Notices by small, versatile 6DoF c++ witch *ada (ada@zoner.work)

@julia@eepy.moe

@MiaWinter@tech.lgbt @navi@social.vlhl.dev @javascript@app.wafrn.net "cooking isn't an exact science, if you are looking for exact measurements do chemistry." is what my cooking club tutor said to me 20 years ago.

seriously!! look at her without her shell!!!!

RE: https://phantomthieves.net/objects/875717ef-62ed-4225-87d6-33774901d240

reminder that the patent has expired

is there even a jsonld library for swift

oh fuck it's 8 years out of date i'm going to buy some vaseline and a rope

so i have two choices, wait for @navi@social.vlhl.dev to finish liblinked, or port https://github.com/dotnetrdf/dotnetrdf/tree/main/Libraries/dotNetRdf.Core/JsonLd to Swift

@julia@eepy.moe time to fork akkoma

@julia@eepy.moe how about a GTS fork

@julia@eepy.moe i should resurrect the wytch project but bleh

developing fedi software seems like such a shitshow with being burdened by keeping up with whatever the fuck mastodon decides is good, and it appears misskey now too as they're just shitting random keywords around the AP object.

nobody respects jsonld contexts so i'd have to prefix everything which isn't bad but demotivates the shit out of me.

FEPs are my only hope but it seems more like wishful thinking with nobody actually implementing them-- so, lol.

@julia@eepy.moe AP will collapse once colliding keys start existing.

@julia@eepy.moe no no i mean this scenario:

imagine this: mastodon implements icon decoration, and then misskey for whatever reason decides to federate their icon decoration. now because mastodon has an entirely different model for this, they silently drop all misskey posts from users that have an icon decoration because it failed to parse the AP object.

@julia@eepy.moe aside: wytch was always going to use prefixes because of my fear of shit like this

but AP as a whole is a house of cards where one whoopsie in the AP object can just break federation.

@julia@eepy.moe I'd say something witty like "json-ld expansion is expensive" (which it is) but then i remembered it's literally required for jsonld-w3sec which is used for authorized fetch.

@natty@astolfo.social @julia@eepy.moe the data it's signing is the expanded jsonld

https://github.com/misskey-dev/misskey/blob/develop/packages/backend/src/core/activitypub/JsonLdService.ts#L28
https://github.com/misskey-dev/misskey/blob/develop/packages/backend/src/core/activitypub/JsonLdService.ts#L102
https://github.com/digitalbazaar/jsonld.js?tab=readme-ov-file#canonize-normalize

@natty@astolfo.social @julia@eepy.moe reminder: canonization has this in it's w3 spec

@navi@social.vlhl.dev EAC probably.

I have a feeling they're going into the direction of boot-time kernel anticheat (i.e. vanguard)

@navi@social.vlhl.dev BE will probably not do that because BE doesn't need to by design. it relies as much on server validation as it does local. it's why BE cheats are less common, also why the cost is more pricey on the developer.

in a way, EAC being as widespread as it is, because it's easy to implement gives a lot of malicious actors enough data to break the system.

some obscurity is needed for sufficient security, after all you don't go announcing your private keys.

@navi@social.vlhl.dev also as shown with vanguard, and battleye-- the goal post just gets moved to a Pi with gadget drivers + a vision ML model or a DMA + a PI

@navi@social.vlhl.dev In the end one major vulnerability in a common kAC will burst that bubble and make developers move towards server authoritative anticheat.