#LoqiKevinmarks has 6 karma in this channel over the last year (29 in all channels)
renem, yaymuffins15, snarfed, tantek__, [tantek], thekingofbandit1, tantek and [jgarber] joined the channel
#[jgarber]Zegnat: Thanks for bringing up the <base> issue. I don’t remember that getting much traction earlier in the week.
#[jgarber]Looking forward to that test making its way to webmention.rocks. I’ll have some code changes to make, but the Ruby lib will be better off for the extra tests. 👏
[chrisaldrich] joined the channel
#loqi.mecreated /DiSo (+25) "prompted by [chrisaldrich] and redirect added by tantek__" (view diff)
#snarfedok. mind explaining in the method docstring? something like, "this method checks the token if provided, but if not, it passes through $user_id as is, since it's a filter." or whatever language you want
aphex`, [kevinmarks] and [pfefferle] joined the channel
#[pfefferle]GWG aaronpk shouldn‘t we implement a simple class to parse h-App? I think it is a very simple and uncomplicated format so it would be a bit too much to include the mf2 parser!?!
#sknebelthere is no "simple format" vs "complex format" from a parsing perspective
#sknebelthe only "complex" thing you can maybe ignore is microformats1
#aaronpkYeah you'll end up creating lots of fun new edge case errors if you try to parse a microformat with anything other than a real microformats parser
#ZegnatMuch like trying to parse HTML with a non-HTML parser
#GWGI was going to start with finding icon and title without microformats
#aaronpkIf you're taking about parsing the html and looking for just those classes you might as well do the whole microformats parsing step
#aaronpkif you're talking about trying to parse it without an html parser... well that is pretty well understood to be a bad idea
#ZegnatYou could have a smaller mf2 parser by assuming the objects you are parsing will not make use of certain things. E.g. h-app authors will never use vcp, and will only use p- and u- properties. But that breaks as soon as someone does uses anything outside of your limitations.
#GWGI just submitted a PR for logging every Micropub error to the error_log if debugging is on. Going to do the same in IndieAuth when I return from work.
#GWGHopefully will be able to track down my problems
#aaronpkI don't love the idea, but the other option here is to use meta tags like the Twitter/FB tags. If too many people building this stuff end up getting turned off of the idea that their authorization server needs to parse microformats then it's worth considering tho
#Zegnatrel-icon exists if it is just about the logo.
#[pfefferle]But I would prefer the simple direct parser way and change to the mf2 parser if it gets more complicated, instead of the other way around
#aaronpkYou'll just end up getting confused support requests if you don't do it right from the beginning
#[pfefferle]Because there are some plugins that use the mf2 parser atm. And the version might differ, so we really can get some strange side effects if the indieauth plugin includes the wrong lib version
#ZegnatRight, I guess there isn’t some sort of dependency management for WP to make sure all the plugins only ever grab one specific version of the mf2 parser
#aaronpkZegnat: want to document what it would look like to use existing html tags to get the same information as h-app?
#sknebelone could embed the parser directly into the plugins, with customzied name etc so it doesn't get loaded accross plugins
#sknebelalthough I'd be curious about the weird side-effects, the parser API should(TM) be stable
#sknebelmh, lang-parsing etc might change and/or would need a strict feature-flag policy to be fully compatible I guess
#Zegnataaronpk, it only requires name/logo/url, right? Name is tricky, could be <title> if we are only talking about HTML. logo and url could be rel-icon and rel-self, probably. Which means they would work in HTML <link> elements, in any XML flavour that does link (atom:link), as well as HTTP Link headers.
#[pfefferle]Perhaps we can easily change the namespace for the class to make it more plugin dependent
#[pfefferle]Will have a look if this is easily doable
#aaronpkZegnat: we do only care about HTML since this is always on a human readable page
#aaronpk<title> might work but sometimes people put a lot of stuff in the title like a small description or tag line, so that could look weird on the authorization screen
#ZegnatDo browsers still do XML with XLST? TXT is also human readable. Though I guess the spec does ask for an HTML page so it is not really a concern.
#aaronpkFirefox dropped XSLT at some point I think? Or was that safari
#GWGI think a fallback to icon and meta might work if h-app isn't available or parseable
#ZegnatOfficially only with namespaces, right, [pfefferle]? So we would have to introduce a new app namespace for OGP? I am not sure what kind of upkeep that requires and might be a bit much to tack onto the IndieAuth spec.
#ZegnatOh. Brilliant. Facebook put the namespace pages behind their API. Can’t access it without a proper token *facepalm*
#[pfefferle]It is no longer called namespace in RDFa, but prefix/profile, but yes
#aaronpkThat requires another http request which seems unnecessary for this
#ZegnatIf you want embedded structured data inside of HTML, that is relatively easy to parse with minimal code, OGP or JSON-LD sound like the ways to go
#aaronpkCan someone go through a handful of Micropub apps web pages and document all the places that they currently advertise the name/logo?
#aaronpkThen we can see if the apps all already agree on some html meta tags or something
#ZegnatWould it be worth opening an issue for this against the indieauth spec? To track this? I’ll be happy to do a little discovery after lunch
acegiak, connection and koddsson joined the channel
#ludovicchabantCool, looking at the chat log I learned about cleverdevil’s ditchbook! Might try it with Twitter (I don’t want to bring Facebook stuff to my website since those were semi-private while my website is public)
#ludovicchabantCool, looking at the chat log I learned about cleverdevil’s ditchbook! Might try it with Twitter (I don’t want to bring Facebook stuff to my website since those were semi-private while my website is public)
ChasedSpade joined the channel
#ludovicchabantWtf, did my message get posted several times?
#Loqi[Zegnat] #23 Client Information Discovery without relying on microformats parsers.
#ZegnatI think I captured the problem with a boiled down parser for h-app, the problem with requiring mf2 parsers, and the list of possible alternatives mentioned in the chat
#LoqiZegnat has 53 karma in this channel over the last year (140 in all channels)
#ZegnatI ran into <meta name="application-name"> as a much better alternative to <title> for app name. And now I am kinda loving the standard-HTML alternative
#aaronpkyeah that option seems to be the lightest weight, fewest new things approach
#ZegnatIt likely introduces hidden meta data, which is something we try to counter with microformats usually.
#ZegnatE.g. you can’t slep a rel="icon" (or similar) on an <img> for your logo
#[pfefferle]> This specification defines a JSON-based manifest file that provides developers with a centralized place to put metadata associated with a web application.
#LoqiIt looks like we don't have a page for "difference" yet. Would you like to create it? (Or just say "difference is ____", a sentence describing the term)
#aaronpkwebapp manifest is for PWAs, but the apps we're talking about are not necessarily PWAs
#aaronpke.g. it would be kind of weird for indigenous to publish a webapp manifest since it is a native ios and android app
#ZegnatUrg, I just read the HTML spec also limited the use of <meta title="application-name"> to “Web application”. Though it doesn’t define what it means by that.
#[pfefferle]aaronpk sure, but it does not have to be a pwa, at least is the spec not saying it, but I understand the point with the extra request... still seems to be the „cleanest“ way for me (does that make sense in English?)
#aaronpkyeah, except you already have to parse the html both to find the link to the app manifest plus also to get the redirect_uri
#aaronpkso it seems easier to at that point just look for a few extra meta tags
#ZegnatI would just wish we didn’t need to invent the meta tags
#[pfefferle]Me too, but searching for an alternative ends always on the web apps manifest page ;)
#ZegnatJSON-LD would let us use an existing vocabulary (schema.org/SoftwareApplication), but I don’t think anyone really wants to exchange the mf2 parser with a JSON-LD parser
#ZegnatI am going through possible <meta> names (https://wiki.whatwg.org/wiki/MetaExtensions) and spotted this one for the category of things I did not know about: <meta name="pinterest" content="nopin"/>
[kevinmarks] joined the channel
#[kevinmarks]I'm going to be indieweb clippy here: it looks like you're trying to invent a microformat. Would you like me to point you at microformats.org/wiki/process ?
#aaronpkdidn't i say something about please document all existing examples of markup that micropub apps are publishing
#[kevinmarks]You seem to be doing the research, but wikify it please, that helps a lot in converging
#ZegnatI am just not sure how this collection is going to show us anything, apart from which tools include h-app (like Quill) and which do not (like Telegraph and shpub)
#aaronpki meant also documenting their use of <title> and other meta tags
#ZegnatSo far not seen a lot of ogp/twitter/meta-data at all. But I’ll have another pass over that. And put my work somewhere public.
#ZegnatLots of early implementations coming out of IndieWeb people, who may have a bit of a bias against those extra meta data elements, is probably not going to help ;)
#ZegnatAlright. Looks like h-x-app outnumbers h-app easily, and there are way less explicit redirect_uri values than I expected.
#ZegnatNow that I have a list of client_id’s, I can see what other metadata they have in common. I have seen multiple rel-icon and also a couple of apple touch icons. Few rel-manifest too.
#aaronpkDropping h-app would not be the worst thing
#tantek__aaronpk - yes if it is done halfway it fails, that's the point
#tantek__the point about using -x- is to allow more rapid active experimentation and following up with formalizing if it works
#GWGI will implement icon searching soon, as most pages have an icon for other reasons
#aaronpkmy next goal with IndieAuth is to get just the client ID part written up as an i-d and proposed to the OAuth WG
#aaronpkand I’m a bit apprehensive of including microformats parsing in that proposal
#tantek__aaronpk yeah it makes sense to keep it minimal
#aaronpksomehow I suspect most of my target audience for this is not going to go for that, as demonstrated by this first occurrence in the Wordpress plugin
#tantek__aaronpk, or rather it demonstrates a good place for a modularity boundary
#aaronpkit’s enoguh of a mental leap to go from “you must register for an api key” to “you don’t need to register because everything is a url” and I don’t want to also throw in microformats into the mix
#tantek__so the "most of my target audience" can deal with just the one piece, and the folks that want the additional functionality can do so on top
#[eddie]Just reading about the h-app, h-x-app stuff. I do have to say, while I’ve found Microformats to be extremely helpful in many cases, mf2 in the IndieAuth process is the biggest friction I’ve ever found. I wouldn’t be opposed to there being a different way to find client info from a url
#[cleverdevil]I put h-app on indiepaper, and it was pretty easy.
#[eddie]Easy to add mf2 to an app’s page. But when writing IndieAuth client code, you either have to add an mf2 library just for that one step or roll your own just for the h-app stage which typically ends up buggy
#[eddie]It’s a good solution, I just mean that if there were a better solution, I would be all for it
#tantek__point being, re-using web manifest likely better than trying to reinvent that wheel, also a good defense against others reinventing that wheel
#[eddie]I do think that would be easier from my experience of working on my Auth Server
#sknebelas discussed above, kind of weird to use for non-web-apps though
#tantek__then we could backport that to an "h-manifest" if we wanted to provide a way to put it all on a page
#tantek__sknebel works fine for icon + name though right?
#[kevinmarks]I don't think it will trigger a prompt without a service worker, but it will be used if you add it manually
#[kevinmarks]Which is a natural extension of "Tantek on my homescreen"
#tantek__kevinmarks "trigger a prompt" is very implementation specific
#Zegnat[eddie]: thanks for the update to the table. I thought I pulled the client IDs correctly straight from the GitHub repos... Don't actually have any of the clients installed to test.
#LoqiIt looks like we don't have a page for "Zero Registration" yet. Would you like to create it? (Or just say "Zero Registration is ____", a sentence describing the term)
#loqi.mecreated /ZoRg (+30) "prompted by tantek__ and redirect added by tantek__" (view diff)
#[eddie]Zegnat: the client ids started as the root before I realized that ios and android needed different client ids, so I might not have the correct info everywhere
#[eddie]If anyone has android and wants to see if Indigenous website with a basic webmanifest creates any strange prompts for “installing the web app”, I’ve added a manifest to the homepage: https://indigenous.abode.pub/ios/
#[eddie]I put in name, icons, prefer_related_applications and related_applications. The last two aren’t required obviously, but helpful if browsers decide to use that to prompt for App Store links.
#[eddie]So tentatively seems like a viable consideration, tantek and aaronpk
#[eddie][manton] Out of curiosity, if I remember correctly Micro.blog did not add the Client Information Discovery step into it’s IndieAuth (https://indieauth.spec.indieweb.org/#client-information-discovery) was that because of the complexity or just for lack of time? If the Client Information Discovery was a JSON file you could grab the url and parse, would you have been more likely to add it or would you still have delayed that implementation?
cottongin14 and [jgmac1106] joined the channel
#tantek__eddie, if we can get some adoption of indieauth+webmanifest working, then aaronpk can add it to ZoRg
#tantek__(just going to keep saying ZoRg like it's already a thing until it is)
#tantek__aaronpk, just practice saying it with straight face and you'll be good
#[eddie]Yeah, I figure before working on getting it working, it would be helpful to get feedback from someone like Manton that seems interested in IndieAuth but that didn’t implement the mf2 discovery
#tantek__eddie, good point. a quick impression from Manton woudl be helpful
#Zegnatgoes to add rel-manifest info to the client information table
#ZegnatFWIW, I can totally see me implementing manifest data in Selfauth, but not bringing in an mf2 parser
#tantek__also if y'all have opinions about webmanifest please do share - I'm tangentially involved with that stuff at w3c and would appreciate improving it
#tantek__(i.e. while you're implementing webmanifest if thing seem difficult or weird)
#Zegnattantek__, do you know what the official discovery process is for finding the manifest? I only recall some non-normative section about <link> elements.
#ZegnatYeah, that’s the only part I saw in the spec. And was a little surprised that discovery was only covered by a single non-normative example in a non-normative “Usage Examples” chapter
#Zegnat7/21 client pages have h-x-app. 7/21 also advertise a manifest file...
[manton] joined the channel
#[manton][eddie] Just haven’t gotten to it yet, and no apps I tested with needed it.
#ZegnatSo in the popularity contest, both are equal. Wow
#ZegnatAlthough not all manifest files might be correct. I haven’t checked the contents on all of them.
tantek__ joined the channel
#[eddie]Gotcha. Thanks for the response [manton] :thumbsup:
#[eddie]That is interesting, isn’t it, Zegnat! Like you said, they might not all be valid but its telling that there are already that many manifests available.
#aaronpk[manton]: which version of this sounds better for you to implement? Parsing for h-app microformats or making another http request to the app manifest and parsing that JSON?
#tantek__eddie, good point about not all might be valid. good opportunity to validate/lint h-app vs manifest?
#tantek__Zegnat yes that HTML spec for meta application-name predates webmanifest AFAIK, and note that web manifest uses it as fallback
#Zegnat5 of the manifest pages do not have h-app, so comparing is out, I’m afraid. But from what I have seen, h-app is usually correct, because the people who added it followed aaronpk’s examples to the tee
#tantek__zegnat so what of the two that have both? do they match?
#tantek__I'd be interested in figuring out if we should either keep h-app or make an h-manifest that just uses the same property names as webmanifest, and either way, create a generator so publishers could auto-generate their webmanifest JSON from the h-app / h-manifest
#tantek__if we can reduce side-file maintenance, we remove a source of errors
#ZegnatThe Monocle manifest does not match, doesn’t include a name.
#ZegnatBut this may be another accidental manifest added by aaronpk without meaning to
#tantek__but on the consuming side, if devs prefer to do the 2nd http request to get the JSON, then it makes sense to standardize on that as the interface
#aaronpki will admit, i had no idea i was adding a web manifest to my apps
#tantek__so publishers can follow aaronpk's instructions and do the simple h-app / h-manifest thing, and consuming code can get their JSON
#ZegnatThe other was Indigenous, and there the name and logo match between h-x-app and manifest
ludovicchabant and renem_ joined the channel; ludovicchabant left the channel
#[eddie]Zegnat: Good question. Right now it’s just a static .webmanifest file. I guess I need to tell my nginx that .webmanifest files are application/manifest+json
#Loqi[Zegnat] The table above has been updated with new statistics. This shows that `rel="manifest"` are equally as likely to be found on client pages as `h-x-app`. This Web App Manifests may be a more logical alternative than previously realised.
Further devel...
#ZegnatMostly because I myself find it easier to look at the github issue than go back and search through who said what on IRC
[jgmac1106], Guest43602, renem, sebsel, [pfefferle] and James_T0 joined the channel
#ZegnatAdded a few more data points to the table. If anyone knows more places that will let you login with IndieAuth (to do anything!) please add them to the wiki table. Even if you don’t have time to document the rest, someone else can do that. Just collecting all the different client_id values is valuable!
#ZegnatNow I am calling it a night. Thanks for all the discussion everyone
#@wp2staticWith static sites allowing free and diverse hosting options and #webmentions as cross ecosystem pingbacks, it’s time to dust-off your old blog/personal site, comment out the #dancingbabies and own your own home on the net again. Web 1.0 Hyper Edition! (twitter.com/_/status/1031289316608172032)
[stefp], snarfed, Lildirt and bradenslen joined the channel
#sknebelhas anyone gone through the new twitter developer registration process?
#sknebel(mostly: do they approve random personal use cases?)
#snarfedsknebel: i've had app review(s) pending for...months now i think
#Loqisnarfed: Zegnat left you a message 12 hours, 31 minutes ago: No, my endpoint discovery code does not have integrated caching. That'll be up to the lib user.