#[kevinmarks]That's my point - at the moment we have a UI for spreading a post to silos, but not to other indieweb sites. Manton and Aaron are saying that we need 2 separate UIs, I'm saying that we should try for a unified one.
[miklb] joined the channel
#[miklb]GWG could probably do that with a gulp/grunt task
#GWGI'm going to use a broader icon pack. It has a JSON of the colors for the various site logos. So, it is structured "title" and "hex" I need to read those, convert to class name and color.
#schmartyOops I missed the destination discussion. I have this problem with micropub clients and posting to both my personal site and my comedy podcast site
#tantekschmarty ^^^ want to start that page with precisely that real world use-case you have of two different indieweb sites that you want to directly post to?
#schmartyThey are both different sites that receive very different posts, but most micropub clients apps only let you log in and post to one at a time, requiring a log out between
tantek joined the channel; tantek left the channel
#schmarty!tell aaronpk i got an error uploading an image to the wiki: "Could not open lock file for "mwstore://local-backend/local-public/0/0e/facebook-page-posting-as.png"."
#Loqiaaronpk: tantek left you a message 54 minutes ago: file uploads may be broken post MediaWiki update - schmarty was having difficulty
#Loqiaaronpk: schmarty left you a message 53 minutes ago: i got an error uploading an image to the wiki: "Could not open lock file for "mwstore://local-backend/local-public/0/0e/facebook-page-posting-as.png"."
#aaronpkso the main question I have around this idea is whether it should be a client-side thing or a server-side thing. e.g. it wouldn't take any protocol changes to make it so I can just authenticate with a bunch of micropub servers at a time in quill, and quill would keep a separate login session associated with all of them together
#aaronpklike, why should I have to sign out of one endpoint in order to sign in with another? why not keep multiple signed in at the same time
#aaronpkin facebook it's less of a full context switch than twitter and instagram
#tantekFB does both full context switch, and per post
#tanteksince that "full context switch" UX seems to be what's commonly implemented (and thus users likely understand), probably better to model how would you do that with Micropub first, before other variations
#[manton]Facebook seems the closest to what I'm trying to do, since it's one user account and multiple pages/sites. On Twitter and Instagram, you have to switch accounts to post somewhere else.
#[manton]If you follow the Twitter-style model, the current Micropub spec is fine since the client could just manage switching between multiple user sessions, each with its own web site.
#aaronpkI think I like facebook's take on it the most, at least for the posting interface
tbbrown and [manton] joined the channel
#[manton]Yeah, TweetDeck's UI is nice, but also might work better for Twitter which usually has unique avatar icons. Not all web sites have favicons, so you'd want to see the text or domain name of the site by default.
#aaronpkThe other big difference with tweetdeck is it lets you select more than one
#LoqiZegnat: tantek left you a message 6 hours, 20 minutes ago: that "pick a destination" screenshot that you shared earlier from FB - could you add it here? https://indieweb.org/destination#Facebook
#petermolnarjust out of curiosity how would you handle a webmention if the source post date is in the future? (eg. scheduled post, but accidentally sends a webmention)
#sgregersknebel: i believe you've been working on parsing feeds with python? could you advise to what is the best starting point for that? some reasonably reliable library?
#sgregersknebel: thx. now that i have the twitter parsing in place, i wonder would it make sense to merge rss feeds and incoming links from twitter into one tool ;)
#sgregerthere is a bit of a semantic difference between "links shared by people i follow on twitter" and "new posts on sites i subscribe", but in terms of my workflow that might not matter all too much.
#sknebeldo you have a feedreader you're happy with? then maybe turning the links from twitter into a feed would be an alternative
#sgregerindeed! i was thinking about that as well. i currently use feedly, but would like to go indie. having my own db/ui would allow seamless integration with bookmarking etc
#sgregeri actually have the PyRSS2Gen docs open as we speak
#ZegnatI seem to recall something about Chrome doing WebM in IMG as a follow-up to animatied-WebP. Maybe they never implemented that. But is the obvious parallel to this.
#ZegnatTook me half an hour to get my nginx redirects working how I wanted. But I think I learned a lot today :D (And zegnat.net + its wiki have moved to https on the new server.)
#aaronpk(in case you didn't click the link, it's implemented in safari tech preview now)
snarfed and tbbrown joined the channel
#tantek.comedited /icon (+681) "expand dfn to include usage for types of posts, "how to" section/headings, unlink inactive homepage, add icon sets section with examples of icon sets, and supporting software (WP Post Kinds)" (view diff)
#sknebelit does produce python dicts that are kind of similar to jf2, but I don't think it can produce jf2 directly (e.g. it parses dates to python datetime datatypes, which I don't think would get serialized correctly by default)
#Zegnataaronpk is there a reason for limiting to https? and no ports?
#Zegnat(That first questionmark is regex, the second is the actual sentence end. Woops.)
#sknebelben_thatmustbeme: it seems to treat e-content (gets turned to "content":"<html>..." instead of an object) and author (not marked as "type":"card") differently
#sknebelbut it's closer than I thought, I had some of it's specialized functions in mind
#tantekaaronpk agreed, keep it as tight/minimal as possible
[kevinmarks] joined the channel
#[kevinmarks]@sgreger python feedparser is the best way to parse feeds in any language. You're welcome to use the version I forked off before the current maintainer split it into loads of files.
#ZegnatI was mostly wondering about the http(s) requirement because of the people who are testing statically hosting their things on e.g. ipfs.
#ZegnatBut especially if you are describing current implementations this look right to me!
#aaronpkthe other reason is they need to be resolvable so clients can do discovery on them, and I'm not about to require that every client support additional (or arbitrary) protocols other than http
#ZegnatGood to have a reasoning in place for the limitations you write into the spec :) Rather ask one too many questions and get the reasons than accidentally limit the spec too much by accident.
#LoqiActivityPub is a decentralized social networking protocol (W3C Candidate Recommendation) based on ActivityStreams https://indieweb.org/ActivityPub