#microformats 2026-09-24
2026-09-24 UTC
Jon, [Leo] and thatoddshade joined the channel
[aciccarello] joined the channel
# 
[aciccarello] For images there's usually the u-featured or u-image on the <img> element. And similar options for links.
# [aciccarello] Or if you have embedded images in your content those could potentially be rendered within e-content
[aciccarello] joined the channel
# [aciccarello] Yeah, I use u-photo in a photo post https://www.ciccarello.me/posts/2026/08/26/supergirl-surf/ but u-featured on a blog post https://www.ciccarello.me/blog/2026/07/20/2026-birdathon/
# 
Loqi [preview] [Anthony Ciccarello] I braved the heat and went and watched the Supergirl Surf Pro Competition again over the weekend. There was some really impressive surfing on both sides of the pier. Iām partial to the flow of the longboarding but of course the short board moves ar... https://www.ciccarello.me/assets/img/2026-supergirl-above-header.jpg [video]
thatoddshade, [schmarty], barnaby and vriska joined the channel
barnaby, thatoddshade and solonovamax joined the channel
# [aciccarello] I need to check if there's a representative hcard parser in the js microformats library
# [aciccarello] Seems really useful.
# 
solonovamax according to https://microformats.org/wiki/representative-h-card-parsing, there exists https://github.com/rockorager/representative-h-card, however this url is dead. I can find https://www.npmjs.com/package/representative-h-card, so you might be able to recover the source code, but the repo is gone
[snarfed] joined the channel
[snarfed] joined the channel
# solonovamax while testing a few of the hosted microformats parsing libraries on https://microformats.io on my website, I noticed a few funny things about what it extracted from the html. because I'm using tailwindcss, some of the html elements have classes like h-full on them, and they all seem to extract those. is there any way to tell them to not do that?
# solonovamax also in my head tag I have some honeypot stuff for LLM scrapers, like a rel=alternate to a .md version of the page or a rel=describedby which points to an llms.txt, both of which will just server a large compression bomb (like an ~1mb file that is 100gb when decompressed with brotli). I'd ideally like anyone trying to parse microformats for my page to never have to risk running into those. is there any way to exclude them? they're not inside
# solonovamax hmm, thunderbird just cut off part of my message lol
# solonovamax they're not inside of an h-card or anything
# 
jkingweb There was some discussion about what to do about Tailwind noise: https://github.com/microformats/microformats2-parsing/issues/59
# solonovamax the tailwind one does feel like the smaller of the two issues though bc I assume that anyone parsing microformats will only extract things with known specs (h-card, h-event, etc.). I'm more worried about someone accidentally hitting the honeypot links tbh
# solonovamax gRegor: maybe something like a microformats=ignore property which microformats parsers would be aware of, but LLM scrapers (hopefully) won't? dunno
# solonovamax those two in specific have to be like that, because I was just following the https://llmstxt.org/ spec (well, I could also always use the Link header for them ig)
# solonovamax I'm reading through all their funny little specs to provide alternative versions of pages for LLM crawlers that are "easier for them to ingest" so I can just serve them large compression bombs created with just dd if=/dev/zero lol
thatoddshade and plantroon joined the channel