Gregory's Server@nasser yeah as far as I can tell it is just a series of conventions that when you encounter a property named X, you should do Y with it
|
13 comments
 @nasser @darius the context stuff embeds typing information about what the properties refer to, so the processor can tell what's an IRI or a string literal or some other typed object etc. and it's exactly like XML schemas except we can mostly pretend that RDF never had anything to do with XML and it was all just a bad dream. i do kind of wonder if fedi clients will handle fully baroque json-ld or are only pretending though.  @jleedev yeah I guess I'm curious as to what the "processor" is here. the implementations I've seen have all been just throwing JSON together as you would normally or parsing it normally with hard-coded rules. I haven't seen an implementation where the context gets consulted, but I'm also not well versed in this technology. @darius if this is annoying LMK and I will take you off these mentions   @nasser @jleedev @darius right; I'm actually kinda curious about the difference between "all the stuff JSON-LD is capable of as a standard" vs "the way JSON-LD gets used by ActivityPub and its implementations" I get the feeling that the latter is a lot more just "a document with hyperlinks" and the ontology stuff doesn't really come into play in a serious way, but maybe I've just been looking in the wrong places @technomancy @nasser @jleedev I've never encountered it as anything but "json with hyperlinks" in an AP context @darius @technomancy @jleedev and, crucially, the differentiation between strings that are hyperlinks to be followed and strings that are just strings is captured directly in the software, as opposed to consulting some kind of schema in a meaningful way @nasser @darius @technomancy @jleedev so are we saying that AP could have just used straight JSON and lost essentially nothing? @technomancy @darius that's neat. I suppose it provides some mechanism to differentiate between hyperlink references to be followed and random strings?  @technomancy @darius @nasser a way of creating and reusing schema and then applying them in a way that reduces to triplet subject predicate object links. so instead of just having an autonomous JSON object it can reference terms from controlled vocabularies and ideally be made interoperable. v powerful way of describing things but also of making compound concepts, actions, etc. some examples and info here: https://json-ld.org/ @technomancy @darius @nasser not necessarily separate - it’s more that JSON-LD can represent a graph of things linked by relations (which can be internal to a document or point outside the document) and also associate the relations with an ontology somewhere @technomancy @nasser @darius think of being able to build composable graphs. there's 1:1 transformation with specs like RDF |
@darius it feels vaguely like XML schemas? which are also URLs and I guess namespaces are nice but, fundamentally this stuff is all just conventions no?