#dev 2026-10-04

2026-10-04 UTC
JefeDelTodos, vikanezrimaya, deathcomestotime, cotsuka, HobbyBuilder, ivyauth and grufwub joined the channel
#
fafau06 currently doing the webmention tests on webmention.rocks, so far i've passed almost everything (still a few left to do) but im stuck on Update Test #2 where it asked to remove a link. im using Telegraph to send webmentions and it didn't allow me to send one because "The source document does not have a link to the target URL or domain" so im wondering if this is normal and does it mean i can't pass this test?
#
fafau06 my site is using a webmention plugin for jekyll so i can probably try sending one from the cli later and see if it differs from sending one using Telegraph
#
fafau06 i did remove the entire <a> element but im assuming that's what you're supposed to do?
oodani joined the channel
#
[snarfed] yeah you probably need to send that one from the CLI, makes sense that telegraph won't do it for you
#
aaronpk heh yeah i guess feature request for telegraph, it should support sending that for a URL it previously sent before
cotsuka joined the channel
#
fafau06 oh lol so it wasn't just supported from telegraph
#
fafau06 but alright thankie! i'll try to finish the rest of the test from cli later
JefeDelTodos and cotsuka joined the channel
#
doesnm_ aaronpk did you used step ca for this? i saw method with them
#
WinnerWind [capjamesg]: saw that you deprecated Arora. If you're not using Arora, then what do you use to make your site?
#
jone42 gRegor re schema.org: i used it to mark up my sites, but now i only use it for making the linter happy (eg meta or link tags are only allowed in body with itemprop)
cotsuka and ZodiacalComet joined the channel
#
WinnerWind https://x0.at/kesjMi7EoV.png
#
WinnerWind wrote my own webmention sending script!
#
WinnerWind its coupled with my SSG though, but the main logic can be adapted to any script
#
offeringofpie dat progress bar ❤️
cotsuka joined the channel
#
aaronpk I don't know what step ca is
#
doesnm_ certificate autority server. It support provising ssh certificates with oidc
#
doesnm_ https://smallstep.com/blog/diy-single-sign-on-for-ssh/
#
doesnm_ seems you invent own thing for this?
cotsuka joined the channel; karjala left the channel
#
aaronpk Other way around. This is using an ssh key you have to log in to a website, not logging in to ssh from a web identity
#
doesnm_ oh ok
jak2k, ax562 and vriska joined the channel
#
carrvo rubenwardy, WinnerWind: if you want an example of mf2 having meaningful use in the wild, I have a few places where I use it to self-parse a webpage and graph the data.
[manton], deathcomestotime, deathcomestotim1, cotsuka, ripple34, JefeDelTodos, thatoddshade, [Tabitha], arbs09 and alex_127_0_0_1 joined the channel
#
[capjamesg] WinnerWind I use a modified version of https://codeberg.org/capjamesg/aurora-v2/, which is simpler than the original version.
#
[capjamesg] I am not really interested in maintaining an SSG for others to depend on right now so v2 isn’t well documented.
JefeDelTodos and arbs09 joined the channel
#
WinnerWind [capjamesg]: that's fair! i keep telling myself that I need to document my SSG more for my own sake but I keep adding feature after feature and documentation falls to the backburner unfortunately
JefeDelTodos joined the channel
#
[tantek] interesting discussion of consent in software and protocols from #indieweb re: ActivityPub, robots.txt etc.
#
[tantek] [snarfed] I also think there's a mismatch between a loud minority on Mastodon sites and what ActivityPub actually supports
#
[tantek] like the whole search of public posts brouhaha
#
[snarfed] yes!
#
[snarfed] (I do still think this conversation is at its core non-technical, and belongs more in #indieweb)
#
[tantek] I think there are "general interest" aspects yes
#
[tantek] like should public posts be searchable?
#
[tantek] I meant to divert the protocol-specific chat here
#
jak2k "ActivityPub federation itself is largely opt-out" Yes, but federation only what you can see and limited by follower structures. If you scrape that for some centralized search and require a hashtag in profiles for opt out, that something completely different!
#
[snarfed] yes and no. fediverse relays intentionally aggregate public posts
JefeDelTodos joined the channel
#
[snarfed] but regardless, centralized search engines are a different point. even just federating to followers, and independently loading posts (eg entering a post URL into your instance's search bar), are generally opt-out, not consent based
#
[snarfed] there are exceptions, but they're rare and non-standard
#
jak2k If you look into my AP object (<https://browser.pub/@jak2k@mastodon.de>), you will see the properties `discoverable` and `indexable`. Those are even more granular than robots.txt! That's awesome!
#
jak2k I wish the "classic web" had that! I'm totally fine with crawling my website for discovery purposes (like for webrings, directories, …), but I don't want anyone to index my writing without explicit consent on a case by case basis. LLMs completely destroyed trusts against bots. :(
#
[snarfed] yes! discoverable and indexable are good!
#
jak2k With relays, you have to differentiate between these words bots that boost without consent (like tags.pub) and real relays that are opt in: https://docs.gotosocial.org/en/stable/user_guide/relay_pushes/
#
jak2k s/word/weird/
ghorsey joined the channel
#
jak2k Typo in the typo correction 🤦 s/words/weird/
cotsuka and ghorsey joined the channel
#
jak2k I think entering a post URL into the search bar to load is not something that requires consent. Entering it into your browser's address bar doesn't require it either! It does require consent though once it's automated.
ghorsey joined the channel
#
[tantek] automation is a spectrum though, and different people have different ideas along that spectrum
#
[tantek] e.g. if I manually bookmark a URL is it ok that I have some software index its contents so I can find it again in the future by content search?
bterry joined the channel
#
[tantek] and if so how about if that software regularly checks that URL for updates, like a reader, perhaps a tags page
#
[tantek] etc.
ghorsey joined the channel
#
jak2k Yes, there is a lot unclear because allowing everything has been the default for decades.
#
jak2k I wish robots.txt was per purpose and not per bot…
#
jak2k And a more structured user agent would help too…
cotsuka and ghorsey joined the channel
#
ghorsey The other issue with robots.txt is its an honor system. It only works of the bot honors the file...
cotsuka and ghorsey joined the channel
#
jak2k Yeah, that's consent. But what would be the alternative? Requiring auth?
#
jak2k Imho, user agent strings should contain `Human` if a request is directly caused by a human and the result is for direct consumption by a human (like when I click a link in a browser that has no LLM integration).
jjuran and ghorsey joined the channel
#
[tantek] Is automatic polling by a reader considered "human" caused even if the human only took one action, subscribing?
#
ghorsey I have an SSG'd site with astro. I'm trying to think about how I can publish web mentions that I reference in my posts or for POSSE to brig.ly or other custom apis. I wanted to ask if there were any examples of doing this with astro, before I delve into deriving my own solution.
#
[tantek] There are folks with SSG sites here that send webmentions
#
ghorsey Ideally, it would be a part of the build process in my CI/CD pipeline...
#
jak2k @ta
#
jak2k [tantek]: A reader is not human but a bot. It could identify itself as `Reader`. Maybe something like `Reader{subscribers=1} FreshRSS/1.2.3 curl/1.2.3`
#
[tantek] How is a reader different from a browser?
#
[tantek] Or different from a browser restarting and reloading open tabs?
#
ghorsey Having an astro integration would be sweet
#
jak2k I'd understand a reader as something like a feed reader as something like a feed reader that fetches automatically in the background.
#
jak2k Woops, I got distracted while typing 🤦
#
jak2k My Firefox is configured to keep my more than 400 tabs unloaded when I open it 🙈
#
jak2k But even if not, that would still be a request that's made specifically because I oped *that* URL for consumption. When I press a button in my reader to fetch the full text of a page, I'd include `Human`
#
jak2k Of course the borders are somewhat fuzzy
ghorsey joined the channel
#
ghorsey jak2k: Re rss atom, I think the ssg astro has an integration to generate it...
#
jak2k Yes, it definitely has! I used it a few years ago.
#
ghorsey Yeah I haven't set it up yet.
#
ghorsey jak2k: you used it a few years ago, what do you use now?
#
ghorsey Full custom site?
ghorsey, cotsuka and vikanezrimaya joined the channel
#
jak2k I briefly used Nuxt, then used 11ty for a few years, and now I'm using a fully custom Rust server. However, I plan to make the Rust server much more modular and reusable. Many pages are currently hardcoded.
ghorsey_, barnaby, ghorsey and ivyauth joined the channel
#
ghorsey Sweet on the rust server. I write a rust based husky library to run tests and formatting checks as a part of my workflow.. I got it working just enough, then my ADHD got me distracted elsewhere and with the indieweb... https://crates.io/crates/c3-cargo-husky
barnaby joined the channel
#
ghorsey I am curious, where are you hosting your rust server? At home, a vps/somewhere?
#
ghorsey Lol sorry about grammar mistakes. I've been drinking watching football (American). *I wrote a library... *
[KevinMarks], cotsuka and barnaby joined the channel