[eddie][cleverdevil] I’ve also been thinking about stories but have delayed because of chicken egg. I feel like stories really needs reader support but you don’t have reader support without something to follow so.... haha
[cleverdevil]Apple Clips does a great job of generating easily shareable videos that are essentially the same as you would find in a "story" elsewhere.
LoqiA story is a singular (one per profile) time stream collection post, that consists of ephemeral photo and video posts that are shown in sequence one at a time and disappear from the collection 24 hours after being added https://indieweb.org/story
[jgmac1106]That definition way too influenced by social media expectations IMO. Why just one per profile? Why must it disappear? To me a story is a collection of media arranged purposefully arranged for meaning with a start and a finish that the author or audience may control.
[jgmac1106]A story is a series of mixed media formats within one post that are purposefully arranged with a clear start and finish and a linear or non linear navigation either the author or audience controls.
[jgmac1106]tantek shifting to dev for a little as I read up on xAPI and then compared it to what chrisaldrich and idea in my summer classes it just reaffirmed to me importance of keeping metadata as small as possible, but I could see parsing feeds of my classes using, read, write, listen, reply and h-review of artifacts with webmention badges to build a cool dashboard
[jgmac1106]and then reading up on xAPI how this already maps onto AS2 and much of the work is done with social readers to create a classroom powered by the web. Just made me smile as I feel like we are all on the right path
aaronpk[jgmac1106]: ignore the name, we're just calling it that because that's what the big social media sites are calling it. The point is we are trying to figure out exactly what about the way those work is appealing and what we can draw from. Not starting with some prototypical definition of the word "story".
[cleverdevil]Was considering something in Known where you could “build” stories with a combination of uploading short video clips, pictures, etc., along with text. Re-ordering and then displaying nicely with a fancy client side JS thing.
[cleverdevil]Clips is really perfect from a UX standpoint for creating stories and they export cleanly as high quality videos that I could upload via Micropub.
[eddie][cleverdevil] I am not sure I’m ready to abandon photos, but that definitely sounds intense adding text and stuff on top. I think my ideal is a photo or video and then you can use apps (regular or Micropub) to bake anything you want into a photo or video and then upload that
[eddie]I have also though having a link and a button might be interesting if you want to link out to something else (like the Zuck.js tool in the Story brainstorming section)
aaronpkInstagram stories are a little more than just photos and videos now, since you can do things like include a link in the "swipe up" action, and also include polls or other reactions as overlays
[cleverdevil]I’ve got a pull request open on Known that I’m waiting on, but once (if?) it’s merged, I’ll start posting some Clips stories to see how I like it 😀
[jgmac1106]My story thing is gong to be like Zeega or Popcorn, but somehow spit out HTML slides, uisng 4X4 CSS Grid where you then get to stack HTML elements by inserting a URL and using z-index and/or allow you to stretch and shrink items across cells...just spits out one file with the CSS and HTML..have no idea how to build it but it will be done
petermolnarI massively streamlined my HTML output: I got rid of microformats1, in-html rdfa, microdata, per entry h-cards (it's now a single link to the main footer). On the other hand, I added json-ld and added back open graph to the meta. The HTML does feel better, but it became much larger due to json-ld.
petermolnarwhile I have serious issues with schema.org, in the end I was able to use it as a reasonable good representation of a post of mine, so I moved from the randomly named template vars over to use the json-ld data in the templates and it made things simpler in the end
ZegnatAlso, even without a single criticism section, I’d argue most of that page can be read as criticism ;) It mostly lists products that were sunsetted, shutdown, it lists some outages, etc.
jgmac1106yeah kevinmarks it is and a bit of javascript would make moving elements around easier but I could see HTML slides just as sections set to 100vh. I should just go and host my own version of Mozilla Popcorn or Zeega (like exampel I shared) that is all I want
aaronpkI went with singular vs plural based on what the wiki implied. So if that explicitly says one or more organizers of the event then I'd translate it into an array in XRay
[jgmac1106]Yes and I would suggest an organization can be an organizer as well... Though there is usually an event (Co) chair in this case that could be p-organizer
aaronpkJust cause mf2 JSON has everything as an array doesn't mean every consumer is going to do something with all values of the array, so it's important to indicate whether it should expect one or more values for each property
sknebeli.e. h-entry saying u-photo "one or more photos", but jf2 saying "photo MUST be a single string value if present", h-entry saying "embedded h-card(s)" for author but jf2 insisting on a single one, ...
aaronpkif what I'm doing goes against the microformats text it's because I found more real examples of that and maybe we should fix the microformats text
sknebelphoto MUST be a single string value if present on the top level feed object, its author property object, or any direct child entry object and MUST be a valid [URL].
aaronpkI also know that my idea of jf2 does not match Ben and Tantek's, which is what that spec represents. They weren't a fan of the spec deciding that some values are always singular or plural
aaronpkif that spec wants to delegate the vocab decisions to mf2 that's fine but mf2 then needs to be explicit about which properties support multiple values
[Rose]Different topic (sorry, hopefully a quick one), is there anything about the Micropub spec about supporting (private) drafts/has anything been done or thought of in that direction? I didn't find anything in the wiki.
sknebelaaronpk: to me the jf2 spec read as if the profile it offers wants to make those decisions for h-feeds with h-entries in them. although I now see it says "non-normative"
@jasonopus↩️ I’ve been playing around with post formats and layouts to achieve a microblogging look/feel and I’m using microformats where I can. Unfortunately, no webmentions since there’s currently no plugin for them in #craftcms. (twitter.com/_/status/1093877374062612483)
LoqiIt looks like we don't have a page for "Craftcms" yet. Would you like to create it? (Or just say "Craftcms is ____", a sentence describing the term)
ZegnatE.g. having an h-card inside h-entry that is not an author (or any other property) would nest the jf2 card object in a children property on the entry object
[kevinmarks]My pov is that deciding plurality per property is a retrograde step as then each implementation has to keep a table of properties to refer to. That's more work than always being plural and one more deref.
jgmac1106so is it more work to signify each property as either a singular or plural when so few things in this world if they are countable can exist as one? (I will then bow out of convo as philosophy probably doesn’t help in the least)
[eddie]Yeah as an iOS developer of a reader app. Swift needs things to me well defined. It would be difficult to not know which jf2 should be singular and which should be plural
[eddie]“That’s more work than always being plural” that sounds like mf2, so then why would jf2 exist? Although I’m not sure of the conclusion to that kevinmarks “one more deref?”
aaronpk[eddie]: let me ask this differently. would you rather have to pull the first value from the "published" array to get the timestamp, or always have "published" be a string?
ZegnatAlso: do you expect readers to show the first or the last timestamp from a published array? Clearly users would expect the same timestamp in all their readers, so implementers have to come to some consistent answer to that question
[kevinmarks]If you're in Swift you are likely declaring a struct with a Date in and writing some marshal/unmarshal code to go to and from iso8601, so yes you're making the n-ary decision.
aaronpkyou're either going to 1) bake the knowledge of whether things are single or plural into your code every time you access a value (when the data you're dealing with is always an array), or 2) bake the knowledge into some library that your code depends on and then your code can use the values as they appear
aaronpkmy hope was that returning the simplified data is helpful to app developers, as it was certainly helpful in writing my own apps. so it's good to hear that it has also helped others writing in other languages.
sknebelsure, but that's somewhat of a strength of going of mf2 json: you don't have a crashing dserializer if you suddenly start changing the number of property values
sknebel(just to be clear, I'm not totally arguing against having it, some examples are very safely one-valued (rsvp), but moving the details out of the clients doesn't mean changes don't affect the clients)
[eddie]absolutely, changes affect the clients. But by moving the details out of the clients it makes the boilerplate code the clients have to write easier
[jgmac1106], [jeremycherfas], [tmiller], [Coty_Beasley] and [schmarty] joined the channel
[schmarty]aaronpk: oops finally looked at why some posts were appearing but my site's feed pages weren't updating. hugo doesn't like when "tags" contains nested objects and i forgot to handle that in my micropub endpoint.
[schmarty]aaronpk: in all my research i think the most-touted way to get a still frame from a video on an iOS device is to watch the video, pause it where you want, and take a screenshot 😂
[schmarty][Rose] good call! i have a shortcut that turns Live Photo GIFs back into videos for looping and posts them via micropub. i could grab one of the GIF frames for a featured (poster) image.
[schmarty]i was able to get micropub edits working with a shortcut yesterday. micropub query, then extracting and sending the content over to Drafts (where i edit and copy content to clipboard), then return to shortcut to package up a micropub edit and send it.
aaronpkso I realized that most people's main complaint about OpenID is that user IDs are URLs and users are expected to enter their URL in sign-in forms. the pushback I always get is that people don't think of their accounts as URLs, they think of email address instead.
aaronpkand for simple domains, "aaronpk.com" is actually easier to enter than an email address, but the reality is, and likely always will be, that most people will not have simple URLs like that
aaronpkand I totally agree that expecting users to enter crap like google.com/userlksdjflkj/2398572983571235 or whatever google did for openid is not good
aaronpkso i'm thinking, instead of ask users to enter their full URL, ask them to enter the server their profile is hosted on. that maps to the mental model of how people interact with these things too.
aaronpkso someone would enter "gitlab.org" into the quill sign-in box, and gitlab.org has an authorization_endpoint defined, so quill knows where to send people to sign in,
aaronpkand if you're not signed in, you sign in as normal however that service does it. maybe email address, maybe username, maybe in the future webauthn
aaronpkbut this way, I have the advantage that this login flow actually requires *less work* for users compared to email-based openid connect discovery
aaronpkfor single-user sites like aaronpk.com, it's great because I can use my real domain just the same as someone can use a shared account like mastodon.social or gitlab.org
aaronpkalso notice that i'm not proposing any changes to the indieauth spec ;-) this has all been possible for a long time, we've just never pitched it to people this way before
Zegnataaronpk, that makes a lot of sense! As long as the whole domain uses the same auth endpoint for every user, no need for the user to supply their userpage for sign in!
[tantek]!tell petermolnar if google, yandex are your jsonld/schema use-cases, then mf1 works great for them too. what (dis)advantages do you see for using mf1 vs jsonld? (e.g. page size, learning curve, DRY etc.)
[tantek]!tell petermolnar I would like to clear up the misconception that jsonld is necessary for google/yandex because it is not. I want to understand how you came to that conclusion and understand what upstream documentation we need to update / fix / create to help reduce that in the future
[jgmac1106]plus realized I needed one for featured images and one for regular images, does use the <small>e element whihc I know not everyone is a fan of
[tantek]I don't think you should wait, I think it's better to put something down with that as a starting point, and then note that you're writing up more docs
aaronpksnarfed: openid connect does have a discovery mechanism, using .well-known. and it's assumed you'll use the user's email address to find that document.
aaronpkI think original openid did work that way, but IIRC some big providers didn't do that discovery thing right so people had to type in the full url anyway
aaronpk[tantek]: that blog post is definitely along the lines of what I want to write up for people, but I want something that starts out with IndieAuth instead of setting it up from the perspective of OAuth
petermolnar[tantek]: that google vs json-ld: I know it's not necessary, and I'm basically just testing things, to see if it makes any kind of difference - so far, it's inconclusive. While the google recommendation is certainly json-ld, it is also very true that if it doesn't make any difference, why bother?
Loqipetermolnar: [tantek] left you a message 2 hours, 55 minutes ago: if google, yandex are your jsonld/schema use-cases, then mf1 works great for them too. what (dis)advantages do you see for using mf1 vs jsonld? (e.g. page size, learning curve, DRY etc.)
Loqipetermolnar: [tantek] left you a message 2 hours, 54 minutes ago: I would like to clear up the misconception that jsonld is necessary for google/yandex because it is not. I want to understand how you came to that conclusion and understand what upstream documentation we need to update / fix / create to help reduce that in the future
ZegnatReleased, the worst UI you can imagine, an ever growing table with radio buttons in them (and no accessibility labels at all): https://vanderven.se/martijn/health/weight/ - if people would like to let me know if stuff works for them that would be sweet
ZegnatI don’t do visitor stats, but I don’t imagine a lot of people even view that page, let alone need the radio buttons. So I shipped it as good-enough-for-me
aaronpkIf you had a websub feed that required authentication then you could get an accurate count. Which is kind of how mastodon can have accurate follower numbers.
aaronpkother answers: some reader services include a subscriber count in their user agent header to let you know how many people on their service subscribe to your feed
[tantek]Especially if there's some incremental, now that you added WebAuthN support, here's all you need to do to add IndieAuth support on top - type guide
[tantek]you may want to pre-seed a /webauthn page on your OAuth and IndieAuth sites with a minimal summary / paragraph just to start getting it indexed
LoqiWeb Authentication is a W3C Candidate Recommendation for an API to access public key credentials, including for a browser, optionally with the use of a hardware key https://indieweb.org/webauthn
ZegnatNot sure how indieauth.com would connect my browser+token combination to my site while also being stateless. Though I haven’t looked into that a lot yet to be honest
aaronpkZegnat: webauthn isn't a replacement for an identity, it's just an authentication mechanism, and it requires pre registration before it's useful
ZegnatThat’s what I was working on. Basically I was trying to get my IndieAuth endpoint to give a sad smiley when key not plugged in, or a big green “continue logged in” button when key plugged in.
ZegnatAnd one feature I have been wanting for the longest time is a state on my endpoint side to have a “remember me” option. So the next time I try to login to the same site (say indieweb.org) it will not even show the page and instantly log me in