As far as I understand, most (all?) fediverse #ActivityPub software does not use the Client-to-server protocol from the specs (https://www.w3.org/TR/activitypub/#client-to-server-interactions) but rather use custom APIs instead.

[email protected]

Any fediverse devs able to explain why? Is there a technical reason/limitation, or is it more about other considerations?

I'm looking for information here rather than speculation, thanks.

[email protected]

@hugh According to one dev for GoToSocial:

"Implementing the C2S API is, I'm afraid, really out of the question, as it's totally underdefined, relies on the client to do almost everything, and would be an absolute nightmare project for us. That's a non-starter."

https://codeberg.org/superseriousbusiness/gotosocial/issues/3846#issuecomment-4018586

[email protected]

@skyfaller Thanks. I guess I’m looking for more detail on what is meant by “underdefined” and “relies on the client to do almost everything” because my reading of the spec is that it relies on the server to do almost everything! I assume I’m missing something, but everything I’ve read about it is very vague.

[email protected]

@skyfaller @hugh

Also #SocialHub #ActivityPub developer forum has a bunch of C2S-related topic. You can use the forum search facility.

https://socialhub.activitypub.rocks

A very detailed investigation on what is needed client-side can be found in the #AndStatus project. It was never completed AFAIK as there were among others no server implementations to test against.

https://github.com/andstatus/andstatus/issues/499

[email protected]

@[email protected] investigation into the use of C2S for cross-instance posting is one of our goals for 2025 with support from @[email protected]

But yes, there's most likely a reason why it hasn't been adopted widely...

[email protected]

@hugh ActivityPub is very badly defined, I wrote an article about it (about s2s, but all of that applies to c2s as well): https://chrastecky.dev/technology/activity-pub-the-good-the-bad-and-the-ugly

Additionally, it's a very chatty api with a lot of http requests, that's fine for s2s, but for c2s it makes the app much slower.

[email protected]

@hugh What bugs me about "underdefined" is it seems to me someone should just go define it.

[email protected]

@skyfaller Well one person’s “under-defined” is another person’s “flexible and simple”. If people get their heads out of micro-blogging it becomes clearer why a more rigid definition becomes limiting, IMO.

[email protected]

@hugh In addition to the concerns expressed here, I argue in my book that Mastodon's status as the single largest ActivityPub implementer back in 2017 is a factor -- after all, Mastodon had an API at that point. People making clients chose to make them compatible with Mastodon's API instead of redoing everything for the then-new C2S spec.

[email protected]

@julian @hugh We will forward your credits to @nlnet, which supports #opensourcesoftware. Although the name is similar, we are a different foundation that develops open source software.

[email protected]

@[email protected] oh! Oops unfortunate name collision.

[email protected]

@[email protected] some mastodon dev discussion re: c2s implementation, a while back:
https://github.com/mastodon/mastodon/issues/10520

"The ActivityPub C2S spec is incredibly barebones. No notifications (as separate from home feed -- it's all mixed together in "inbox"), no search, no autocomplete, no domain blocking, no muting (as opposed to blocking), etc etc. You'd end up defining so much custom vocabulary and endpoints that you might as well just use the Mastodon REST API."

- gargron

"The activitypub c2s api is cool but it comes from a wildly different perspective, and it would be a lot of work to write clients that support that, with no real clear benefit to the mastodon user experience. Furthermore, it would be incredibly hard for the mastodon code base as it exists today to support a good implementation of C2S and the mastodon API side-by-side—practically, it would mean basically re-writing the mastodon server from the ground-up. And even once you've done all of that work, there aren't any existing C2S clients out there that provide a comparable user experience to the mastodon front-end."

- nightpool

[email protected]

@[email protected] (no comment from me as to the accuracy of those perspectives, just figured worth sharing as stated reasoning from one project's devs at the time)

[email protected]

@chrastecky @hugh

Thanks for writing this article! I tooted about it:

https://social.coop/@smallcircles/114419670153108301

And it was good opportunity to follow-up on two #SocialHub threads relating to your examples in "The Bad":

https://socialhub.activitypub.rocks/t/exposing-edit-history-via-activitystreams/2930

https://socialhub.activitypub.rocks/t/update-note-quirk/4545/14

https://socialhub.activitypub.rocks/t/distinguish-between-posts-and-direct-messages/2283

[email protected]

@cascode @hugh gargron's perspectives are still accurate and gives a good (partial) list of the many underdefined aspects of C2S. Related discussion at SocialHub: https://socialhub.activitypub.rocks/t/nextgen-activitypub-social-api/4733/4

[email protected]

@[email protected] A few do, Pleroma historically supported it. Not sure if they still do.

Part of the problem is that it's kind of unwieldy to work with. C2S apparently requires developers to handle all logic on the client side, rather than the server. I've also heard that there's vagueness in the spec on how to do certain things.

The Mastodon API ended up being a lot easier to work with, and overtook the client ecosystem. It actually ended up becoming so popular that most Fediverse clients and platforms have adopted it. It's also one of the factors as to why so many Fediverse platforms are full-stack servers, rather than clients.

I personally still think that C2S has potential, it just requires capable hands to implement.

[email protected]

@rwg @hugh Pragmatically speaking, that's true because people needed to build/show something. But there are gaps in the specs such that broad interop is technically not possible as it stands ( https://w3c.social/@csarven/114178482096355457 ). I'm not saying this to dismiss any work (especially one I was involved in since the days of Laconica/StatusNet). I'm coming from the perspective of how different classes of products can interoperate, regardless of the umbrella or spec "camp" they're part of on the web platform.

[email protected]

@smallcircles
> what is needed client-side can be found in the AndStatus project ... there were among others no server implementations to test against

Seems like Pleroma had it working before mid-2020;

https://pleroma.social/announcements/2020/05/10/pleroma-security-release-2-0-4/

The Epicyon server has support for AP C2S too, so that could also be used to test apps trying to implement it;

https://libreserver.org/epicyon/

Was any of this mentioned in the SH thread?

#ActivityPub #APC2S

@skyfaller @hugh
@bob

[email protected]

@strypey @smallcircles @skyfaller @hugh @bob https://rdf-pub.org is providing c2s. I started Testung with #andstatus but there where open questions regarding oauth if i remember right.

[email protected]

@strypey @skyfaller @hugh @bob

I did not mention a #SocialHub thread. There are multiple discussions where various aspects were discussed, that might still be useful. The search facility is best way to find them.

As for AndStatus the github issue lists their step-by-step progress in investigating what was needed, and what the challenges were. One of them was unavailibility of appropriate server back-ends to test against, mentioned *at the time* as challenge.

Would ️ more #ActivityPub C2S dev.

Darkscribes Community

As far as I understand, most (all?) fediverse #ActivityPub software does not use the Client-to-server protocol from the specs (https://www.w3.org/TR/activitypub/#client-to-server-interactions) but rather use custom APIs instead.