#barnabythere are 46 examples of /contact in the indiemap data too, although those are likely split between additional profile info and contact forms, and I don’t feel like reviewing all of them manually
#Loqibarnaby has 28 karma in this channel over the last year (45 in all channels)
#barnabyso IMO the real-world usage of /about indicates that rel=about would be a familiar way of marking up links to expanded h-card pages, but we can come up with a solid proposal for exactly how the same/similar-domain/prefix matching rel-me would work, I think that would be preferable, in order to avoid adding an additional rel value
#barnabyit looks like all usage of those rels in the dataset is from one domain!
lanodan joined the channel
#[snarfed]barnaby bigquery browser URLs aren't shareable like that, you'll want to save the query as public and then get its link with the Share button
#[Jamie_Tanna]Doesn't seem to happen for my other followers
mro and [MiaCrow] joined the channel
#jonnybarnesLoqi posts stuff here from Twitter based on hashtags right? Wonder if we could do something similar for Mastodon?
jonnybarnes and barnaby joined the channel
#aaronpkthere's no global mastodon search for obvious reasons
#aaronpkbut we _could_ make an indieweb account that follows a bunch of people and then looks for keywords in their posts. not quite the same thing tho
#sknebelalso very much a cultural thing, anything scraping-like is *not* welcome
#aaronpktho an "indieweb.org" account that follows you wouldn't be scraping, and someone could block it if they want, so might be acceptable?
gRegorLove_ joined the channel
#sknebela few days ago someone had the bright idea of "oh hey its possible to build a a fediverse search engine, lets do that". iteration one got shut down after 7 hours
#aaronpkyeah i dunno if it's actually worth the trouble
#sknebelrelaunched it now with some more considerations made, curious to see how that goes
#barnabyI wonder if the “fediverse” will end up suffering from it’s apparent culture of not really wanting to be on the web
#sknebel(main thing was that now they respect robots-tags, which at least mastodon has explicit UI to configure for your profile)
#barnabyI’ve seen some fairly prolific people on twitter criticising mastodon’s lack of search functionality, and would have pointed out that you don’t need a fediverse-specific search as posts are just web pages, but it looks like that’s less and less the case, by design
#sknebelso I dont think there is a very strong effort to clamp down on stuff
#aaronpkrandom question, i'm getting a *lot* of stuff in my activitypub inbox that's in reply to people i follow but from people who I do not follow. why/how is that ending up at my server??
#kandr3s[m]Mastodon generates an RSS feed for any hashtag by adding ".rss" to the URL - Couldn't Loqi subscribe to #indieweb in a few selected instances?
#aaronpkas said before, there are technically ways to do it, but whether it's actually a good idea is the question
#aaronpkoh that's not even just posts from that one server, it's posts that server has seen from any server 😮
#barnabyyeah, it could work for the moment to just hook Loqi up to that feed and/or the same from indieweb.social and assume that that’ll pick up the majority of indieweb-relevant stuff
#aaronpkstill feels weird that when i click "unfollow" it sends a message to that person
#aaronpki feel like you should be able to silently unfollow someone. so more like how websub works really. where you have to continually re-confirm your subscription, and you can just not renew it if you want at some point
#barnabythat’d make follower counts a little harder to maintain, if you care about such things
#sknebeleither the system has a concept of "a follows b, and this state is known" or it can promise "you cant know if someone unfollows you"
#sknebeli.e. twitter didnt tell you, but of course there were apps to track it
#sknebel(with blocks it's arguably worse, and there the "solution" is a social one: if your instance is known to notify users about blocks (vs just enforcing them, which users can of course discover) it'll be removed from the "polite" parts of the fediverse)
#aaronpkfrom a websub-like perspective (assuming each user has to fetch your feed with some kind of authentication rather than the all-public hub-centric model) the "follower count" is the number of users who have fetched your feed within the time interval that you require they re-subscribe
#aaronpkwebsub requires renewing your subscription every X seconds as defined by the hub, so if someone just stopped renewing the subscription you would just stop counting them as a subscriber
#aaronpkrandom thing i just realized, this push-vs-poll thing is also coming up in the SCIM spec. SCIM is about syncing lists of users and groups between different software (like your HR system syncing the list of employees to Slack to auto-create users and such). there's a debate about whether this should happen via polling for changes or a push-based delivery mechanism
#Loqigiving credit is a collection of cultural practices related to acknowledging and attributing text, hyperlinks, quotes, utterances to others, typically by name, as a way of recognizing their contribution(s) https://indieweb.org/giving-credit
#barnabythe first idea could potentially be expanded to automatically detecting phishing attempts from client_ids using, say, punycode or subdomains for imitation. look through all known client_ids and see if the new client_id looks like it’s trying to imitate one
#barnabyor just show a warning whenever a client_id containing ascii-imitating punycode is encountered
#LoqiIt looks like we don't have a page for "silently unfollow someone" yet. Would you like to create it? (Or just say "silently unfollow someone is ____", a sentence describing the term)
#barnabyI know what muting is and I suspect aaronpk does too. the discussion was about feeling weird that unfollowing someone in activitypub-land involves actively sending their server a message saying “I unfollowed you”
#[tantek]1I'm saying that the user-level feature already exists by another name, regardless of whatever protocol does whatever
#@cagrimmettHonestly, Mastodon is not the answer. It is a crappier Twitter and I don't expect it to get better. Instead, I think I should spend my time reading more indie blogs and liking/responding via webmentions. I know this isn't a great setup for everyone, but I feel it is right for me. (twitter.com/_/status/1590005736461209600)
#barnaby^ huh cagrummett works on wordpress at automattic
#[tw2113_Slack_]“I don’t have a newsletter, subscribe to my RSS feed.” hahaha
#[tw2113_Slack_]i don’t expect any mix of mastodons to replace twitter, but it’s also not trying to do that themselves. Everyone just thinks that because the ui/behavior is similar.
#aaronpki wasn't saying the user feature didn't have a name...
#barnabyyeah we were discussing protocol-level details and their emotional consequences. in activitypub-land there is a practical difference between unfollowing someone and muting them
#aaronpkthat's also why i posted it in #indieweb-dev and not #indieweb :)
#barnaby[tantek]1: re you noting AP/mastodon’s regression vs atom/rss wrt fetching past content when you follow someone, here’s a gh issue discussing it, in case you’re interested/want to reference it somewhere https://github.com/mastodon/mastodon/issues/1547
#Loqi[valentinp72] #1547 Previous toots not loading
#barnabywith the reasoning that “that’s just how federation works”… hmm
#aaronpki feel like loading the most recent X posts would go a long way to not looking so broken
#[tantek]1"that's just the way (insert plumbing / jargon) works" is never an excuse for poor UX. fix the plumbing
#@stevehunt4hiopHere is another surprise. If you follow someone on another instance, you will not get access to their previous posts, unless it so happens that they were already being federated to your instance. Your instance will start to collect their new activity but will never go back... (twitter.com/_/status/1589913026832236545)
#[tantek]1barnaby, the context was getting inbox deliveries from accounts you DID NOT follow, and thus reasoning about "unfollow" is the wrong approach, because technically that should be a no-op in a protocol where you're not already following them, hence my point that /mute is the answer there
#aaronpki think those were two different discussions?
#aaronpki was specfically commenting on the fact that in order to unfollow someone i was previously following, i have to actually send a message to them that says i want to unfollow them
#aaronpkyes, i could write something that tells my server to drop their messages on the floor so i can silently ignore them, but that's not really a scalable solution
#@stevehunt4hiopHere is another surprise. If you follow someone on another instance, you will not get access to their previous posts, unless it so happens that they were already being federated to your instance. Your instance will start to collect their new activity but will never go back... (twitter.com/_/status/1589913026832236545)
#[tantek]1I suppose "newsletters" are like AP in that respect, where when you subscribe to a newsletter, it's not like you're emailed the most recent N newsletters, you only get new messages
#LoqiA permalink is a URL which typically represents and retrieves a single post (also explicitly called a post permalink) https://indieweb.org/Permalink,
#[tw2113_Slack_]my oldest/longest is still michaelbox.net
#@kn_wler↩️ Please invest in accessibility from the get go. Don’t make it an afterthought.
Also, consider making it an open platform that supports Webmentions. (twitter.com/_/status/1590040335195213824)
#[Murray]1My longest-owned domain name has never been used 😂 😭
#[KevinMarks]1The hashtag search on Mastodon is across the full federated feed on the instance, so it's not just looking in the people on that instance but the union of their follows, so searching the big instance will get a fairly good approximation to the whole thing.
[benatwork] joined the channel
#[benatwork]just realized I've owned a domain for 21 years that I don't use at all. an english word and everything. I should put something there
#[KevinMarks]1Also, per the chatty mastodon protocol thing, I think it caches images locally by default too, so you can end up with a lot of traffic and storage use if someone follows an image heavy feed
#t0nic@barnaby your cert has expired on waterpigs.co.uk
#barnabyt0nic: thanks, fixing it now! my webhost finally started offering unlimited free LE certs, so I was waiting for my manually-updated ones to expire before changing over
AramZS joined the channel
#barnabyI had a whole PHP script which would automatically update my old LE certs and then send me an encrypted email with the new cert and key in, but I still had to manually copy that into the web admin UI for ~10 domains
#barnabyas managed lifetime LE certs used to cost 15€ per *sub*domain, which was too steep for me, knowing that they cost nothing
#barnabyI kept on pestering the host to offer unlimited managed certs on premium plans, so I’m glad they finally did!
#barnabyit was that change which finally prompted me to put my affiliate link for my hosting company on my homepage as part of my data disclosure. lack of unlimited free TLS certs was the only thing I was really dissatisfied with from them so far
#[KevinMarks]1I'm doing the quarterly manual update of LE certs for a client site at the moment (it's only visible to chosen IP addresses, so certbot can't really check it, so manual DNS challenges it is)
#[schmarty]1automatic DNS challenges are now part of the certbot tool which is exciting.
#barnaby[snarfed]: the access_token parameters are required for all granary twitter requests, right, even for public tweets? are they long-lived? can I safely put them in a config file and reuse them for many requests without re-authorizing with twitter?
#[schmarty]1(i use a hacked-up solution with acme.sh + plugins to do automated DNS challenges for letsencrypt on some non-public servers for work and am excited to switch back to certbot)
#barnabywhen I remove the access token parameters from the query in the sandbox UI it still runs the query just fine, but when I open the generated query URL it complains about not having access token parameters
#[snarfed]barnaby yes, token etc are required for twitter. not sure about their lifetime, I can look
#barnabyokay, for the moment I’ll store the ones I have now, and see what happens
#barnabynow that twitter rejected my API access request I’m relying on granary.io for getting reply contexts
#[snarfed]I don't see that sandbox UI behavior though, if I remove the token and secret and click GET, it 403s
[benatwork] joined the channel
#barnabyhuh, interesting. for me it works fine, and the generated GET URL doesn’t have those parameters either, so it doesn’t look like they’re being added silently by the backend
#rubenwardyThis tool doesn't appear to find my syndicated copies, I suspect it's because I have rel="nofollow syndication" rather than just rel="syndication"
#barnabyrubenwardy: I think the h-entry tester is looking for the syndication properties on the h-entry (from links with u-syndication classname) it found rather than rel-syndication links, as it’s for testing the h-entry markup and the rel links are a slightly different thing
#barnabyit would be a good idea to add rel-syndication checking to the tester though, and surface those links, suggesting to add u-syndication to them if not found on the h-entry itself. I’ll raise an issue for that
#barnabyah I see, they’re already marked up with u-syndication, jsut not child elements of the h-entry. That’d be another good feature for indiewebify.me actually, looking for microformats classnames outside of the root and prompting the user to move them (or the root classname)
#barnabyrubenwardy++ for the useful usability testing data ;)
#Loqirubenwardy has 1 karma in this channel over the last year (2 in all channels)
#barnabylooking over indiewebify.me makes me want to work on it again, but I really gotta get the new version of my site working before I continue procrastinating on community projects
#[KevinMarks]1wekk, if you use it to test your new site…
#barnaby[KevinMarks]1: the stuff it tests is all simple markup that I can do correctly in my sleep! I mostly want to make it better and more useful for other people
#sebbui'll need to check if i need to update my copy of selfauth and php-mintoken
#barnabygRegor++ good to see wider support for indieauth-metadata discovery!
#LoqigRegor has 16 karma in this channel over the last year (71 in all channels)
#gRegorAye. indiebookclub is going to show a notice if your site doesn't have the metadata endpoint, emphasizing the app still works, but might want to update your IndieAuth or contact the developer
#barnabygreat, definitely a good plan to gently nudge people towards updating their indieauth implementations
#gRegorProbably fairly easy to wrap that in the indieauth metadata endpoint. A separate PHP file could set up an array of the metadata and output it as JSON. Then add a <link rel="indieauth-metadata"> to that PHP file.
#barnabydiscovery is currently out of scope for taproot/indieauth, but I added the metadata endpoint to the examples and the example app, so if people base their implementations off that it’ll be up to date
#[Jamie_Tanna]barnaby I believe jacky has been doing some interesting things around "x has previously used this app" and other things in the indieauth consent screen
#barnabyoh cool, good to know, I’ll ask them when I next see them
#LoqiFortress is an IndieAuth provider and consumer service that allows one to have drop-in support to use their site as an identity on the Web https://indieweb.org/Fortress
#[jacky]ah yeah so that has _some_ looks of a consent screen
#[jacky]I think I had images before but I can upload some in the near future!
#barnaby[jacky] “Per-site app usage reporting, Reviewing apps used, Shareable as a h-feed to help promote apps” on /Sele sounds very interesting and similar to the ideas I drafted here https://indieweb.org/consent_screen#Ideas, would be interested in your thoughts on them, and screenshots of existing UI examples if you’ve already implemented something similar
#barnabyI think there are plenty of interesting ways we can combat the potential for phishing attacks which comes with using public oauth client apps
#sebbu[Jamie_Tanna]++, the indieauth endpoint could keep track of all successful and failed attempt at login (which service, how many times, the date/time of the last 10 attempts)
#[jacky]Indeed. A lot of the ideas of the features I have right now are still in development but I think I can do some screenshots and videos of it by running such cases
#[jacky]I've been using https://playwright.dev/ to write out tests that replicate behaviors I'd want to confirm (like making sure that it finds endpoints, can be setup, etc)
barnaby and geoffo joined the channel
#barnabyaaronpk: when you get the chance, could you merge https://github.com/aaronpk/XRay/pull/113? should be very quick to review, and I feel a bit bad leaving XRay dev-main with an easily fixable bug, even if it’s not likely to affect anyone yet
#Loqi[barnabywalters] #113 Used default options for fetching as well as parsing