#Phykshi, I've been hanging around for a while, to remember to have a look at microformats and h-feeds and finally got some time to handle it.
#PhyksI'm wondering about the implementation of h-feeds, as if I understood correctly, that only requires to add some classes on key elements in HTML (h-entry, and so on)
ShaneHud_ joined the channel
#Phyksbut then would that mean that a feed agregator wanting to parse my feed would have to handle the full HTML of my blog and parse it to extract the relevant data ?
#aaronpkPhyks: there is a well-defined parsing algorithm that takes the additional classes you add and returns a nice data structure
#Phyksbut I'm wondering how does it compares to parsing Atom feed directly
#PhyksAFAIK, parsing HTML requires much more computation power and network bandwidth than a simple feed, no ?
#aaronpkthe difference in parsing HTML vs XML is so slight that in practice it really doesn't matter
#aaronpkand HTML feeds tend to be smaller than Atom or RSS
#Phyks(actually, I was wondering whether I should add this on the homepage, or create also a separate light page with only the necessary elements, ATOM-like)
#Phyksactually, maybe it's a vision formatted by my habits with Atom/RSS, but it seems to me that using my homepage is definitely not optimized as up to half of the actual content can be really useful for the feed for instance
#Phyksand I was wondering whether it was still the way to go, or I missed something
#eliemichelIn my opinion, h-feeds can't completely be machine-friendly as soon as they are human-readable
#eliemichelhumans do not want to read at the same time 50 articles
#aaronpknot really. if you're polling for updates, you really don't want to get the last 50 articles every time
#gRegor`Eh. "machines" can easily be made to make multiple requests to get up to X posts, if that's what they want
#eliemichelwell, if you want to automatically gather the whole articles of a blog, you should browse many pages if they are organised in a human-confortable way
#Phyksaaronpk: as I see them, "feeds" and "syndication" is a machine to machine process, whereas HTML usually serves machine to human content. That may be why it seems strange to me.
#eliemichelMaybe you don't want, but that seems to be what Phyks want
#gRegor`Phyks: I think the pros outweigh the cons. Even if only half of the HTML on your homepage is mf2, the benefit of having one version of your data and not having to maintain an XML feed, is worth it.
#eliemichelAnd that's what I was looking for some days ago
#Phyksbut still, it seems to me it is more human oriented that machine oriented
#aaronpkyep, a common problem is that sites will update the human-readable version and forget to update the machine-readable version, leading to broken rss feeds
#aaronpkso focusing on making the human-readable version also be machine-readable solves that
#Phyks(ie machines do not care about readability, etc., the only matter is efficiency and low bandwith requirements)
#Phykshalf of the info on my blogpage is useless for the feed
#Phyksbut still, the feed is a bit larger than the html file
#snarfedalso, if you're worrying about your web site's speed, html file size is pretty damn low on the list of things to optimize. *many* other things matter more.
#gRegor`The pin13.net parser aaronpk referenced runs php-mf2, so feel free to test it out there
#Phykssnarfed: not really worrying for my speed, but for the other speed. An increase of 10% of the size, would mean 10% more time to refresh the feeds, and that's a real problem for an agregator
#snarfedPhyks: no, that's not the math. lots of stuff matter for network time and cpu time for processing. file size is only one (often small) factor for either
#aaronpkI think even a 10% increase in time to process would not be a thing worth optimizing
#snarfeddesign/write the simplest code possible until something is actually too slow. only then optimize
#Phyksaaronpk: I'm working on a feed agregator right now, that's how I came across mf2, and reducing the refresh delay is a real pain. No need to increase it :)
#Phyks(especially when I have to download more than 100 feeds each time :)
#aaronpkaha! cool. well you're going to get a much bigger speed boost by being smarter about polling vs the (hypothetical) 10% increase in parsing time
#Phyksyes, I know about polling, that is planned as well
#aaronpksuch as using PuSH to be proactively notified of new content
#Phykssnarfed: I read a lot about PubSubHubbub, but did not come accross superfeeedr. But one of the main problems remains the implementation on server side…
#Phyks(RSS / Atom already has some mechanisms to reduce polling, but they are barely used as far as I could see)
#eliemichelMy problem with h-feeds is that I have just no page where full articles are listed. On the home page I put only summaries. So to be able to get the full feed I would need to query each article-related page…
#PhyksI display the date in a specific format, for instance dd/mm on my blog, so this means I should add an element with the full date and hide it in CSS ?
#snarfedPhyks: sure. rss/atom still need polling. the point is that superfeedr does it for you, so you don't have to. (they amortize the polling across all of their clients.)
#Phykssnarfed: yes, but that mean one more centralized service
#PhyksI'm not in favor of feeburner… nor am I for superfeedr…
#tantekkylewm - it looks like a duplicate event post - not sure what is going on there. aaronpk - what's the reasoning for two indie events for the same event? did you repost KevinMarks's indie event?
#KevinMarksI mean known doesn't POSSE events itself yet
lukebrooker, KevinMarks__ and RichardLitt joined the channel
#tantekKevinMarks: right, if you manually POSSE a FB event for your indie event, Bridgy will send RSVP webmentions from the FB event back to your original indie event
mdik, tantek and scor joined the channel
#KevinMarksHm, so I need to connect bridgy to my known site?
#KevinMarksCan bridgy understand that I have more than one indieweb site for a given Facebook account?
#kylewmKevinMarks: Bridgy will only search one domain for posse-post-discovery, but if your posse copy includes a link to the original it should still find it (i think)
#KevinMarksSo do I need to sign up with known.kevinmarks.com as well as kevinmarks.com?
#kylewmnope, I don't even think you could if you wanted to
#kylewmamusing quote ... her: "oh there is some explanatory text for dumb people", me: "hey not everyone knows what a bookmarklet is", her: "all women do!"
#Phykstantek___: thanks ! I'll add myself. My personal website is http://phyks.me as linked by eliemichel (content mostly in French and occasionally in English)
#GWG!tell acegiak I see your Whisperfollow plugin uses the JSON API. I have an idea for using it to add a reply/like/etc endpoints to Indieweb Taxonomy
#LoqiWhisperFollow is a WordPress based social reader create by acegiak that currently supports H-Entry, RSS, Atom and PubSubHubbub http://indiewebcamp.com/Whisperfollow
#ben_thatmustbemejust with the link or shortlink i can already pull up the post, its just the UI for the micropub page, and UI for syndication links being displayed
#ben_thatmustbemeaaronpk, i was trying to think of things that url= would be used for other than setting the URL for a new post. and that would actually line up perfectly
#ben_thatmustbemeso url= would be the URL of the post, if it doesn't exist, the MP endpoint should create it with that URL (if it supports that), if it does exist, update it
#voxpellitantek: original idea was that the config wasn't necessarily tied to webactions but the indieweb in general, but true, the use case is just for web-actions right now I agree that it might be a better protocol
#tantekcool. I also think that it has plenty of room for growth too.
#tanteksince "web actions" in general describe doing something from one site across to another site (e.g. your own site), and that's exactly what registering a protocol handler is for in general!
#voxpelliyes, I like that everything don't have the "indie" name as well
#tantekthanks to voxpelli's innovative use of protocol handlers and postmessage, we've got webactions working cross-site WITHOUT having to install anything!
#voxpelliThe one with the postMessage() could be locked down as well, but would be advisable to at least have an alert() or something telling you if you're not logged as you will have a hard time noticing
#tantekaaronpk - or it can be on your *logged-in* home page - i.e. only send that registerProtocolHandler script to the client if your server knows that it's you that is logged in
#Loqiindie-config is a method of using protocol handlers and postmessage to setup your indie website to both notify the browser that it can handle webactions and then do so http://indiewebcamp.com/indie-config
#voxpelliaaronpk: a cool privacy hack is to use a confirm() in the iframe to ask if one wants to whitelist the current page, save a whitelist in localStorage and only send a postMessage() if the document.refeerer matches a page in the localstorage
#voxpelliaaronpk: If you run the registerProtocolHandler() on the same page that's already registered nothing will happen – it will know that everything is registered as it should
#tantekand it should load *your handler* not your homepage
#aaronpkmy handler only needs to be the JS that runs postMessage. so you didn't intend on me also putting that on my home page?
#voxpelliaaronpk: right now my script lazy loads the iframe
#voxpelliI had a problem in Chrome when quickly prototyping where it insisted on only loading the index page, that's why i opted for the index page in my code
#voxpellididn't have time to dig deeper into it and understand why
ShaneHudson joined the channel
#tantekindeed. aaronpk, interestingly enough this protocol handler / webaction stuff worked/works better in Firefox - so you might want to try it out with that first while you're working with it
#tantekonce you have it working there like it should, then we can document the Chrome problem(s) and file bugs accordingly.
#voxpelliyeah, I will look into separating out my indie-config stuff from my web-component to aid with that, the web component is much more complex and should take all the stuff we talked about in Brighton into consideration
#aaronpkthis is the reason for the state parameter
#aaronpkyou can set a session ID as the state parameter so that you can store the client_id for your request server-side
#sammachinaaronpk: is there any way to pass additional parameters through indieauth, say have a page on site A but you indie auth with site B and then get redirected back to site A
#sammachinbut there could be multiple site A's and only one site B so it needs to know where to send you back to
#aaronpkyou should be able to do what you need with a combination of redirect_uri and state parameters
#aaronpkkeep in mind that you'll (eventually) need to "register" the redirect URLs, so you might be better off finding a way to do this with "state"
#voxpelliaaronpk: tantek: if using a absolute URL in registerProtocolHandler then any URL seems to work in Chrome, with relative it seems to always resolve to the top path
#sammachinaaronpk: hmm that might be harder as for this I won't always know the redirect urls, the idea is that a user can drop a widget onto their site that uses my backend app and indie auth
#KevinMarkshm, known does, so we can try with that
#tantekthis is why it's important to get the "How to" setup steps documented by someone who has successfully done so but was *not* @IndieWebCampUK - I think aaronpk's notes will be essential
#aaronpki'm trying just launching web+action:// links on my own, and it's really messede up
#voxpelliyes, much more messed up, but I'm able to load it when trying locally
#aaronpkbecause I have both the chrome and firefox versions registered to open web+action:// urls, firefox prompts me to ask which I want to use to open it
#tantekfortunately there's been lots of thinking about the UI by various folks
#tantekand this weekend I just figured out decent UI for in-stream web actions
#tantekmore like researched / determined based on prior art (e.g. Twitter)
#tantekfunny thing is, I got stuck in needing to update some storage code to make this work on my home page, so now I'm considering a pretty big overhaul there.
#aaronpkwell the good news is it didn't take me long to add support for web+action links on the author side
#voxpelliThinking I should add the loading exmaple somewhere, would linking to the Gist work or do we want it into the wiki itself?
#sammachinaaronpk: is state implemented in indieAuth? I'm setting it on the initial form but it doesn't seem to be in the callback or the response when I validate the code
#kylewmso the thing I'm confused about now is why does web+action need to do the configuration back and forth... as opposed to registering the intent url itself as a handler for web+action?
#aaronpkyou can find out what iOS apps are installed on a phone that way
#aaronpkcause a lot of apps register their own URIs, like facebook:// twitter:// etc
#voxpelliwith this solution, as I confused aaronpk with earlier, you can instead control your own privacy by yourself deciding if you want or don't want to do the postMessage on a site by site basis
#voxpellibut I guess we shouldn't be thinking about that too much that until people get used to the simple concept :)
#voxpelliMy train is arriving back home now so I have to leave
#tantekaaronpk - really re: "you can find out what iOS apps are installed on a phone that way" ? - that would be very handy for my People Focused Comms UI
#ben_thatmustbemeit should be super simple for me to implement full editing via micropub now
#tantekcan you show me a code snippet illustrating what you mean? E.g. is "facetime:" registered -> hide/show Facetime button
#aaronpktantek: that is totally possible from within an iOS app. not sure if safari implemented the isProtocolHandlerRegistered thing
#ShaneHudsonYeah you all did great, got some interesting discussion going in irc :) I think some people were starting to get the point of it, Leo certainly did
#sammachinanother indieweb question, can I have more than one domain but sign in using the same twitter account?