#[snarfed][tantek] I vaguely remember a page you wrote or contributed to about standards body best practices and guidelines, personality types/trolls who are attracted to them and cause problems and how to deal with them, etc. any idea where that is?
#[snarfed]haven't found it by searching pbworks or our wiki or google yet
#[tantek]snarfed, I used to have the top result for Troll Taxonomy as well, though I think my page on it has aged out to the 2nd page of G SERPs: https://tantek.com/w/TrollTaxonomy
#@tmcwcan a blog be an activitypub server, so i don't have to do some rss->mastodon zapier hook or whatever, you could just follow a website directly in mastodon? (twitter.com/_/status/1593799110280232961)
#@tomcritchlow@vgr Ha yeah strong brand alignment 🤣 (Though don't tell anyone that I dislike about 50% of the current indie web implementation - activitypub might be an evolutionary dead end) (twitter.com/_/status/1593837739862282242)
slyduda1, mro, barnaby, jamietanna, gRegor and claudia1 joined the channel
#[schmarty]haha! i like that php-di docs start immediately with autowiring 😁
#barnabynot used league/container but it looks like it does pretty much the same stuff
mro joined the channel
#barnabyphp-di has a nice slim integration where it’ll inject dependencies into closure request handlers, so you get IOC dependency injection without having to make classes for absolutely everything
#barnabyyep! I like this approach of having every handler start out as a closure, and only “upgrading” groups of related handlers to a class when there’s a clear need for it
#aaronpkoops, i need to not send both a webmention and activitypub post when i reply to someone
#[schmarty]lol i did a double-take to make sure this was still actively developer after hitting this quote on their homepage: "[takes] advantages of all the latest features of PHP" (it then lists out 3 features each from php 5.4, 5.5, 5.6)
#[schmarty]does your site need to sniff that the post you're replying to is activitypub or not? 🤔
#barnabywho got a duplicate response? receivers deduping responses themselves is the more robust method, as in theory anyone can send a webmention for any page
#aaronpkit happened to someone here the other day, and just happened to me on my test wordpress site
#[schmarty]"the mentioned post accepts webmentions and is an ActivityPub object" is a weird case for sure!
#barnabybetter to file an issue with whatever plugins are implementing the responses to get them to dedupe on their end
#aaronpki could not send the activitypub post if i've successfully sent a webmention, but that means i have to totally change how my AP replies are sent
#barnabyideally webmention vs AP should both end up with the receiver knowing that there is a response to their target at some source URL
#aaronpkbut yeah they should be able to dedupe on the other end since the URL of the reply is the same from both
#barnabyso they can just compare the URL and see “yeah I’ve seen this already”
#aaronpki guess that's a feature request for the wordpress activitypub plugin then
#barnabyI’d say so, yep. No idea how closely related it is to the webmention plugin, or how they could communicate between them, but hopefully there’s a solution
#barnabyyep I just saw simonw post their article about AP. good to see more implementations, hopefully this can nudge AP away from being “whatever mastodon decides to do”
#barnabyat least for all the under-specified details
#aaronpkShould we restart meetings at the W3C social CG to try to actually improve the standard and interop?
#barnabyif there are people motivated to do so? I don’t really feel like working on AP myself
mro joined the channel
#barnabyas I have no experience implementing it and hope to keep it that way with BF for as long as possible
#aaronpkoof now i remember why i shortcut this by converting to a webmention, there's a lot of bits and pieces to pull out
[iambismark] joined the channel
#[snarfed]The Multiple Domains Per Server section on that post is entertaining 😆
#aaronpkah crap i have to handle photo responses too
#barnabyconverting incoming AP messages to something resembling what XRay outputs?
#aaronpkactually skipping that step and inserting into my responses DB directly
mro joined the channel
#[schmarty]gRegor: found a maybe issue on indiebookclub.biz. when i sign in it tries to show a profile photo but the src URL is `Array` : `<img src="Array" alt="profile photo for" style="height: 1em;">`
#barnaby[snarfed]: is it possible with BF to have a single page listing people you follow, and sending BF a webmention whenever you update it? I’d much rather use a blogroll-type approach for managing BF follows than have a channel full of individual follow-of posts for each one
#barnabybecause if there’s no precedent or standard approach for fetching “fediverse posts” from their URLs, then how would a “fediverse post viewer” work?
#Loqicontacts refers to a set of people, organizations, groups and anything else with one or more addresses (often URLs) for communication https://indieweb.org/contacts
#[snarfed]sknebel hmm. if I click the star icon on a post on https://mastodon.art/explore when logged out, I get a pop-up that still lets me type in my instance
#barnabyanyone who (semi-?)automatically manages a list of contacts on their site (say, from response contexts and incoming responses): how do you manage contact deduping and aliasing? do you do it at all, or treat, say waterpigs.co.uk and twitter.com/barnabywalters as two different identities?
#aaronpkbarnaby: i have contact "posts" and i can add multiple URLs for each contact
#aaronpki don't update automatically, it's all manual
#aaronpki can also add alternate nicknames, so when i type @barnaby in a post, it links to your site. and when that's syndiated to twitter it changes it to your twitter username
#barnabyokay, got it. I’m currently trying to design my contact data model based on the ~1000 people I’ve interacted with in one way or another in the history of my website
#barnabyso I think either I drop most of that data into a separate moderation rule set (which I need anyway) and start fresh with a manually-managed contacts list
#barnabyor try to come up with some way of semi-automatically merging identities
#aaronpk[snarfed]: there's no "type in my instance" on that screen 🤔
#[snarfed]I'd love to see the discussion around that evolution of the remote follow/interact UI
geoffo, slyduda1 and gRegor joined the channel
#[KevinMarks]That was changed in 4.0 so you can probably find the discussion in github
[manton] joined the channel
#[manton]I should read that discussion because I was surprised by the 4.0 UI change here… First of all not prompting for the instance, but second of all having “Copy” use a URL instead of the username. I think it’s confusing for people to have to mix and match both usernames and URLs.
#[manton]I actually had to update Micro.blog this week to allow pasting in a Mastodon URL instead of just the username, so they both work.
#barnabyin the case we were discussing (remote interactions) it’s a link to a post which is being copied, not something which can be represented as an @-@ ID
#gRegor[schmarty], Does your indieauth return info for profile scope? Otherwise it's falling back to representative h-card, maybe there's multiple photos and I just need to use the first one.
#gRegorah, maybe I'm not handling the new mf2 photo value/alt structure!
#barnabymf-cleaner has functions for that now, in case you use it
#gRegorYeah, looks like I use it, so probably something else
#barnabyif you were using getPlaintext to get photo properties before then the new version will handle img-alt just fine. If you were requiring ^0.1 you’ll need to change to ^0.2 though
#IWDiscordRelay<jacky#7226> I'm thinking that I'd push people to _only_ set up their sites using metadata endpoints with Sele instead of the three (or two, excluding TicketAuth)
#IWDiscordRelay<jacky#7226> I don't know for sure how many clients support it but I figure a means of nudging support would be by making it an option
#IWDiscordRelay<jacky#7226> aaronpk: If you're down to restart it, I'd gleefully join
#IWDiscordRelay<jacky#7226> I'd love to be more involved (the other committee I was a part of dissolved but it was nice)
#gRegormetadata endpoint on by default and an option to turn on the others for backcompat could be nice.
#gRegorWith ibc, it's now nudging people if they don't have metadata endpoint to update, but will still work without it
#IWDiscordRelay<jacky#7226> barnaby: I keep the scheme to avoid any assumptions when fetching info over HTTP or HTTPS
#IWDiscordRelay<jacky#7226> and I have a skeleton of a 'resource resolver' that leans on the scheme to know _how_ to get info (leaving it open to do things like resolving dweb-esque URLs and read into things like Tor)
#barnabyah, interesting. in my case I’m always keeping the true URLs themselves in the raw data, but would be using the normalized versions for querying
#barnabyso I’d never fetch anything based on the normalized URL, they’re only for internal use
#barnabyand would mean that, say a http site sent me a webmention, then upgraded to https and updated their response, sending another webmention. I wouldn’t end up with duplicates and the update would be handled correctly(?)
#barnabyat least, correctly from a user-facing PoV
#barnabyI can’t think of a better method to avoid that sort of duplication without constantly re-fetching all sorts of stuff and checking the effective URL
#barnabyI suppose another method of normalizing URLs *internally for queries* without dropping scheme would be to normalize https to http, and assume that any site serving over https serves redirects over http, which might not apply the other way around
#barnabythat way there’s still at least a scheme, for differentiating content served other other schemes
#barnabybut I don’t currently see any need for the latter
#IWDiscordRelay<jacky#7226> I agree, in that case, it wouldn't be much of a need
#barnabyand as the normalized URLs are only in the DB for querying, not the raw data, I can regenerate the DB whenever I want if I need to change it
#barnabyokay, so I’ll go with this for the moment and see if I run into any problems
#Loqijacky has 32 karma in this channel over the last year (83 in all channels)
geoffo joined the channel
#barnabyand I’m already using just the path segment of URLs to refer to local content internally for querying, so heavily normalizing URLs for external content isn’t so weird
#IWDiscordRelay<jacky#7226> I've also cleaned up the dashboard a bit (and started playing with some simple styling)
#IWDiscordRelay<jacky#7226> That page is def a js;dr one (as well as the request consent screen). Thinking about refactoring parts of it into Web Components and what not
#gRegorWith webmentions, I use paths only for my URLs so I can accept a wm for http / https of the same URL. I redirect http to https, though I wouldn't assume that for incoming wms, so I use those URLs as-is
#gRegorI also create an md5 hash of the source+target URL and use that indexed column instead of source or target. I haven't looked at the implementation in quite a while, but it's been working pretty smoothly
#barnabygRegor: I also used to store each individual source+target mention together, but am switching to a model where I have one record for each external piece of content, which can be a response to more than one of my URLs. Mostly to deal with the fact that a response can also be a mention, and I don’t want to end up with two records in that case
#barnabyI’m also using the same table for reply/like/repost contexts, embeds, and potentially even an integrated feed reader if I can make it work
#barnabyto simplify everything. Say someone replies to one of my notes, mentioning me, and then I like their reply. Previously, the external piece of content would be stored three times in my site: once as a reply, once as a homepage mention and once as a like context
#barnabyand then potentially even more times in a reader
#barnabyso I’m trying to reduce that down to exactly one copy, which I can easily keep track of, update, version etc
#[tantek]aaronpk, could you upload that screenshot of the div pop-up thing saying "Favourite Mastodon • Art ... " etc. ? There's some very interesting UX copy there, as well as the suggested flow, and the suggested copy/paste option that seems very similar to the citation interface on the bottom of my posts