#[Joe_Masilotti]If there’s any indie web-specific feature requests throw them at me!
KartikPrabhu and [Emma_Humphries] joined the channel
#[Emma_Humphries]I'd love to add a step to a Jekyll build to call out to download these images.
[schmarty] joined the channel
#[Joe_Masilotti]If you’re using them for og:image you don’t even need to. You can point it directly to Mugshot Bot’s servers and the image will be generated on the fly the first time the post is shared. There’s details in “Option 3” but it requires a paid subscription.
#aaronpki think the indieweb angle there was not relying on an external service at runtime, but using it to generate and then store the image
deltab, nickodd, samcmullin, justHaunted, [tantek], KartikPrabhu, gRegorLove, [jeremycherfas], globbot, [capjamesg], jamietanna and [KevinMarks] joined the channel; justHaunted and nickodd left the channel
#[KevinMarks]indieweb features would be to parse for microformats and use those to make a richer preview
#[Joe_Masilotti][Zegnat] that sounds very similar to what I had before I made mugshot bot! I’m not using pupeteer (it’s very hard to scale) but something similar to screenshot the images.
#ZegnatIt is mostly because it feels bad to tell someone “this is not a bug, nor can we help you with it, your webserver is just misconfigured”. And then when you tell that to people on a managed hosting plan that basically means you are telling them they picked the wrong hosting company? Just bad vibes all around.
jjuran joined the channel
#Zegnat[Joe_Masilotti]: yeah, they definitely imagined running it on your own side or as part of the build pipeline, to get around scaling issues.
#ZegnatIt was just the first time I think I saw the idea of generating dynamic preview images from content really laid out in a blog post :)
#GWGaaronpk: It's to support their new application passwords
zack1 and [schmarty] joined the channel
#[schmarty]zack1: the short answer is that i can't think of any microformats2 consumers that work on a client-assembled DOM
#[schmarty]it's definitely possible to do so, but would require automating a real browser environment or running a headless browser to allow the JavaScript app to build the page content, then run a microformats2 parser over the result.
#[schmarty]i am not aware of anyone that has actually used a setup like that because ultimately it works against the ideas of /microformats as an extension of HTML
#[schmarty]and i find the definition on the wiki a bit snarky but agree in principle that something on the web isn't machine-readable without /curlability
#zack1I agree with the curlability sentiment; I'm not sure why the project went with a client side solution
#aaronpkYeah tbh if you can't convince the project to make a server rendered version (then upgrade the interactions with js if needed) then I'd just put a hidden blob of html in it
#aaronpkgood JS apps are even finally doing the first render server side to deliver actual html now
#aaronpkof course they have funny names for it like hydrating and such but at least they're thinking about it now
#[schmarty]i'm guessing but i feel like it comes from framasoft's history of in-browser javascript projects (e.g. webtorrent, then PeerTube), i also feel like a newer AP-driven fediverse projects have an "APIs and apps" sensibility.
#[schmarty]the way they frame mobilizon also gives me the sense of "we're reinventing everything", and it doesn't surprise me that "building with modern JS frameworks" would be a default tehre.
#zack1yea a lot of people on the fediverse are against browsers and web pages/apps so the developers there tend to prefer to build an API others can build apps for
#[schmarty]ooh as an aside this question put me in mind of a tweet i saw from voxpelli this morning about parsing pages and i'm enjoying this article about how much of a mess something as "simple" as link previews can be for security. https://www.mysk.blog/2020/10/25/link-previews/
#Loqi[Mysk] Link Previews: How a Simple Feature Can Have Privacy and Security Risks
#[KevinMarks]I did something similar for svgshare.com, where the png previews are made by a headless browser from the svg's but there are easier ways to do that now
#sebbuaaronpk, talking about js, i wanted to use the (internal, not documented) api of a site, so i browsed and used the site normally while using the network analyzer... so there's REST endpoints, which returns HTML content
#sebbu(because i can write regex and i'ld be too lazy to load the dom and do xpath instead)
[tw2113_Slack_] joined the channel
#petermolnarisn't the wordpress block editor doing something similarly nasty, but they even came up with their own way of doing so, inside html comments?