#KevinMarkshm, google pagespeed insights on my site says I should compress the twitter avatars that are injected by webmentions more
jihaisse, friedcell and sanduhrs joined the channel
#voxpelliKevinMarks: They should probably be sent through an https image-resizer/cacher of some kind eventually :/
#voxpelliI've tried to avoid needing to cache images, but as not all are https it results in mixed content warnings if I don't, so will probably be forced to
#KartikPrabhuwas considering using Google's image proxy as an intermediate
#voxpelliKartikPrabhu: maybe one could create a standalone nickname-cache of some kind and have that handle it all? or maybe there are other images than images one need to proxy/cache?
#voxpelliI would be happy to delegate things to a third party service
#voxpelliKartikPrabhu: yeah, feels like a waste of time for eg us two to make two individual ones :)
#KartikPrabhui don't know would be a gravatar type replacement
#KartikPrabhuit could be argued that "it is a waste of time for us to make 2 different websites" too
#voxpelliwell, within webmention etc there's some more useful dogfeeding of specs etc – not sure there's any use in dogfeeding ones own image proxy thingie to become https-compliant
#KartikPrabhuvoxpelli: yeah maybe. I haven always been unsure of how much central services are in the spirit of the indieweb including Bridgy
#voxpellimy thinking is that they can help people get started before they have had time to set things up all for themselves, so lowering the threshold of starting to selfdogfeeding things
#ben_thatmustbemecweiske: nope, same feeds, just includes private posts as well, they get a webmention but they get no content from the absolute URL unless they are authed
#ben_thatmustbemegetting server to server working for that is something i would like to get worked out at some point
#kylewmaaronpk: re: the micropub source query stuff, it seems like a big shift ideologically. we can no longer pretend HTML is the API and micropub is just a publish layer on top of that. it's definitely its own protocol with its own in/out serialization format.
#aaronpkHTML is still the API for going between websites
#aaronpkbut there isn't really a way to get the real source material from the HTML to use for editing
ttepasse joined the channel
#kylewmya i get that. it sort of feels like "reality setting in" on micropub.
#aaronpkbut I suspect if you do a quick survey, the vast majority of people will say that the HTML that ends up in the published version of the post is different from what was originally authored
#[kevinmarks]Seems like the Markdown to html converter is a different service
#aaronpkinput is an image URL, output is an image URL. if the contents of the input image changes, you'll get a new output URL, and the old one will still work
#voxpelliaaronpk: [kevinmarks]: thinking whether one maybe would want to serve avatar based on h-card profile in nickname cache rather than included image, as a more "intelligent" and up to date form of caching
#voxpelliaaronpk: [kevinmarks]: and only have the image parsed from the WebMention as a fallback for when the profile itself don't have a provided image
#[kevinmarks]Right, I was thinking url to avatar service rather than email to avatar service
#tantekaaronpk: re: "aren't in the HTML, like changing the custom slug of a post" - yes my custom slugs are in the HTML! as part of the asserted permalink (uid) of the post!
#aaronpkvoxpelli: yes the author of the h-feed is the simple case. but as I learned from going through the authorship algorithm for h-entry, there are quite a few other cases to consider
#Loqireacji is an emoji reaction, the use of a single emoji character in response to a post, introduced as a feature by Slack[1] https://indiewebcamp.com/reacji
#voxpellisingpolyma: regarding the errors, thinking that one could maybe add indie-config support to Subtome as an additional/alternative backend so that it's still useful, thoughts?
#singpolymaI haven't wrapped my head around indie-config yet
#singpolymaalso, most users won't have indie-config, so it probably doesn't help them
#voxpellisingpolyma: not in the same way, it loads a custom protocol handler within an iframe and waits for that protocol handler to (optionally) do a postMessage() back with ones config
#voxpellibut utilizing localstorage rather than a custom protocol handler
#singpolymaif subtome can't do anything useful without localStorage, my preferred fallback would be to just be a link to the feed, since at least in firefox you can sub in a feed reader directly from the feed view, etc, anyway
#aaronpkgRegorLove: what are you looking for in an avatar caching service?
#aaronpki don't quite have my head around what Census will do exactly, so collecting ideas
#tantekvoxpelli: agreed that indieconfig would be a good fallback for subtome
#tantekaaronpk: Census has information about all the people based on what they self-submit(publish) :)
#voxpellicreates an issue for that to document one possible way around the issue
#gRegorLoveca3db (?) sounds pretty ideal, aaronpk. I was just going to set up something similar to that, but only for my own site, hosting the images on my site as well.
#aaronpkgRegorLove: you can map a subdomain to the s3 bucket, or send 302 redirects from a URL on your site to the s3 bucket
#gRegorLoveI'm not tooo worried about size, so not sure I'd set up s3 for storage
#aaronpki'm really curious how much data is in my s3 bucket now, especially after the home page mentions launched on bridgy and @t gets a ton of mentions
#gRegorLoveBut yeah, if I could pass a URL and get a persistent author image URL back, that'd be great
#aaronpkit's a pretty simple thing really, you could probably pretty quickly rewrite that in PHP
#gRegorLoveYeah. I had a rudimentary cache set up in Nucleus, but I migrated to ProcessWire last IWC
#sknebelwhat's the argument for caching the historic profile picture vs updating it?
#aaronpksknebel: well the particular problem i was solving was that the URLs that twitter/instagram/facebook use end up 404ing after someone changes their picture
#sknebelaaronpk: but webmention.io returns a new URL if the image changes, right? and doesn't overwrite the old one if it finds a different image
#[kevinmarks]Also, the avatar at the time may be relevant to the conversation
#aaronpksknebel: there is no concept of user profiles in webmention.io, each mention stores the image URL in the record
#[kevinmarks](Halloween or Xmas avatars for example)
#singpolymaI've worked around the subtome issue on my own site for now by opening a new tab (window.open) instead of using an iframe
#gRegorLoveKeeping one image updated for a twitter account would be tricky, given avatar changes and username changes. Would have to use the API to store Twitter user IDs
#voxpelliIf salmentioning gets going and lots of people use image caches then there will be lots of duplicates if one doesn't do an intelligent gravatar like cache rather than a simple avatar cache, but thinking about that is probably just plumbing
#aaronpknothing wrong with each person having their own copy of avatars of people they interact with
#aaronpkthat sounds way better than any sort of centralized service
#voxpelliaaronpk: the same image from 3 different people will have three different URL:s though
#gRegorLoveTrue true. so when aaronpk gets tired of maintaining Census we're not high and dry. :)
#aaronpkthat's why ca3db de-dupes on the *content* of the image, not the URL
#voxpellino need for a centralized storage though, more a matter of trying to show avatar of external h-card before that one of the embedded h-card
#voxpelliaaronpk: how can you dedupe on content if everyone applies their own compression resizing? unless you do something like pHash which sounds pretty overkill :P
#aaronpkyeah it won't work if people resize the images
#kylewmvoxpelli: maybe a good reason to use the pattern <a class="u-photo" href="original"><img src="resized"></a>
#aaronpkalthough you could still get close by comparing a really downsampled version if you wanted
#voxpellimy own avatar is 280x280 so when I comment on someones post my avatar will be a bit large unless someone resizes it
#voxpellibut 280x280 is somewhat small compared to what eg. a profile pages want – eg. Slack doesn't allow smaller avatars than 540x520 to be uploaded on their service now I believe
#Loqiidentengine is open source and software to discover and construct user-consolidated-identities from homepages, social silo profiles, based on following the rel-me microformat https://indiewebcamp.com/identengine
#calumryanNo Brighton Homebrew tonight unfortunately it seems –Â adactio has cancelled but hopefully back in 2 weeks. Following responses from my IW talk last month there’s possibly some interest in running a London Homebrew
#voxpellicweiske: the wiki note on public page upgrading suggests returning a unique private rel-self within the upgraded feed and PuSH clients will then use that url when subscribing
#voxpellithat way the access token will more or less become embedded in the URL sent to the hub
#voxpelliaaronpk: wonder if the alternative proposal should perhaps also be highlighted higher on the private webmention page so there's one page to find it all on?
#aaronpkthe private messaging page really isn't about webmention at all though
#voxpelliconsidering also that eg. "indieweb messaging" points to that one
#Loqireacji is an emoji reaction, the use of a single emoji character in response to a post, introduced as a feature by Slack[1] https://indiewebcamp.com/reacji
#snarfedthere's a decent chance it just needs the USERID_POSTID format, which occasionally 400s or 404s today (like what kylewm saw in that issue) but mostly works
#aaronpki don't get it. I search "wordpress amp" on my phone and I get the little result cards, but when I click one, it just takes me to the regular site which loads slow