@acute_aura@weh_kaniini I'll be completely honest, I gave up on AP and W3C huge committee seemingly designed to be implemented as big java blob standard, I like to point to json-ld normalization algo for how not to design for adoption Doesn't help AS is somewhat attached to tootsuite at the hip either. (twitter.com/_/status/1366892654252982275)
@kevinmarks↩️ Sorry, I know kaniini, but I don't think we've met before. There was a split in approach in the social web working group that that document explains to some extent. Webmention is from the non JSON-LD side of the split. (twitter.com/_/status/1366904359502626817)
@weh_kaniini↩️ but microformats is just a different form of semantic web craziness. parsing HTML or XML is expensive by comparison to just using JSON everywhere. it really is the right way to do it, and the AP spec is far more elegant than WebMention and friends. (twitter.com/_/status/1366905785310183429)
aaronpkgargron's tutorial on setting up activitypub was the best i've seen, it's still quite a bit of work and some fiddly cryptography bits that you need to find the right libraries for to get the message signing right
jackybut like the fact that you can't even insure the same _sorting_ when used threw me off (I remember this being an issue with JSON-LD signatures for some time - I think they have an approach now but it dealt with n+1 parsing of the object recursively)
aaronpkyou know what, now that i think about it, i'm pretty sure JSON-LD was the absolute wrong choice for AP and they should have stuck with JWTs instead
@kevinmarks↩️ Indieweb uses JSON where it makes sense - in Micropub and Microsub for example; Webmention was designed to work with any web resource, not just JSON ones.
Microformats define a way to get consistent JSON out of HTML web pages, and plenty of tooling to do it (twitter.com/_/status/1366909146369581056)
[KevinMarks]JSON dictionaries don't round trip unless you add sorting rules, yes, as different languages do different things with key order -php and js are the order you create them in, python in hash order, Go deliberately randomised
aaronpki'm just thinking that AP would have been a lot easier if they were pushing JWS blobs around instead of a combination of JSON-LD plus HTTP message signing
aaronpkthere's like 3 different versions, one of which is marked "do not use", mastodon went and added stuff to the draft they found when they picked it up. I happen to be talking with some people who are continuing another fork of it at the IETF which should hopefully pick up some steam again soon
[KevinMarks]as you say, some of that is inevitable as when you sign anything you end up with key management entailed. I wish we could get hashing more unified too
[tantek], petermolnar, [snarfed], benwerd, shoesNsocks and [fluffy] joined the channel
[fluffy]Okay so I’ve just added the whoosh full-text index to Publ and it works amazingly well. But now I’m trying to figure out, from an API standpoint, if I should expose the search result metadata, or if it should just be a list of entries. Does anyone have any strong opinions? I don’t really see the value in exposing the actual metadata except for extreme nerdiness, like I don’t see the user story for letting people know the match score or