2019-03-03 UTC
# 00:00 swentel I'm referrrring to that image right underneath "The tree of objects describes the list of blocks that compose a post." # 00:01 swentel however, plugins can probably introduce new block types, and it's hard to know I guess what is 'content' # 00:02 GWG swentel: This is the problem. Right now, I ignore all Gutenbergy stuff. # 00:02 GWG But that will get harder in the future. # 00:03 GWG Also, people will want to create 'reply' blocks and such as a future UI # 00:03 [tantek] I want fragmentions for picking a specific img in a post! # 00:04 GWG swentel: I have a class that takes a post object and expresses it in MF2 and allows setting MF2 properties. It is an abstraction. # 00:05 GWG I use it so that I can update/amend it and not have to change the higher level code much # 00:05 GWG But I have been staring at it for the last few days(sickness hasn't helped), trying to figure out the best approach to updating it to work more effectively. iasai joined the channel
# 00:07 [tantek] oh I know I just saw above "<!-- wp:" :face_vomiting: # 00:07 swentel GWG, can I see the old class, trying to get a mental picture by looking at code (don't know much about wp internals) tw2113 joined the channel
# 00:08 GWG If I can't count on 'post_content' == mf2 content, I'm not sure what to do # 00:08 GWG Micropub stores content as a property as well, sort of a 'backup' content. # 00:09 swentel right, post_content could contain all sorts of dedicated mf2 properties now? # 00:11 GWG Micropub does add them if Post Kinds isn't installed # 00:15 swentel hmm I'm afraid it's too late to completely thing along now KartikPrabhu, iasai and [kevinmarks] joined the channel
# 00:40 [tantek] only if it's made visible because the image didn't load (or had some other problem) # 00:41 [tantek] really I want fragmentions for images when there's more than one in a post (though I suppose I could synthesize IDs), and alt text seemed like a valid hook to use KartikPrabhu, snarfed and iasai joined the channel
# 01:21 jacky it kinda goes back to what people are collecting from people when they sign into a service # 01:21 aaronpk we've gotta get an indieauth provider into gogs/gitea/gitlab # 01:21 aaronpk yes often the service wants stuff like the user's name and profile photo, and sometimes even email address # 01:21 jacky [tantek]: I think so. In the case of ElixirStatus; they just pulled my name and photo # 01:22 [tantek] I feel like GWG's Quill update has demostrated that can work with IndieAuth (name and photo) # 01:23 GWG tantek, I credit swentel and aaronpk for the original conversation. I just implemented first # 01:24 GWG I would like to see how hard Indieauth for Nextcloud might be # 01:24 GWG There's a place it might be really popular # 01:29 aaronpk but this plugin talks about doing registration stuff which is what regular oauth stuff needs but indieauth doesn't need # 01:33 Loqi [nextcloud] user_external: 👥 External user authentication methods like IMAP, SMB and FTP # 01:34 jacky granted this sounds like it'd require a whole new plugin # 01:34 jacky which again isn't too much of a problem since it only really just needs a username snarfed joined the channel
# 01:37 aaronpk seems like the real trick would be whether URL characters are allowed as usernames, and if not, how to create a username from the indieauth URL # 01:38 jacky going to open more issues to see if I can get indieauth in more places # 01:39 aaronpk argh you're making me want to whip indieauth.net into shape # 01:39 jacky the only thing for me is time and (programming) language barriers # 01:40 jacky but I know that I could pitch + land indieauth in pixelfed # 01:40 aaronpk pixelfed is written using laravel which i'm pretty familiar with snarfed joined the channel
# 01:42 jacky ugh I feel like I'm bad at this lol - I can do bug reporting but feature requests? blergh # 01:49 aaronpk tbh I'm more excited about projects like these becoming an IndieAuth provider # 01:49 aaronpk once there are more providers then there is more reason to have consuming sites support it iasai joined the channel
# 01:55 jacky this is my most productive saturday code-wise all year lol # 02:06 aaronpk anything self-hostable that provides some form of identity online # 02:06 GWG aaronpk: That's why I thought of Nextcloud. # 02:06 GWG Home Assistant already uses a variant of Indieauth. # 02:07 aaronpk yeah home assistant used the client ID part of indieauth # 02:09 jacky hmm my mf2 parser doesn't like h-entry with nested h-cites it seems # 02:11 GWG I translate doesn't like to..in need of enhancement # 02:11 aaronpk hm sounds like it didn't follow the parsing algorithm right # 02:11 GWG My code doesn't like a lot of things it should # 02:12 GWG jacky, for example, I discovered that Monocle didn't like my jf2 output # 02:15 GWG I feel like I need someone to check my work to make sure I didn't miss anything iasai, snarfed and [tantek] joined the channel
# 03:09 Loqi [Amolith] @jalcine That's exactly the kind of thing I'd like to look at hosting! Thank you for the suggestion! iasai joined the channel
# 03:26 GWG I'm eager to host more aaronpk services myself iasai joined the channel
# 03:57 GWG Okay, how does that outline look? # 03:57 GWG Tried to lay out what swentel and I are doing. jjuran and iasai joined the channel
# 05:00 Loqi [tsileo] microblog.pub: A self-hosted, single-user, ActivityPub powered microblog. iasai, barpthewire, [frank] and [kevinmarks] joined the channel
# 10:48 jeremycherfas Is there an EASY way of doing an A/B test on a website without any messing about? I mean, deciding at random whether to serve A or B to a site visitor. [mrkrndvs] joined the channel
# 10:51 jacky jeremycherfas: what would be 'messing about' be classed as? # 10:52 jacky ah well you could do some silly 'uniqueness' tool (like on the visiting IP) whether or not to bucket a person in groups # 10:52 jacky stick it into their session (or not if you don't care that they stick networks) and go from there # 10:52 jeremycherfas I've a feeling anything other than GA is going to be complicated. And most likely what I learn will not be worth learning. # 10:53 jacky just need something to make people fall into a group # 10:54 jacky are you looking for a service or wanted a method? # 10:55 jeremycherfas To be honest, it doesn't actually have to be random. Time is a neat one. I could do something different if minute/modulo 5 is odd or even. barpthewire joined the channel
# 10:56 jacky has too much experience with that from his time at Lyft # 10:57 jacky the number of experiments these companies run on people eduardm, KartikPrabhu and [kevinmarks] joined the channel
# 11:59 [kevinmarks] Bear in mind that the stats of a/b testing are tricky and you need a lot of volume. sandhawke joined the channel
wagle joined the channel
djmoch and [tonz] joined the channel
# 14:38 Loqi It looks like we don't have a page for "web application firewall" yet. Would you like to create it? (Or just say "web application firewall is ____", a sentence describing the term) # 14:40 sknebel Web application firewall are software that inspects HTTP requests and tries to block malicious or suspect requests, often with a high risk of false positives due to broad defaults, and limited effectiveness against a motivated attacker. iasai and [kevinmarks] joined the channel
[jgmac1106] and sandhawke joined the channel
# 15:17 GWG sknebel: Is there an easy way to get nearby POI from OSM in a query? # 15:17 sknebel it's an interesting question what kind of info you'd need in that query - a lookup might tell you all kinds of stuff, but I guess e.g. the street address isn't needed in this lookup and could be only on the local venue page if there is such a thing # 15:19 GWG sknebel: I tried to define it very specifically. The return is an h-adr, but the only mandatory properties are the label, latitude and longitude. But it could return any h-adr property. # 15:19 sknebel I'm still somewhat interested in finding a solution that doesn't use OSM community resources too much, even if we find servers we technically could use for free # 15:19 GWG It's just a proof of concept. Indigenous for Android + Micropub for WordPress/Simple Location actually do support the geo property, but not the venues property yes # 15:20 sknebel right, it's even on the wiki, i totally skipped over that part iasai joined the channel
# 15:21 GWG sknebel: Not set in stone though, I just fleshed out original 2015 comment. # 15:21 sknebel sgreger had the idea that maybe one could get the data out of OSM vector tile sets pstuifzand joined the channel
# 15:22 sknebel but I'm not sure about that yet. poked a bit at the formats used there, but didn't get quite through # 15:23 GWG Well, the discussion made me feel better about borrowing the label and URL from Foursquare # 15:23 GWG Until I can start building my own database [tantek] and tw2113 joined the channel
# 15:53 tw2113 last i knew and understood, that's supposed to tell the "powers that be" that the value presented there is the one they should index/consider most important? # 15:53 tw2113 and the page with the tag is just a copy of sorts # 15:55 [jgmac1106] okay thank you, I am the only power to be to me so will skip it, hypothes.is listed as a best practice for your articles..they also suggest OGP but I skip that # 15:56 tw2113 the powers that be that i claimed would be the search engines # 15:56 vilhalmer I think it only really matters if a given article has more than one url [kevinmarks] joined the channel
[tonz] joined the channel
# 15:58 Zegnat It is also really nice for when you have a page available on both example.com/123/ and example.com/123 (without trailing slash) to let people know which one of those two URLs you want as the “real” one # 16:00 sknebel (although in that specific example I'd question why there's not a redirect from one to the other # 16:01 sknebel but yes, if you have multiple urls with the same content, point a rel=canonical to the canonical one so it's clear they're just different represenations of the same thing # 16:01 sknebel e.g. hypothesis then shares the annotations between them nloadholtes, leg and snarfed joined the channel
# 18:41 Zegnat I wonder if *any* browser vendor is actually working on these specs # 18:43 Zegnat Firefox and Chrome have completely incompatible implementations of native messaging # 18:48 Zegnat I like Firefox’s implementation, but Chrome’s matches the spec # 18:48 sknebel incompatible as in "same code can't work in both" or as in "need to duplicate stuff"? # 18:49 Zegnat "same code can't work in both". Method signatures are different. # 18:49 Zegnat Firefox returns a Promise, Chrome needs a callback function specified in the method call. # 18:50 Zegnat Guess I can always pass the callback function, and then check if the return is a Promise, in which case I call the same callback function on .then() … or something # 18:52 sknebel yeah, at least sounds like you could wrap that once centrally # 18:54 Zegnat Feels very much like you would just end up writing for one of the browsers, and then wrap whatever doesn’t work in the other # 18:54 Zegnat But that’s not what I was promised with WebExtensions :P # 18:54 sknebel you can also go file a bug with mozilla to please give you the callback form too # 18:55 GWG I am still struggling with my storage problem # 18:55 GWG It keeps leading me in different dire5 # 18:56 GWG The date was not consistent, and that code was in one place # 18:56 GWG Now the type isn't consistent and I need to figure out a Micropub issue # 18:57 GWG If the post was by Micropub then it stores the h property in a 'type' key # 18:57 GWG I am wondering if removing that would break compatibility # 18:57 GWG I don't know anyone using it for anything right now [jgmac1106] joined the channel
# 19:03 GWG I am obsessing a bit much about this KartikPrabhu joined the channel
# 19:19 GWG It is really holding me up though iasai and [grantcodes] joined the channel
# 19:59 [grantcodes] I use that for my extensions. One codebase for all browsers, but it probably does mean you need a build step iasai, [frank] and [Rose] joined the channel
# 20:45 Zegnat I’d rather not introduce a build step ... unless you know one that will actually result in properly packaged extensions for the different browsers # 20:57 sknebel hm, of course a simple OSM query about anything one might consider a POI is now happily reporting all public trashcans and park benches in a wide radius # 20:58 GWG sknebel checked in a trashcan. Not very compelling # 20:58 sknebel I'm always confused people leave reviews on google maps for s-bahn train stations # 20:58 GWG Would make for an amusing post, but not as intended # 20:58 GWG sknebel, people review everything now # 20:59 sknebel At least at IWCs people check in at the weirdest things too # 20:59 sknebel so allowing checkins at trash cans and plant pots might be a feature for some # 20:59 sknebel or at each of the individually tagged bicycle parking thingies in front of the university library # 21:00 sknebel (at least as soon as someone makes a virtual-coin-generating aggregator :D) iasai joined the channel
# 21:02 GWG sknebel, keeping up with the founders # 21:02 GWG tantek takes Microformats everywhere KartikPrabhu, iasai, [chrisaldrich] and [tantek] joined the channel
# 21:34 Loqi Ok, I'll tell them that when I see them next # 21:40 Zegnat I am so close now… I should be going to bed but it looks like I am just minutes away from getting a WebExtension talking to a PHP script that executes an AppleScript that talks to some software :D # 21:46 [tantek] Zegnat, are you checking the latest version of the spec (editor's draft) that you're talking about (FF vs Chrome vs what spec says) ? # 21:47 [tantek] And if so, if you do prefer the FF implementation, can you please file an issue on the spec stating that the spec should be fixed to work the way FF does? And definitely provide reasons for why it is better. # 21:47 [tantek] That kind of direct webdev feedback is essential to getting specs improved # 21:48 [tantek] it's not even an editor's draft. goodness sakes it's definitely open to changes then iasai and snarfed joined the channel
# 22:20 Zegnat I can go to bed! Firefox->PHP->osascript->third-party software, completed. snarfed and iasai joined the channel
# 23:10 jacky implementing an indieauth provider was easier than I thought # 23:11 jacky it's just the wiring up of relmeauth that's taking up most of my time # 23:11 jacky making sure each one I want to support _actually_ exposes a rel=me link somewhere # 23:12 jacky aaronpk: yeah b/c it's going to be a easy way to get adoption # 23:12 aaronpk cool are you making something like indieauth.com then? a provider that doesn't use passwords? [schmarty] joined the channel
# 23:21 [schmarty] jacky: nofollow should be fine? I don't think a RelMeAuth parser has to take notice of that. iasai joined the channel
[tantek] Kevin Marks can we extend /fragmentions to support alt text?
[kevinmarks] Hm. Does page search in the browser search alt text?
jacky shouldn't be too hard https://help.nextcloud.com/t/nc12-build-in-oauth2-client-authentification/14852
Loqi [Jacky Alciné] Wanna Federate the Web? Stop Using Silos for Signing Inhttps://v2.jacky.wtf/post/use-indieauth-more
@jackyalcine https://v2.jacky.wtf/post/use-indieauth-more
Wanna Federate the Web (and Make It More Friendly)? Stop Using Silos (like GitHub, Twitter) for Signing In (twitter.com/_/status/1102030722791948289)
jeremycherfas Is there an EASY way of doing an A/B test on a website without any messing about? I mean, deciding at random whether to serve A or B to a site visitor.
@jgmac1106 ↩️ @tbc0 I kinda like the thread UI in my microblogging. I try to webmentions to keep posts threaded by topic. (https://quickthoughts.jgregorymcverry.com/s/2l5s82) (twitter.com/_/status/1102189792060682240)
sknebel mod_security is a [[web application firewall]] for the [[Apache]] web server.
[jgmac1106] question for people what is rel="canonical for?
[grantcodes] Zegnat: Mozilla makes a polyfill
aaronpk I just added a changelog section to the indieauth spec https://indieauth.spec.indieweb.org/#change-log
Zegnat Any idea what the status is of those drafts anyway, [tantek]? I noticed you were the last to supply a commit in 2017 to the main draft (https://browserext.github.io/browserext/)
[schmarty] jacky: nofollow should be fine? I don't think a RelMeAuth parser has to take notice of that.