonRobinson[m]
where can I follow the work or participate?
#
kaniini
#litepub on freenode
#
JasonRobinson[m]
👍
#
kaniini
#freenode_#litepub:matrix.org
#
kaniini
our basic plan is to create a rigid profile of activitypub for real-world s2s applications, that is easy to implement and minimalist. no JSON-LD, etc.
#
JasonRobinson[m]
makes sense to me
tantek joined the channel
#
csarven
Can conversations here outside of the office hours be trademarked? I'm particularly interested in this bit: "if you follow it, everything will just work".
#
kaniini
csarven well, that is the goal of litepub
#
kaniini
it is meant to be a rigidly defined profile of AP that if an implementor follows, there will be no missing pieces
#
kaniini
hopefully, the AP spec will be able to refer to litepub in the future as a profile that is useful for implementors
#
csarven
All the power to you - and no sarcasm intended. I think you've made your position fairly clear. I'm not sure if this channel can offer you anything further; many people have already dedicated plenty of time to hear you out and find a way to collaborate.
#
kaniini
i'm open to collaborating, but i believe a separate working group is the only way forward. i feel strongly that JSON-LD advocates would disrupt any attempt to defang JSON-LD from the spec.
#
kaniini
and, to be clear, my issue isn't even JSON-LD itself, it's @context
#
kaniini
we intend retain extensibility support in a way that JSON-LD parsers would be able to leverage effectively
#
kaniini
just, without @context :)
#
tantek
kaniini - there is no longer any working group in this space. The Social Web WG closed earlier this year: https://www.w3.org/wiki/Socialwg
#
tantek
that may be part of the confusion, e.g. when you say "a separate working group is the only way forward"
cwebber2 and puckipedia joined the channel
#
tuxether[m]
> and the two specifications may interoperably combined.
#
tuxether[m]
is that grammatically correct?
#
pantherse
kaniini, what exactly is your point of brining litepub up with this group? You were offered a chance to voice your opinions through the proper forum; but, you shot it down. Then you make statements like "whoever were invited to draft the initial spec clearly lacked real world experience engineering these things"
#
pantherse
You say you want to collaborate; yet, you take on a combative tone. I know I'm coming across as a jerk; but, I'm trying to figure out what's going on here exactly.
cwebber2 joined the channel
#
kaniini
pantherse i believe there are some in the audience who agree that activitypub needs to become a more rigid standard, wrt the S2S protocol. what i am not interested in is hearing about why activitypub / activitystreams need to be semantic, use JSON-LD, etcetera.
#
kaniini
pantherse i do not believe that the W3C mechanisms in terms of SocialCG can deliver a rigid specification that is usable for implementors to create robust and secure AP software. some proposed opening a bug to strip all JSON-LD discussion from the specification, but this is deflection: everyone in the room knows that the bug / pull-request will absolutely be rejected, because the proposed solution to the problem "but i need semantics in my AP message
#
kaniini
parser" would be "no you don't, because you have to handle the side effects regardless"
#
pantherse
Then why keep beating this issue over the brow? I'm not trying to pick a fight as I'm still learning about ActivityPub; but, I want to understand the hostility coming from you as someone weighing my options.
#
kaniini
i'm not beating this horse anymore, i have called it dead and moved on, hince litepub.
#
kaniini
there are areas where collaboration are possible, in terms of defining future additions to activitystreams vocabulary, for example
#
kaniini
and, it is possible that litepub itself could become the basis of an activitypub 1.1 specification in the future. who knows.
#
kaniini
but i do not believe the SocialCG can incubate a rigid specification. every attempt to discuss steps about bringing rigidity have been met with deflection or fantasy hypothetical scenarios that will never actually happen where rigidity may be harmful.
#
kaniini
i also strongly believe rigidity is the only way to deliver a protocol that will truly deliver on the promise of universal interop in the real world
#
kaniini
many projects are presently stalling on federation because figuring out what paths to take is too hard, due to activitypub lacking rigidity.
#
saranix
pantherse: "as someone weighing my options." mind telling us what you're building?
#
saranix
kaniini: "many projects are presently stalling on federation because figuring out what paths to take is too hard" -> this has been going on for like a decade and predates activitypub by far
#
saranix
kaniini: but I see what you're saying, that activitypub has failed to fix the situation
#
kaniini
no
#
kaniini
pixelfed, for example, exists because activitypub promised universal federation
#
kaniini
but it's stalled, because it turns out that the lack of rigidity makes federation support unattractive to code
#
kaniini
i can point to every single project on the fediverse besides Plume in the past 6 months. they all stall when it comes to federation.
#
saranix
right, but you're ignoring all the projects that stalled because the only "federation" was openid/pingback/rss/etc. -- it's an old problem
#
kaniini
yes, and that's more specs that lack rigidity
#
kaniini
ActivityPub *is* an improvement over OStatus, for example
#
tantek
in particular, cross-implementation federation is hard
#
kaniini
it is hard, but the solution is rigidity
#
tantek
kaniini if I understand you correctly, by rigidity, you mean stricter conformance requirements
#
tantek
e.g. fewer "optional" aspects
#
kaniini
i mean separation of optional aspects from the core protocol
#
kaniini
and "there is only one way to do it" for the core protocol
#
saranix
I have a counterargument to rigidity. That's how we end up with something that can only microblog, or can't understand nomadic identity
#
kaniini
saranix that's why specs move.
#
tantek
I think conformance requirements are different from extensibility though
#
kaniini
saranix and litepub is extensible in the exact same way as activitypub
#
kaniini
you can do json-ld without @context, so we can present litepub extensions as activitypub extensions, without @context
#
kaniini
but yet, normal JSON consumers can clearly understand them too.
#
pantherse
saranix, I'm working with a group that wishes to do an ActivityPub-based system along the lines of the old USENET system. The primary target audience are open source software developers; so, we don't always have to rely on commercially-owned collaboration platforms like github or slack.
#
tantek
I have a read a similar critique of "optional JSON-LD" by Henri Sivonen: https://hsivonen.fi/no-json-ns/ - I'm curious how much this overlaps or is different from the current criticism.
#
pantherse
Our current path is to implement using ActivityPub, with options to expand later.
#
kaniini
pantherse well, the point of litepub is to produce specs that can be used by actual developers to create something which will work
#
kaniini
in other words, litepub provides the answers that activitypub doesn't
#
kaniini
is the plan
#
kaniini
it also restricts JSON-LD to the point where it is only used for extensions, and only in a format where it really is just normal JSON if you want it to be
#
tuxether[m]
JSON-LD is sort of difficult to implement in a statically typed language like Haskell.
#
kaniini
correct, hince why litepub tames it :)
#
saranix
pantherse: what group is this? do you have anything to look at yet (even user requirements list)?
#
kaniini
anyway have to run for now. bye. litepub is not meant to be combative, but instead a specification which builds ontop of activitypub and gives implementors the actual answers they need to work with the software already out there.
#
pantherse
saranix: Here's the project page https://newguard.icei.org/hathi-project/
#
pantherse
I still need to get in touch with cwebber2.