#microformats 2023-07-05
2023-07-05 UTC
[tantek] joined the channel
[aciccarello], [KevinMarks]1, [pfefferle], btrem, [jacky], IWSlackGateway, [KevinMarks], Loqi, [Murray], eitilt, [tantek], [snarfed], firepoet, vladimyr, [timothy_chambe], [marksuth], whyisitnotworkin, gRegor, [will], JKing, [Jo] and [capjamesg] joined the channel
# 
[capjamesg] Should we have a microformats-LD spec?
# [capjamesg] The rationale is that JSON-LD is easy to copy and thus easy to add, just like open graph data.
# [capjamesg] One could have a gallery of formats that you just copy then adjust to your template.
# [capjamesg] This is an established pattern across the web, in large part due to Google's JSON-LD support.
# [capjamesg] There was a long discussion about this in HWC today.
# [capjamesg] *probably in large part (I don't know the full history)
# [capjamesg] The primitives of mf2 are useful: well-defined and use case-driven. But, adding mf2 directly to HTML comes with a range of drawbacks.
# 
vladimyr Idea is to move mf2 to json-ld?
# [capjamesg] Having to fix microformats when you change HTML structure, deciding what is in and out of scope for parsing (i.e. if we parse `srcset`, what happens if someone wants another HTML attribute: do we acquiesce because it is useful. This is a real discussion that's going on in `mf2py` land.)
# [capjamesg] vladimyr I think it would be more offering another way to use mf2.
# [capjamesg] Use it in HTML directly, return an mf2 JSON file via content negotiation, or represent content via a JSON-LD object.
# vladimyr Sounds interesting, do you have supersimple example of how it might look like, hCard for instance?
# [capjamesg] The need is simplicity for developers, for both consumers and publishers.
# [capjamesg] A JSON-LD h-card could be generated from a web view without having to match things up to your markup; a consumer could easily see what to expect and codify as needed.
# 
[KevinMarks] LD is never simpler.
# [capjamesg] Say more?
# [capjamesg] (This was a long-winded discussion in HWC, so this is good to document!)
# [capjamesg] [tantek] metaformats are not for what I am looking.
# [KevinMarks] We already have 2 json forms - mf2 and jf2
# [KevinMarks] LD is inherently more complex and opaque, and ambiguous.
# [capjamesg] jf2!
# [capjamesg] That's probably better
# [capjamesg] Can I add jf2 to my web page?
# [capjamesg] *better as in closer to what I was thinking about, not inherently better by any means
# [KevinMarks] They still don't have an agreed way to normalise it, and they've been trying for 10 years.
# [capjamesg] jf2?
# [capjamesg] Yeah, JSON-LD is not immune to that either.
# [KevinMarks] As long as you add a parse test to your build
# [KevinMarks] I admit that I occasionally do a bit of a non DRY thing where I generate a JSON like structure to pass to a nunjucks/jinja2 template to make mf2 and then also pass some bits through as declared variables in a script tag, if I want to make a map from them with leaflet, say.
# [KevinMarks] But that's not the same thing.
# 
[tantek] re: "established pattern across the web, in large part due to Google's JSON-LD support" <-- established pattern of generating/publishing garbage until proven otherwise. No one else consumes it for any practical purpose than Google Search and even there you can't verify it is making any difference, you can't actually test it is having any practical impact (their validators don't even claim to make any promises about what will show up
# [capjamesg] [tantek]++
# JKing Is the Discord gateway to this place broken? Or perhaps the link on the wiki to the channel? I tried to join via said link and was greeted with an empty void.
# [capjamesg] Hm!
# [capjamesg] It should be working.
# [capjamesg] Try joining with https://discord.gg/2pHeSycn?
[manton] joined the channel