i always kinda vaguely thought micropub was okay but rereading it just now with far more experience than the last time i read it (years ago), i am realizing it's actually quite hairy and it's not immediately clear what the data model even is supposed t...
-
infinite love ⴳreplied to infinite love ⴳ last edited by
at least in activitypub the data model generally assumes that your AP server will be carrying out side effects for certain activities, so you can do CRUD and collection management using the AS2 vocab, and it's mostly coherent from a surface level. it's not terribly hard to imagine a CMS built on top of AP. but i can't make any sense of the micropub data model. the more i think about it the more confused i get. now, the api itself seems to do a lot of things generally well, but big-picture idk...
-
infinite love ⴳreplied to infinite love ⴳ last edited by
my main criticism here is that there seems to be a lot of behaviors that the server is possibly or expected to carry out, but it's never entirely clear what those behaviors are because the underlying microformats are so freeform and there's a lot of stuff that's auxiliary -- syndicating to other platforms, for example. i'm not personally interested in much if any of that stuff, so it's unclear what value implementing my own micropub server is supposed to get me.
-
infinite love ⴳreplied to infinite love ⴳ last edited by
so i'm wondering if the course of action might be to just develop an api that's even more basic (inb4 nanopub)
of course before i actually get to that point, i need to figure out a data model that makes sense first. it's possible that whatever i come up with can be mostly transformed or serialized into other existing data models but so far i'm not finding any that feel appropriate -- AS2 could be contorted to fit but i don't think it actually works well for this bc it overly relies on types...
-
Sebastian Lassereplied to infinite love ⴳ last edited by
micropub _is_ ok.
You are ignoring where these two protocols come from.As anything from https://indieweb.org/ :
micropub is self-dogfooding and it should be as easy to implement as possible with just the knowledge of html + indieweb microformats.
It should serve the implementor and some friends but not a wider LD-world.ActivityPub is the opposite …
[disclaimer] As described in the german book or the pinned Talks or recent Berlin fediday: We are writing a fully conformant CMS on top of AP which was a primary motivation for ActivityVocabulary https://www.w3.org/TR/activitystreams-vocabulary/#motivations
It is just wonderful to imagine it -
infinite love ⴳreplied to infinite love ⴳ last edited by
...and one of the things that i've come to realize while working on my website recently is that types actually do not matter, for the most part. the data model of the web is resource-based, but in most cases you don't actually care what type those resources are. it probably matters more if you're trying to build an overlay network or linked data set, but if you're not doing that, then it's kinda pointless to spend so much time boxing yourself into that additional data model.
-
infinite love ⴳreplied to Sebastian Lasse last edited by
@sl007 i get html and microformats but i see no value in micropub for my use case, because it's bringing in a lot of assumptions that i don't feel like adopting. this post was made because i initially thought micropub could fit my use case when i was looking at it years ago. in rereading the micropub spec, i must conclude that it does not actually fit my use case.
this doesn't actually have anything to do with LD or AP, which also do not fit my use case. at least, not this one.
-
@trwnh yeah micropub and microsub are both super-lacking in a lot of ways
I really dislike how micropub really wants the target resource to be named by its URL rather than by a UUID, for example
and microsub has a super restrictive view of what a "channel" is and how it should be ordered, good luck building a compatible reader that lets you do something crazy like showing unread items oldest-first
-
@fluffy i really want to collect a list of user stories and use cases that have been neglected by existing protocols and find a way to address them in whatever i end up doing.
i'm not opposed to urls as references but i think that you ideally want relative urls, and then you resolve them against a variable base (which gets you portability and location-independence as long as your data model is locked in and doesn't change)
-
infinite love ⴳreplied to infinite love ⴳ last edited by
@fluffy part of why i'm engaging in this renewed push of updating my website is because i want to hammer out the data model that makes sense for me, and hopefully it makes sense for other people too.
-
@trwnh at a higher level this is why I'm moderately pissed at Mastodon's bungling of AP objects of type different from Note.
-
@oblomov the world if every object was rendered with name/summary/content