#[kevinmarks][manton] - a like becoming a post, insofar as it notfies others and can be seen, is a common pattern on silos - Google Reader, Facebook and Twitter have all done this in effect
#[kevinmarks]the trickier problem is combining the POSSE toggle with a like.
KartikPrabhu joined the channel
#[eddie][kevinmarks] I think manton’s concern is more along the lines of a person’s blog becoming filled with “Liked X page” without them realizing that would happen. Because if the blog doesn’t support a like post, and does fallback behavior, that is what they would get.
#[manton]On the like-of fallback, I think it's great that it exists, but a fallback is by definition not the best user experience. That's where I think discovering what a server supports can help.
#tantek[manton]: I disagree with the "by definition" assumption
#tantekin this case, the fallback is actually based on analyzing "like" activity text from existing social media
#tantekand while "existing social media" may not be "the best user experience", it is nonetheless "good enough" to bother worrying about in practice when building a replacement
#tantekthat is, parity in that aspect is sufficient, because there are plenty of advantages in other areas
#tantekand in this specific case, perhaps "fallback" is not the best framing, as it is more like "plain text version", or as a more concrete example, "tweet equivalent", i.e. what would be posted to Twitter if you checked the "[x] crosspost to Twitter" button in whatever app/silo you were using to post.
#tantek(evidence: nearly every social site/app has a "auto-post to Twitter" option for every thing you do, and such auto-posts naturally post some plain text equivalent to Twitter)
#tantekwhich I take as evidence that the UX of a providing a plain text equivalent is good enough, since people are actually *choosing* to do so
#[manton]Yeah, "text version" is maybe a more accurate description of what is happening. Although I think "fallback" also describes it. Anyway, what we call it doesn't really matter too much... My concern is just what the user sees and whether it fits their expectations.
#tantekThere are enough other places where a "text version" is "expected" (e.g. RSS) that it's been reasonably normalized.
#[manton]Without debating whether Micro.blog should have public likes, which is a separate discussion, the fact is it doesn't have them right now. So let's say someone is using Micro.blog for years and then downloads Indigenous and all of a sudden a new "like" button appears! It's unexpected. I don't think the user will know what it will do.
#tantekI think a better way to view likes etc as "upgrading" people's UX
#tantekah that's a different but related question for sure
#tantekwhether a client chooses to show buttons for different "kinds" of posts, not necessarily knowing how well a server "supports" them
#tantekthe assumption is that all servers at least support plain text notes, like Twitter
#aaronpkis happy this discussion is grounded in UX
#[manton]I see... Yeah, all servers should absolutely support plain text notes. But I view "like-of" and "bookmark-of" (and others) as special cases. Not just a simple plain text note.
#tantekso if people's expectations are that "posting to micro.blog" (from whatever client) are similar to "posting to Twitter", then plain text expectations are already there
#tantek[manton] my point above about cross-posting from other services is that they are *not* special cases in practice, with users, with existing services
#tantekthat is, people do things like cross-post from YouTube when they like a video
#tantekthere's likely more research/documentation we can do there (specific examples from specific bookmarking services cross-posting what text to Twitter)
#tantekbut it's that research that should inform our opinions on this rather than abstractly viewing different kinds as special
#tantek(e.g. internet scales reporting people's weighings are plain text, even though a "weight post" is arguably a very special case)
#tantekSwarm checkins, sleep reports, etc. the list goes on. every special "kind" of post has plain text equivalents figured out that people are using actively on Twitter
[mrkrndvs] joined the channel
#tantekand that they're *ok with*, most importantly, thus expectations are ok
#tantek(in other news, I'm at the point where I have to read the Bridgy Publish to GitHub instructions)
#[manton]Weight posts and RSVPs and things like that are much closer to real posts, though. Likes and bookmarks are more like actions you take on a post.
#Loqiresponses, or interactions, in the context of the indieweb, refer to all the different ways and things people explicitly do to and with others’s posts, from written replies to quick likes, in other words responses = replies + reactions https://indieweb.org/response
#tantekthough RSVPs *are* actions you take on a post.
#tantekthough an RSVP+comment can be more substantial
#[manton]Somewhat related, the way reacji's are handled makes a lot of sense to me.
#tantekperhaps a pull request has so much additional structure that that will require far more markup than just a /reply so we don't have to worry about distinguishing them for now?
[tantek], snarfed, bengo and renem_ joined the channel
#aaronpkconsiders making a repo for micropub extensions in the indieweb org in order to have a place to discuss them better
#ludovicchabanthey there! so after adding basic micropub support to my blog engine, I'm moving on to webmention... I find that when I receive a webmention, besides the checks required/recommended by the spec, I need to also do a few other things like scrub the source page to figure out who mentioned me, and a summary/text blurb of their post, so I can display it like a comment under my own post... are there libs or other resources I can use
#ludovicchabanteven if there's no lib that does the work for me, any resource that gives best practices would be useful -- I find myself scratching my head at the various p-author/h-card/u-photo things I can find at the source URL, and what to do with them
snarfed and chrisaldrich joined the channel
#tantekludovicchabant: you can start with using post type discovery to figure out what kind of webmention it is
#ludovicchabantinteresting -- the webmention spec doesn't specify where the hyperlink should be on the source page, so it's possible a page mentions one of my posts, but that mention lives outside of their `h-entry`
#ludovicchabantthis is the case for webmention.io test mentions for instance
#KartikPrabhuludovicchabant: webmention is pretty agnostic about microformats so there could just be a simple link to you site. The microformats only comes in when you want to process the webmention to get data
snarfed and tantek joined the channel
#tantekludovicchabant: and indirectly you can get mentions from various content silos / social media sites - specifically if you POSSE to them, you can get responses to those POSSE copies
#tantekI'm working on the first part of that with GitHub, POSSEing to GitHub issues
#Loqi[Tantek Çelik] The “Authorize Bridgy” prompt page has a big green button that says to “Authorize snarfed” which seems a bit incongrous.It would be better for branding/trust reasons if this prompt for Bridgy could use the github.com/bridgy organization, and ... https://indieweb.org/images/1/1d/2018-02-09-github-bridgy-auth-snarfed.png
#tantekand speaking of different prompt, that's exactly where I'm currently stuck in my UI
#tantekwhich doesn't look any different that if I was going to POSSE to Twitter or GitHub, so despite by backend possibly doing the right thing, now I *really* have to update the front-end to communicate *where* a post will be POSSEd to
#tantekI knew there was going to be extra work with this particular POSSE destination
[tantek], snarfed and [mrkrndvs] joined the channel
#tantekalright, updated my posting UI so it makes it clear when POSSEing to Twitter or not (omitting character count / elllipsing if not), and at the same time added confirmation of what type of post it is publishing (article, note, reply etc.)
#tanteknow to see how Bridgy Publish to GitHub fails when I haven't auth'd Bridgy yet
#Loqi[tantek] The “Authorize Bridgy” prompt page has a big green button that says to “Authorize snarfed” which seems a bit incongrous.
It would be better for branding/trust reasons if this prompt for Bridgy could use the github.com/bridgy organization, a...
#tantekonly weird thing - the image didn't come through
#snarfedooh images, you may be the first to test that
#tantekok so I have basic (text) POSSEing replies to issues on GitHub working thanks to GitHub!
#tantekthat was A LOT of code reworking and I ended up simplifying a bunch of code and a few minor frontend posting UI improvements as well
#tantekthe next test will be to see if I've broken any "normal" (previously working) posting :) but that'll have to wait til I get home
#tantekand then I'll try publishing a new issue to GitHub - that *should* work too with the code I've written so far but we'll see
#snarfedbtw i expect the "silent" failure, wm to bridgy publish without a github account, was a 400 with a descriptive error message. feel free to let me know if not!
[tantek] joined the channel
#[tantek]Yes I should write code to check that, de-auth and try again
#[tantek]LMK if there’s something special I should be doing to make the img work
#ZegnatOh, I wonder if that change might cary over to ampmail (mailamp? What was it called?) That would then open the gate to JavaScript-in-email-in-Gmail
KartikPrabhu joined the channel
#aaronpkyes that was the main criticism of bringing amp to email
KartikPrabhu, eli_oat, [miklb], snarfed, [kevinmarks] and [xavierroy] joined the channel
#ludovicchabantZegnat: can’t remember which version I’m using exactly but it’s the one you get when you install the package with pip
#KartikPrabhuludovicchabant: the pip is v1.0.5 . we are working on adding some updates to it and push a new version. 1.0.5 works nicely for most cases though
#KartikPrabhuludovicchabant: also a tip "pip freeze" will output all the installed libs and their versions
#Loqi[cleverdevil]: snarfed left you a message 1 hour, 41 minutes ago: for a mass dump, should be easy to throw that in a loop that reads from file(s), or in a python -c ... command
#LoqiFacebook is a popular content hosting silo and activity aggregator most well known for being the largest centralized social network on the web https://indieweb.org/Facebook
#snarfed[cleverdevil]: btw there are type-specific conversion fns (event_to_object, album_to_object, etc) if you have different files for different types, or know which type at runtime
#[kevinmarks]If python 2 is being deprecated, Google needs to get python 3 running on appengine
#[kevinmarks]They have had most of the core committers there for over 10 years
#snarfed[kevinmarks]: yup. one of the longest running issues on their tracker. :P flex has supported 3 for years, and they've been working on getting 3 into standard for years. i know they do plan to ship it.
#ZegnatIf you are doing some planning on separate tasks and things, I would probably recommend splitting IndieAuth off of the Micropub implementation as well.
#ZegnatThe biggest challenge is coming up with an API so new plugins for custom post types can leverage Micropub. I had some ideas about it but felt all of them needed too much reworking for me to do it over a weekend.
#Loqi[snarfed] i'm repurposing this issue for the big new redesign we want to do for coordinating with other plugins and themes that render mf2 data. (@dshanske, please holler if you think i shouldn't!)
we're going to implement @kraftbj's design sketch in https:...
jackjamieson joined the channel
#[cleverdevil]Yes, I totally agree @Zegnat. That's where i get out of my depth. I know precisely how I'd do it in Python, but not so much in PHP.
#ZegnatOh, let me read the WP discussion, might have some ideas for Known :)
#ZegnatSo one of the things that WP already seems to be doing, that Known isn’t doing but I wish it did, is that it stores the submitted mf2.
#snarfedyup that was definitely a good call way back when
#ZegnatI’d love to modify Known to use mf2 to store its posts internally. But currently it does PTD on the mf2 first, converts it to one of their post-type-classes, and that post-type-class gets stored.
#ZegnatWhen I came to the conclusion that changing the storage format was the first thing on the agenda, I tapped out. As I am not using Known myself and wouldn’t manage that in a weekend as a little side project to help other people out
#[cleverdevil]One of my longer term goals would be to actually make it possible to store everything on-disk as mf2, with some smart index files, eliminating the need for a database entirely.
#[cleverdevil]But, that's sort of a personal/pet issue, not really relevant to the Indieweb goals.
#[cleverdevil]Priority one, IMO, is to make Known 100% the most compliant and complete implementation of IndieWeb standards.
#snarfedthere is a partial or complete known flat file backend, right?
#ZegnatI think Known has an odd db design. But that seems to have some legacy reasons.
KartikPrabhu joined the channel
#ZegnatI would love to sit down and talk with some people who actually use Known as their daily driver and see what they think. Completely with you [cleverdevil] that it would be lovely to have Known super compliant!
#Zegnat(I actually always thought more of Knowns internals were IndieWeb.)
#[cleverdevil]It basically is that way because they started with MongoDB, and then shoved things into MySQL.
#[cleverdevil]It is also very slow for a lot of things as a result.
#snarfed!tell tantek thanks for the nudge on bridgy's github auth prompt. i was reluctant because i didn't want to move the repo, but realized i could just use the org for the app. prompt now says "Authorize bridgy"! https://snarfed.org/github_bridgy_org_auth.png
#[eddie][cleverdevil] I also saw that mapkyca resolved an issue I filed in Known. The token endpoint now returns JSON instead of urlencoded. Yay for standard compliance!
#[cleverdevil]Of course, now Likes are broken for me in Micropub, LOL.
#aaronpkthat's essentially what I do for my site too except my canonical data is stored in files instead of another table
#ZegnatYep, I don’t really have anything against that pattern. I just wish Known would sture the mf2 json in there, instead of a serialised post-type class
#[kevinmarks]You could possibly store both and gradually change over.
#@jobsonnentag@Blot__@kevinmarks@desparoz I apologize upfront for my newbness, but neither. The JS from WMIO to retrieve webmentions needs a target url. I thought I could pass in the {{url}} tag by itself, inside the WBIO JS, which would sit head of my entry.html. This didn’t work so I tossed in a static URL for testing. (twitter.com/_/status/966053841425793030)
#snarfedfor anyone here signed up for github on bridgy, backfeed (aka listen) is available to beta test if you want! you can turn it on your user page