tantek_, [tantek], [jgmac1106], KartikPrabhu, [kevinmarks], AngeloGladding, [matthilt], eli_oat, renem, [snarfed], [miklb], [mrkrndvs], [eddie], cweiske, snarfed, bhh, leg1, [pfefferle], loicm, snarfed1 and swentel joined the channel
#ZegnatHmm. Wondering how to send a notification to someone posting to GitHub from their website (e.g. tantek) that I have mentioned them in a reply I wrote on GitHub.
#ZegnatI block everything in uMatrix. images, css, javascript. To save on data-usage. Though the saving from CSS is basically 0, it does sometimes stop multiple third-party CDNs from loading
#Loqi[Peter Molnar] How to install microG an odexed stock android ROM
#petermolnarbut I'm having weird issues with it, namely with location precision and even more weirdly with gms connectivity, however, I'm not convinced that last has anything to do with it
#[snarfed]@Zegnat bridgy github actually does backfeed replies to your replies (technically replies "after" your replies, yes) as well as replies to your issues. eg https://publog.stuifzandapp.com/posts/589
#Loqi[Peter Stuifzand] It is possible to create extra activities for "like", however I noticed that not all apps supports this. I think Twitter will only shows different apps, and not the specific actions.
#[snarfed]not sure why that one also got backfed itself, but that's a separate question
[matthilt] joined the channel
#Zegnatdo you also get mention backfeeds? E.g. because I mentioned `@tantek`, do you send a mention to his homepage?
#petermolnarI can't believe rss vs atom is still a thing
#kaushalmodipetermolnar: So I support both; users can pick whichever they like :P
#petermolnarkaushalmodi: you should publish rss 0.9, rss1, rss2, atom, json, rss3, activitystreams, mf2feed (did I leave anything out?), so you'd make truly everyone happy...
#LoqiIt looks like we don't have a page for "jf2feed" yet. Would you like to create it? (Or just say "jf2feed is ____", a sentence describing the term)
tantek, myfreeweb, snarfed, loicm, plindner, eli_oat[m], grantcodes, zoglesby, schmarty, mindB and peterlawson[m] joined the channel
#Zegnatdgold, did you do the Instagram download? Are those “comments, likes” you mention on the wiki comments and likes you have left on other photos, or also comments and likes others left on your photos? Big distinction, possibly worth elaborating.
#dgoldZegnat: I don't know how to answer that - I never liked or commented on anyone elses' posts, so the file is empty
maingo joined the channel
#dgoldoh, wait, other people did like mine, so I guess that means that its _your_ comments/likes
#ZegnatThat’s too bad, though perhaps expected as those comments/likes aren’t “yours”
#aaronpkuh, the likes.json file leaves something to be desired
#aaronpkit's just a list of timestamps and usernames
#aaronpkcomments.json is similar. looks like it is the timestamp of the photo i'm commenting on, my comment, and the username of the person whose photo i'm commenting on
#aaronpkthere is enough data to make an app that would parse those files and send a bunch of micropub requests to your site to import all your history tho
#aaronpkfor your photos/videos anyway. not your comments
#aaronpksomeone please make a desktop app that does that :)
#aaronpkshould be doable with electron or whatever
singpolyma, KartikPrabhu, tantek, sebsel, tomasparks and cweiske joined the channel
#Loqicweiske: kaushalmodi left you a message 3 hours, 59 minutes ago: Is there a way to extract comments specific to just my domain from commentpara.de?
#singpolymaheh. I realise this is not a new question. happy to read wiki pages or past discussions :) reading over jf2 spec now
#Zegnatjf2 tries to normalise certain behaviour for easier consumption. “mf2 json” isn’t really a format, just a name for the canonical serialisation of mf2 parser output.
#ZegnatAt least that is how I would describe the two.
#singpolyma"normalise" as in, remove all the array wrappers mostly? and also chop off "h-" looks like
#Loqijf2 is a working prototype of a simpler JSON serialization of microformats2 intended to be easier to consume than the standard Microformats JSON representation https://indieweb.org/jf2
#ZegnatIt also defines a stricter vocabulary than mf2, I think?
#LoqiIt looks like we don't have a page for "microformats2 JSON" yet. Would you like to create it? (Or just say "microformats2 JSON is ____", a sentence describing the term)
#Zegnatmicroformats2 JSON is the canonical output format of the microformats2 parsing algorithm. As such it can be used to compare parsers and create test suites. It is also used as the official serialisation format of microformats objects, and relied upon by specifications such as Micropub. http://microformats.org/wiki/microformats2-json
#ZegnatNow I remember part of why I hadn’t finished the one-php-file token endpoint companion to selfauth: I didn’t want to write the code that does the auth endpoint discovery.
#singpolymasnarfed: in bridgy you have a p-numeric-id property -- I assume that's just a thing you made up? Should it be p-x-numeric-id? (I don't care that much, just trying to understand)
#snarfedsingpolyma: nah it's the underlying silo's numeric user id for a user
#Loqisnarfed has 59 karma in this channel (381 overall)
#singpolymaSo, I feel the ergonomics of jf2, but I think micropub using mf2json makes it enough of a real thing to tip me that way for now. I mean, my experiment will probably go no where and then why did I do all this thinking anyway? ;)
#singpolymasnarfed: Interesting... can bridgy only generate /post/twitter/* pages for people who have auth'd? Even if the tweet is public?
#snarfedsingpolyma: you can put any tweet id at the end of the url
#snarfedbut yes it needs an authed user's username in the url to know which token to use
#snarfedor try https://granary.io/ if you want a generic service for reading silo data in mf2 (or other formats)
#singpolymawould just any token work most of the time?
#singpolymaI'm working on building a thing that can do twitter/mastodon-style "show the whole conversation" for distributed conversations across silos/fediverse/indieweb
#singpolymamy plan is to start by using it as the webmention receiver for my site (so just conversations my site is in) but then extend from there
#singpolymaso, it'll be a weird degenerate webmention where I ignore `&target=` entirely and just parse the source looking for in-reply-to, etc, myself
#aaronpkZegnat: if you're writing a token endpoint for selfauth you don't need to do endpoint discovery
#ZegnatNot integrated in selfauth in any way, should work with anything, is the idea
#singpolymaprobably most times I see a parent post that is a silo it'll be someone who uses bridgy anyway, but I'd like to get the whole chain even if they POSSE manually
#aaronpkAh hm. tokens.indieauth.com isn't one file but it's pretty small. Still I'm not sure how may more of that kind of thing needs to exist. Ultimately thetoken endpoint and auth endpoint should be closely linked
#ZegnatI still feel like the link between e.g. the micropub endpoint and the token endpoint is more tight than to the auth endpoint.
#ZegnatI did snarfed. I also wrote a pretty need little function to get the q parameters off of conneg ;)
#aaronpkIn any case you don't need to do endpoint discovery if you just provide a config option for it
#ZegnatBut link headers and their variable amount of parameters, and possibility to have multiple rels defined, the regex just isn’t flowing tonight, hehe
#aaronpkthe reason mine does discovery is because I host it as a service that doesn't require registration
#aaronpkAlso I wrote all those regexes for you with tests already :-P
#ZegnatI probably should have looked around, aaronpk, hehe
#ZegnatI actually think I wrote a (partial?) regex for the javascript mp helper too. :facepalm:
#aaronpkBut yeah just leave out discovery for this if you're making something you expect people to install and set up
#ZegnatMostly doing this because I want to preempt the first issue being “how do I run this for multiple domains” ... as we got on selfauth now
#aaronpkThe flip side to this question is how do you avoid people using your instance of this endpoint
#ZegnatThat is going to be a whitelist of endpoints.
#aaronpkRight so just define the auth endpoints in that whitelist
#ZegnatSo if a client sends a $me (which may even be pre-redirects etc, hard to whitelist, maybe?), I discover the endpoint. If the endpoint for $me is on the whitelist for the token endpoint, I go and check the code.
#skippyhey, the less i can rely on rando services hosted by other people, the better.
#skippyoh, i just saw the XPath stuff you're doing. Sorry. Didn't mean to send you down THAT rabbit hole!
#ZegnatThe XPath in that gist was some of the quickest code to write. I am more than familiar with DOM at this point.
#ZegnatI have taken aaronpk’s implementation of the relative URL resolver with me for bed time reading now ;)
#skippyi think i asked this before, and I dont think I fully understood the answers: if I send a "repost" to my Micropub endpoint, what is the "normal" or common behavior? Does the endpoint fetch the content of the original for display on my side? do I manually copy-paste the content and send that along with the repost-of URL?
#ZegnatI think the few people who have implemented reposts post it the same way as they do a like: just the URL. Then the server fetches context to display. But not sure.
#pstuifzandYeah, just the repost-of url and micropub endpoint should fetch and parse the content
#aaronpkyep that's what I do. the client sends just the URL and leaves it up to the server to decide how to handle that
#aaronpkfor a long time I was showing reposts and likes on my site as just the URL even. once I finished XRay then I made my micropub endpoint go and expand those posts to show the reposted/liked contetn
#Zegnatskippy, if you use the German Firefox Klar you get better privacy defaults (unless they changed that). In case your choice for Focus is because of privacy.
#aaronpkI was paying attention to the indieauth plugin not webmention
#GWGaaronpk: I did the REST API thing around the Summit in 2016
#pstuifzandI doesn't seem to be a problem with the query parameters of the url. I wrote a test to check it, and it comes out as expected.
#GWGaaronpk: After Aaron Jorbin came to IWC NYC that year, we were talking and he was explaining to me you can use the endpoint creation functions of the REST API to create other types of APIs
#pstuifzandGWG: Do you have an example of the content?
#snarfedagreed! and lots other higher priority things in wordpress land. debug micropub "no location header" problem, lean on pfefferle to merge wm and semantic linkbacks, figure out how to integrate gutenberg blocks with https://github.com/snarfed/wordpress-micropub/issues/85 , etc
#Loqi[dshanske] #85 redesign to coordinate with other mf2 plugins and themes
#snarfed(and that's just the small subset i'm familiar with)
#GWGsnarfed: I think the spec in Gutenberg is nearly settled down. but I really didn't want to do JS. Oh well. As for leaning on pfefferle...I don't want to lean on anyone...
#snarfedand figuring out micropub redesign vis a vis blocks shouldn't require any JS iiuc
#GWGsnarfed: My next project is improving Syndication using that hook we put in, and putting in a platform to syndicate that I can hook 'providers' into easily
#snarfedanyway. should probably be in #indieweb-wordpress
#GWGaaronpk: So, are we scaring people off by talking dev in there?
#aaronpkso -dev ends up being where the plugin development talk happens
tantek joined the channel; GWG left the channel
#aaronpkI don't know, but I feel like i've seen an uptick in wordpress user chat there, from people who aren't involved in developing the plugins, rather using them
#pstuifzandit's a custom scheme to make android call the right activity when the browser redirects to that url
#GWGSo, a few things there. Your SSL certificate is for peterstuizand.nl,
#pstuifzandI didn't expect the client_id to used in a GET request
#GWG"The client SHOULD publish one or more <link> tags or Link HTTP headers with a rel attribute of redirect_uri at the client_id URL."
#pstuifzandThat seems to be the problem. I need to figure what I'm going to do there.
#GWGaaronpk: You wrote the spec. I have the endpoint checking to ensure the redirect_uri is on the same domain, and rejecting it if it is not. A future feature I want to put in is checking for alternative redirect_uris
#pstuifzandI do have the wrimini.net domain now, so probably that will become the client_id in the future
#GWGIt shouldn't allow this to avoid arbitrary redirects.
#aaronpkGWG: what i've done so far to avoid the strict checking is to show a warning to the user on the authorization screen that says that the place they'll be redirected to may not be safe
#GWGaaronpk: But would you agree, either way, your client_id should be a valid website?
#aaronpkthe plan is eventually the authorization server will fetch the client ID URL and check that the redirect_uri is registered there, but in the mean time (since the only user of this is myself), the warning suffices
#GWGpstuifzand: If you have a valid certificate, yes.
#GWGBut it would still run afoul of my code checking in compliance with the specification.
#aaronpkyeah, ideally the server fetches the client_id to check the redirect_uri. this is the mechanism that takes the place of pre-registration of apps.
#pstuifzandSo even if the domain is the same, it should also send Link headers?
#GWGpstuifzand: So, all you have to do is publish one or more <link> tags or Link HTTP headers with a rel attribute of redirect_uri at the client_id URL.
#aaronpkbut if the client_id and redirect_uri is the same, then it's safe to redirect to without fetching the URL
#aaronpkat some point i'd like to start promoting the idea of using the PKCE extension for native apps, but that requires more work on both the client and server
#GWGaaronpk: I read about that in your book. That's the code thing?
#aaronpkit's an extra step the client does before beginning
#GWGaaronpk: Wait, I was thinking of device flow. Never mind
#aaronpkthat would be fun too, but has more limited uses right now :)
#GWGaaronpk, what would be the use case for device flow?