good morning fediverse.
-
@trwnh absolutely suspect there will/would end up being a number of "and now we need to handle this weirdness that these other folks insist on" scenarios, and some that I'd probably choose to push on and not support in more than the most cursory ways. I have a feeling that said looming pachyderm is going to have to start innovating in the space again or get left behind as folk build things that are closer to what people actually want.
-
@djsundog it’s more like “we need to do the same weirdness because they wont understand the normal stuff AT ALL”
in that regard, i guess the other question i have is whether you envision this operating at the level of the activities themselves or on textual content. one of the bigger problems of fedi is no one cares about activities, mostly just the objects themselves. both approaches are valid to some extent, but they are fundamentally different and you have to choose.
-
I cover this space in detail. I've seen the work on building plugin systems and the responses by the Mastodon team.
-
@trwnh it's not a decision I've made yet but I tend to think I'd lean towards acting on the Activities (I didn't go into it in the blog post, but I really think I want the client to actually implement AP C2S) and then catching them before delivery to nodes that are text mungers and make sure to send them the munged form that they need to make it easier for users not to think about? that's the balance beam I think I'd prefer to straddle, anyways, lol
-
@djsundog yeah i’m bringing this up because it’s been kind of a huge issue in me trying to do my own thing.
i’ve said before that fedi dev is basically “if your friends jumped off a bridge would you jump off too”. you have to do so much that is not only wrong but also unnecessary and unnecessarily complex too. that way lies madness. i’d like my main approach more webby less fedi and that’s why i put my fedi plans on hold to work on my website first.
-
@trwnh totally legit approach in my book! I'm most hopeful that if there's enough external pressure applied to the more reticent parties around more reasonable approaches to interop that eventually they'll cave rather than continue to deal with the barrage, but only time will tell ha
-
@djsundog the whole system could do with a radical simplification but the cargo cult is necessary just to do hello world
it might honestly make sense to view fedi as a separate target for crossposting, with how far it diverges from web stuff. my hope is that we can extract some network effects out of it but for the most part the network is sadly captured.
i’m thinking abt a separate server component for taking activities and stuffing them into a Create Note-shaped hole
-
@[email protected] @[email protected] @[email protected] @[email protected]
There are pros and cons to the plugin architecture... but there are orders of magnitude more pros than there are cons.
NodeBB went with a plugin system really early on. It basically democratized the extension process so that if someone wanted custom functionality, they could build a plugin without help from the devs (huge!!!). If the plugin system wasn't extensible enough, they'd contact us, and we'd add in a hook to support their use case, and that was the extent of our involvement.
The main con is that it hamstrings you a bit when you want to break things and refactor code, and the hooks start passing in the wrong things causing oodles of plugins to break. Something like 90% of third-party plugins found in the wild plain don't work with the latest version of NodeBB.
So, pros and cons, but mostly pros. If I did it all again, I'd still support plugins from the get-go.
-
@julian @laurenshof @ChrisWere @trwnh architecting for extensibility is always a win for community in my book because of the benefits you mention. I'd far rather see project leads take on the cons than becoming the bottleneck that prevents progress (a much bigger con in my book, particularly in the social software sphere)
-
@julian @djsundog decided to go open a GTS issue for discussion of plugins there! https://github.com/superseriousbusiness/gotosocial/issues/3575
-
