@mariusor i don't check the forum every day, so sorry for "necro-posting" within less than a week, i guess?

is the request for me to just drop any mention of you when i am replying to things others said?

personally, you can treat everything i say as an opinion, but the difference between "opinion" and "fact" is a lot finer than most people assume. you can agree or disagree with what was said. if you agree, that might make it a fact to you. otherwise it's a matter of assertion vs objectivity

@julian @mariusor intended effect = you receive a notification in your inbox that i like your thing

side effect = you can add it to the likes collection on that thing (or you might not -- that's your concern more than it is mine)

@julian close enough, i guess? it's the stuff that happens after you POST (think ifttt, service processing, that kind of stuff)

they can be automatic or manual, often automatic in the way most people use the term but i personally allow manual processing of "side effects" too

@steve i would probably think my outbox was broken or malicious or not doing compliant^1 activitypub.

^1: depends on a conformance profile to be meaningfully defined, but "publish what i POSTed" seems to be a reasonable enough expectation in the minimal case...

@kopper @hongminhee

generally agreed except

> you have to map the expanded representation to your internal structure which will likely be modeled after the compacted one

this is compaction but manual instead of using a jsonld processor to do it. maybe the more precise argument is "don't bother with auto/native compaction"?

with that said: you also lose out on flattening and framing, which are pretty cool features for transforming the serialization. if you don't care about those, ok fine

@benpate it's close enough, though. you can imagine content models as being like:

- atom:entry
- as:Activity
- html:article

and it is possible for "the same thing" to adhere to more than 1 content model

@benpate micropub is okay; its content model is mf2-based and it acts like a cms for h-entry h-card h-event and h-cite. that could be useful for powering something that maps pretty close to 1:1 onto atom entries (and h-entry is intended to be that 1:1 mapping onto atom:entry actually!) but i haven't gotten around yet to finding that useful because i do less "blog"-like work and more "wiki"-like work.

@benpate so i guess the answer i have rn is:

- build content models that make sense and are at least internally consistent
- build content formats that capture those content models naturally and with minimal effort
- build processing models for publishing and consuming that content format as easily as possible (bearing in mind easy != simple)

@benpate if all of this more or less lives inside a sort of "container" with its own "filesystem" then it becomes far more portable, which is what i'm trying to make fall out of my work naturally. you can take your outbox "around" to whatever service provider is willing to host it. with inboxes, it can be the same thing with taking your inboxes "around" to whatever provider is willing to host them. similar to how IMAP lets you both download *and* upload messages.

@benpate using activities as an example of a content model: the outbox is a collection you can POST to and publish activities. this is ~doable as a self-contained microservice, but is there an even easier or more minimal way that requires even less overhead and maintenance? what is the sum total of all requirements needed to make it actually work?

@benpate that is what i'm trying to figure out for at least myself before writing up any recommendations for others. the hardest part for me so far is actually how to organize everything, what to name things, etc -- approaches and conventions moreso than tooling or technologies. i want it to be as easy or even easier to post to my website than to post to social media. a lot of that involves figuring out workable content models!

@benpate personally my goal is to guide people from fediverse into personal websites, not the other way around ^_^;

@ricferrer @benpate the "end goal" for me is to have them discriminate by content type so i can open pdfs in my pdf viewer and activities in my activity viewer

@ricferrer @benpate the app is a browser too but i mainly use either firefox or subway tooter

@ricferrer @benpate right now, if you link me x accounts and i find n interesting, i copy-paste n links (n <= x) into my fedi address bar (search bar) to access them in my "layer 2" browser-in-a-browser

i am not entirely convinced it would be better for every page to have to publish 2x links as opposed to publishing x links and you copying n links. 2x > x + n

what could be improved is opening n links in a different app (pwa included) but you can't avoid that unless you make your browser auth'd

@benpate how do we get people to use fedi without an account?

@julian yup

you could in theory do web+activity to mean "http get with accept header for activity+json" but what mastodon does is i think just intercept the click with js

@benpate i think 3b is where my issue is -- the solution for me looks a lot more like

1. on a web page
2. click share (in js or in browser)
3. your browser or os sends some stuff to an app of your choice you already have (including registered pwa targets)

bonus: filter share targets by content-type

@julian @benpate @rimu @ricferrer most people don't have fedi

@benpate @ricferrer @evan @julian @rimu if i'm already in my browser i don't want another browser. the "open web" IS the platform. that's where the interactions SHOULD be happening... and could be with a bit of work.