#dev 2023-04-29
2023-04-29 UTC
# [0x3b0b] I found 200 characters to be short enough I don't find it overwhelming, long enough it seemed adequate, and at the font and window sizes I usually view my own site it's about three lines which is around the same height a preview will take if it finds an image to put next to it. So that's my take on it as a consumer, I guess.
bret joined the channel
# [tantek] cool. roughly summarized here: https://indieweb.org/link-preview#summary_text_limits — feel free to edit/improve upon that!
# @HeeySora ↩️ well it's not a beta for nothing
no DMs, no blocks, lacking a lot of things (understandably!)
the protocol is super interesting though and has less opportunities for negativity than even mastodon/WebMention/etc (twitter.com/_/status/1652121476982530054)
bret and gerben joined the channel
# [KevinMarks] DID means "Doesn't Interoperate, Dorks"
gRegor, IWSlackGateway, oodani, [KevinMarks], [KevinMarks]1, jan61, gerben_, Kaja_, wagle_, [TMichelleMoore], [tantek], gxt__, geoffo and [snarfed] joined the channel
# [snarfed] today's the big day! Twitter API apocalypse day. nothing new has happened yet as far as we can tell though. maybe Elon forgot about it? https://twittercommunity.com/t/t-minus-6-days/191342/4
[snarfed]1 joined the channel
ludovicchabant joined the channel
# ludovicchabant hey here, I'm using mf2py to parse my website, and I see that if I write an `alt` description for an image, it gets included in the post's `name`
# ludovicchabant I assume it's because of the mf2 rules, and not a bug: http://microformats.org/wiki/microformats2-parsing#parsing_a_p-_property
# ludovicchabant but I'd like to, well, not do that and keep the `alt` description separate... is that possible?
# ludovicchabant (context: I want to use the `alt` attribute on images to use as the "description for the visually impaired" when I syndicate over to Mastodon)
# [snarfed] Hmm! alt text is a somewhere newer mf2 parsing feature. Background: https://github.com/microformats/microformats2-parsing/issues/2
# [snarfed] Based on https://github.com/microformats/microformats2-parsing/issues/2#issuecomment-392608361 , it should be separate from your h-entry's name
# ludovicchabant [snarfed]: ah right that looks indeed like what I want, thanks. It looks like it was approved a few years ago? So the problem is mf2py not having implemented it yet, or is there something else?
geoffo joined the channel
# ludovicchabant Ah thanks [snarfed] ! Didn't spot that :D
# [snarfed] lucivocchabant I filed https://github.com/microformats/mf2py/issues/182
# ludovicchabant [snarfed]: thanks!
# Soni do we have search yet?
# Soni webring-like search
# Soni being search pals with someone sounds fun
# bkil You mean creating web rings where all search queries would be limited to returning pages from member websites?
# bkil For search pals in general: https://indieweb.org/YaCy
geoffo joined the channel
# Soni you index your own website, and then you share that index with your search pals/webring. the webring doesn't have to crawl your website and you still get effective search. and the fact that it's shared means there's a way for folks to find your website through search.
# Soni (honestly fuck crawlers. waste of resources tbh. we should change how this whole thing works tbh.)
# bkil I like your idea. It would play well with my concept of a backend-optional, client-side search & indexing system.
# bkil One limitation that you need to keep in mind is that you should expect a "normal", popular thematic web ring to contain hundreds of web sites, and even if a single fetch per index was somehow sufficient, a hundred HTTP requests is pretty expensive to answer a search result, especially as you iterate on it. I'd hence probably implement some form of volunteer mirroring as well to better scale it.
geoffo joined the channel
# bkil I think YaCy is designed to query all of its peers in the federation in real time as it is implausible to replicate a whole WAN-crawler - it's more about physics than being naive. The reason it failed to gain traction was more about how it handled ranking and compositing results. It could probably be easily adjusted for this Small Web use case to always replicate the index and only query locally.