Ancestors

Written by Avery (She/Her) on 2024-12-29 at 21:35

so what i'm gathering is that the ap standard is more of a suggestion then actual rules

=> More informations about this toot | More toots from shitpostalotl@axfedi.derg.rest

Written by Julia :v_bi: on 2024-12-29 at 21:39

@shitpostalotl@axfedi.derg.rest yeah that sums it up

it tries to define this crazy fucking abstract model that is impossible to implement in practice, so actual implementations just implement parts of ActivtyStreams, usually with some extensions tacked on

=> More informations about this toot | More toots from julia@eepy.moe

Written by Avery (She/Her) on 2024-12-29 at 21:40

@julia

=> More informations about this toot | More toots from shitpostalotl@axfedi.derg.rest

Written by Julia :v_bi: on 2024-12-29 at 21:40

@shitpostalotl@axfedi.derg.rest w3c

=> More informations about this toot | More toots from julia@eepy.moe

Written by Avery (She/Her) on 2024-12-29 at 21:41

@julia nah like why is it impossible to implement?

=> More informations about this toot | More toots from shitpostalotl@axfedi.derg.rest

Written by Julia :v_bi: on 2024-12-29 at 21:43

@shitpostalotl@axfedi.derg.rest because it defines at least six different compact (standard JSON) shapes for most types of interactions, under the assumption that implementations will expand the JSON-LD into an RDF graph and then query it. Also every object has no guaranteed shape. Everything is technically allowed to be renamed, and the only property you can count on is @context.

=> More informations about this toot | More toots from julia@eepy.moe

Written by Julia :v_bi: on 2024-12-29 at 21:45

@shitpostalotl@axfedi.derg.rest so for example, instead of having a content property on the compacted JSON-LD, you're allowed to rename that property to literally anything, just by defining an alias in the context. And, per spec, implementations have to accept this. In practice? they do not.

=> More informations about this toot | More toots from julia@eepy.moe

Written by Avery (She/Her) on 2024-12-29 at 21:50

@julia so what i'm getting from this is that they are using a fucked up format and not enough things are actually defined definitively so a bunch of stuff is platform-specific implementation details?

=> More informations about this toot | More toots from shitpostalotl@axfedi.derg.rest

Written by Julia :v_bi: on 2024-12-29 at 21:51

@shitpostalotl@axfedi.derg.rest yes. implementing AP is largely looking at what is supported by the other implementations and building according to that

=> More informations about this toot | More toots from julia@eepy.moe

Written by Julia :v_bi: on 2024-12-29 at 21:52

@shitpostalotl@axfedi.derg.rest in most cases, mastodon

=> More informations about this toot | More toots from julia@eepy.moe

Written by Avery (She/Her) on 2024-12-29 at 21:53

@julia let me guess: most people implement the same subset of the spec as masto and there is no agreement about how to add more things in addition to that?

=> More informations about this toot | More toots from shitpostalotl@axfedi.derg.rest

Written by Julia :v_bi: on 2024-12-29 at 21:53

@shitpostalotl@axfedi.derg.rest prettyyyy much

=> More informations about this toot | More toots from julia@eepy.moe

Written by Avery (She/Her) on 2024-12-29 at 21:58

@julia that's fucked. please tell me that most softwares use a sane internal representation for all this?

=> More informations about this toot | More toots from shitpostalotl@axfedi.derg.rest

Written by Julia :v_bi: on 2024-12-29 at 22:00

@shitpostalotl@axfedi.derg.rest most? *oma just stores the JSONLD in the database as JSONB with a few keys for caches but that's a special case

• Key and mastodon have reasonable DB schemas and whatnot but yeah

=> More informations about this toot | More toots from julia@eepy.moe

Written by Avery (She/Her) on 2024-12-29 at 22:08

@julia the software family known for db rot does something nonstandard with the db. that makes sense i guess.

=> More informations about this toot | More toots from shitpostalotl@axfedi.derg.rest

Written by Avery (She/Her) on 2024-12-29 at 22:09

@julia is JSONB just uniquely shit software??

=> More informations about this toot | More toots from shitpostalotl@axfedi.derg.rest

Written by Julia :v_bi: on 2024-12-29 at 22:10

@shitpostalotl@axfedi.derg.rest JSONB is just a binary representation of JSON- the issue here is that they're just plopping the JSON into the database directly instead of translating it into something reasonable

=> More informations about this toot | More toots from julia@eepy.moe

Written by Avery (She/Her) on 2024-12-29 at 22:11

@julia like as a string??? or the json fields being the db columns???

=> More informations about this toot | More toots from shitpostalotl@axfedi.derg.rest

Written by Julia :v_bi: on 2024-12-29 at 22:14

@shitpostalotl@axfedi.derg.rest as a binary representation of the JSON data, but pretty much the same as a string

It has some fields from the JSON data copied into the columns to make it more performant but it's not great

=> More informations about this toot | More toots from julia@eepy.moe

Toot

Written by Avery (She/Her) on 2024-12-29 at 22:17

@julia i've decided that i dont want to know enough about db design to understand what this means

=> More informations about this toot | More toots from shitpostalotl@axfedi.derg.rest

Descendants

Written by Julia :v_bi: on 2024-12-29 at 22:17

@shitpostalotl@axfedi.derg.rest valid

=> More informations about this toot | More toots from julia@eepy.moe