#dev 2019-11-01

2019-11-01 UTC
#
[jgmac1106] Was thinking the h-entry on the home feed that was really just a link to articles page could be a way to show new subscribers all my stuff... Will just keep recent article feed
[qubyte] joined the channel
#
aaronpk [grantcodes] I'm getting a weird ssl error on https://photo.postrchild.com/
[KevinMarks] joined the channel
#
[KevinMarks] This might also be a case where a p-summary explaining that its a link roundup post could work, assuming that the webmention receiver builds on the name/summary/content split
#
aaronpk guess what, it helps when you spell "visibility" right
#
aaronpk not "visibiltiy"
#
aaronpk :facepalm:
#
aaronpk that was like 15 minutes of debugging wondering why this wasn't working
#
aaronpk aaand my posts from the last 2 days are un-privated
#
[tantek] oooh
#
aaronpk definitely want to re-think that feature having gone through this
#
aaronpk I was able to hack it quickly thanks to ownyourswarm's "private" checkin mode
#
aaronpk but that means I had to go back and un-private each post
#
[tantek] pretty sure we have a brainstormed feature for that
#
[tantek] what is embargo
#
Loqi embargo is, in the context of the IndieWeb, the publishing practice of requiring exclusive rights to publish an article by an author on a publisher’s website, some amount of time (typically days, sometimes months) before the author publishes it on their own personal site https://indieweb.org/embargo
#
aaronpk really what I wanted was to delay all my feeds, but still have the posts public so they'd be sending webmentions and such
#
[tantek] that sounds like emphemeral unlisted posts
#
aaronpk then with one setting I could un-delay the feeds and the posts would all appear as normal all at once
#
aaronpk kind of, but again it's not a per-post setting that I want
#
aaronpk I basically want a giant "pause" button for my website, where it keeps storing posts I create, but if you look at my home page or any feeds, it looks like I haven't posted
#
[tantek] that does sound like a new feature / concept
mblaney joined the channel
#
mblaney aaronpk my posts get created with a feed attribute, so I could imagine doing something like what you're suggesting by leaving that field blank and then filling it in when I want them to show up.
maxwell, gRegorLove and [jgmac1106] joined the channel
#
[jgmac1106] Is there any h-card markup for multiple last names or changing names for use cases like: https://twitter.com/Arley_McNeney/status/1190120444805840896?s=20
#
@Arley_McNeney When I got married, I thought I'd keep publishing under Arley McNeney and then just do everything else as Arley Cruthers. But now I teach as Arley Cruthers, and have gotten into academic writing where I have to state my university affiliation. Which last name to use? (twitter.com/_/status/1190120444805840896)
gRegorLove, KartikPrabhu, cweiske, [Lewis_Cowles], krychu and [fluffy] joined the channel
#
[fluffy] so I notice that now on publ.beesbuzz.biz Firefox seems to keep on rendering pages without the stylesheet having been loaded yet, which seems really weird. Anyone have any ideas what’s going on with that?
#
jamietanna[m] <[fluffy] "so I notice that now on publ.be"> [fluffy] have you checked the network tab / console to see if it's actually loading the stylesheet?
#
[fluffy] It does load, it just takes a little bit to do and the page shows as unstyled while it waits
#
[fluffy] It’s not a behavior I’ve seen in a long time
#
[fluffy] Once it’s cached it stays loaded but because of how my templates are set up it needs to be cached separately for each directory
#
[fluffy] That said it’d be easy for me to make that site not work that way. But my main site does work that way and I haven’t noticed that issue happening there.
#
[fluffy] But my main site also runs on dedicated hosting while the Publ subdomain is on Heroku, just because I want to have an active site that follows that use case.
#
[fluffy] And that might affect the timing or something to do with how the cache domain works or something.
#
[fluffy] It’s just weird to me that the browser isn’t waiting to retrieve the stylesheet needle or starts rendering the page.
#
[fluffy] *before it starts wtf autocarrot
#
[fluffy] Hmm looks like someone else is having a similar problem that also started recently: https://stackoverflow.com/questions/58254161/firefox-renders-html-before-applying-css so I’m suspecting Firefox bug
#
[fluffy] Ok whatever I’m going to bed
[jeremycherfas], strugee, IWSlackGateway and [qubyte] joined the channel
#
[qubyte] (this question is not leading to criticism of any kind, it’s pure curiosity 🙂 )
[jgmac1106] joined the channel
#
aaronpk because multipart is the simplest way to upload a file
#
Loqi aaronpk: [grantcodes] left you a message 2 hours, 26 minutes ago: yeah just use the non Https version for now. I moved it to heroku without knowing I would need to pay extra for https
#
aaronpk anything else is making stuff up from scratch and would require more work to write code to receive it
#
aaronpk most web frameworks handle multipart uploads because that's how <input type="file"> works in HTML
#
GWG aaronpk: Sent you the location tracking data for WN
#
aaronpk nice just saw that
[rem] joined the channel
#
[qubyte] 50:50 on most web frameworks. In Node, Rust, and possibly others it’s a bit of a pain (though not that bad in the grand scheme). A plain file upload in Node for example looks like (from a request)
#
[qubyte] const chunks = [
#
[qubyte] ```
#
[qubyte] meh
#
[qubyte] I can’t use a keyboard.
#
[qubyte] No matter. I see where the spec is coming from now 👍
#
aaronpk I would say that's the fault of node and rust for not having a good web framework
#
[qubyte] For the latter, fair (it’s early days). For the former, there are a number, but Node emphasises streaming, and multipart makes for complicated streams.
#
[qubyte] There are modules though, and I’m using one (so yup, no biggie).
#
[qubyte] I’m actually more interested in possible extensibility. Could some future version of the media endpoint accept more than one file in an upload? I suppose that might add constraints on the field names. 🤔
#
[qubyte] Anyway, thanks for the info. It’s nice to be able to read a spec _and_ ask questions about it!
#
aaronpk what's the use case for accepting multiple files?
#
aaronpk it's modeled after Twitter's media upload API https://developer.twitter.com/en/docs/media/upload-media/overview
#
aaronpk so if anything, the only change I would want to make is to support a chunked upload method like twitter has
#
aaronpk the media endpoint is really meant to be just a temporary holding cell for media to improve the UX of micropub apps, nothing fancy
#
[qubyte] Yup
#
[jgmac1106] anyone here cross post from micro.blog to Twitter, I can't figure out what I am doing wrong in setting it up
#
[qubyte] I get keeping this uncomplicated. The only reason would be to send stuff in plural than a request per resource, but there’s little reason to do that.
#
[qubyte] As for chunked uploads… Happy not to have to accommodate that.
#
aaronpk yeah I don't think you're going to get any UX benefit trying to shove multiple files to it at once
#
[qubyte] yup
#
aaronpk chunked uploads on the other hand have a clear benefit of being able to resume an interrupted upload
#
[qubyte] I kinda dislike overloading the word “chunked”, but I see why.
#
[qubyte] It’s an attempt at symmetry.
#
aaronpk overloading?
#
[qubyte] See chunked transfer encoding (of responses).
#
[jgmac1106] Now I want a Baby Ruth.....ohhh wait Halloween baskets...probably lots of Baby Ruth in the house...(sorry for off topic interjection...goes for candy breakfast)
#
[qubyte] lol
#
aaronpk hm, seems pretty similar
#
aaronpk tho at a different layer of the stack
#
[qubyte] yeah
[KevinMarks] joined the channel
#
@TheRealNooshu Idea for a VSCode extension: live microformat previewer. Extension parses the currently edited page and pulls out the microformat data in real time. That would be super useful right now. Does it already exist? I can't seem to find it anywhere if it does. (twitter.com/_/status/1188937368197324802)
#
[KevinMarks] Nice idea https://twitter.com/TheRealNooshu/status/1188937368197324802?s=20
#
[qubyte] At the risk of inventing something new… It seems like you could use headers and PATCH to offset and overwrite/append things to a partial upload.
[LewisCowles] joined the channel
#
[LewisCowles] qubyte, aaronpk, why is a file upload happening and not a URL?
#
[LewisCowles] is it a PESOS vs POSSE thing?
#
aaronpk what do you mean?
#
aaronpk this has nothing to do with syndication
#
[LewisCowles] Oh so known needs to receive a file which it posts to both?
#
[LewisCowles] > Known->micro.blog->Twitter
#
aaronpk this is just about creating a post on your own site, nothing to do with syndicating to micro.blog or twitter
#
[LewisCowles] I see I’ve blended chat logs.
#
[LewisCowles] I see
#
[qubyte] Ah sorry. We’re chatting over each other.
#
aaronpk see this section for a high-level description https://www.w3.org/TR/micropub/#media-endpoint
#
[LewisCowles] Yeah, I was pointing at https://www.w3.org/TR/micropub/#uploading-a-photo-with-alt-text
#
[LewisCowles] but I must be confused
#
aaronpk that's the other reason for it, to be able to include photos in JSON micropub requests
#
GWG Trying to build a compass request in JavaScript inside Android's tasker app
#
@tomcritchlow Today's project: gonna finally see if I can wrestle webmentions to the ground. Never been able to get it working properly but I like the idea of rolling my own glitch webmentions server like this: https://qubyte.codes/blog/a-new-service-to-handle-webmention-dispatch-for-you (twitter.com/_/status/1190253123295744000)
#
[qubyte] The reason I use a media endpoint is for the alt text (JSON on the regular endpoint) yeah.
#
[jgmac1106] sweet it worked. I can go from Known->micro.blog->Twitter https://twitter.com/jgmac1106/status/1190247437342035969
#
@jgmac1106 Doing a test trying to go from Known ->@microdotblog->Twitter rather than just Known->Twitter and syndicating only one tag to microdotblog. (twitter.com/_/status/1190247437342035969)
ketudb joined the channel
#
aaronpk making progress on my photo album workflow...
#
@tomcritchlow ↩️ ha - I feel like that's webmentions in a nutshell... So close to the right idea but just too damn hard to get it working.... Will report back (twitter.com/_/status/1190267014104977408)
KartikPrabhu joined the channel
#
GWG Bingo
#
GWG Southwest wifi to tasker to compass
#
aaronpk Nice
#
GWG As long as there is wifi
#
GWG Wish there was an app, but this will do
#
GWG I would have done it faster if I was familiar with Tasker and not trying to type on a phone
#
@qubyte ↩️ Your mileage may vary. In the end I made a new one based around the code I mention in that blog post. It's much more picky about microformats though. https://glitch.com/edit/#!/lean-send-webmentions (twitter.com/_/status/1190270288795189250)
#
@qubyte ↩️ @tomcritchlow That webmention server is how I discovered your tweet btw (via http://Brid.gy). (twitter.com/_/status/1190270685328945153)
#
@qubyte ↩️ @tomcritchlow That webmention server is how I discovered your tweet btw (via http://Brid.gy). (twitter.com/_/status/1190270685328945153)
#
[qubyte] It wasn’t of course. I was conflating sending and receiving.
#
@tomcritchlow ↩️ Setting up Webmentions TL;DR: - Enable h-card markup - Add http://webmention.io support - Write custom JS to display http://webmention.io API - Sign up for http://Brid.gy for social webmentions - Sign up for http://webmention.app (twitter.com/_/status/1190275208495669248)
#
@tomcritchlow ↩️ Setting up Webmentions TL;DR: - Enable h-card markup - Add http://webmention.io support - Write custom JS to display http://webmention.io API - Sign up for http://Brid.gy for social webmentions - Sign up for http://webmention.app (twitter.com/_/status/1190275208495669248)
#
@tomcritchlow ↩️ (cont.) - Set up IFTTT for RSS -> http://webmention.app to send webmentions Fin? Maybe? I count 4 different web-apps necessary *and* two different code markups needed to get this all working correctly. Webmentions is a failed system. (twitter.com/_/status/1190275209900756992)
[frank] joined the channel
#
@tomcritchlow ↩️ I was *just* about to come back around and thank you and Remy for building these servies for starters. So thank you! Yes, this is a partial failure of static sites but it still just feels like an over-engineered solution. I know lots of bloggers and so few of them use webmention (twitter.com/_/status/1190276452517847040)
#
@jgmac1106 Disabled my crosspost to Twitter as I try to figure out how to add the microsub endpoint in Known. (twitter.com/_/status/1190277022867771393)
#
@tomcritchlow ↩️ @aaronpk Is it much easier for non-static sites? Is there a "webmentions in a box" WP plugin? (twitter.com/_/status/1190277327047012352)
#
@aaronpk ↩️ Yep! https://wordpress.org/plugins/webmention/ It's completely self-contained, no external services, and integrates with the native WordPress comments system. (twitter.com/_/status/1190277552704589824)
#
@tomcritchlow ↩️ Just to counter-balance the frustration - I want to give huge props to Remy for building http://webmention.app and Aaron for building http://webmention.io. Love the fact that there is a community building solutions here. (twitter.com/_/status/1190277717662601217)
[KevinMarks] and [xavierroy] joined the channel
#
@tomcritchlow ↩️ Ok! First webmention officially sent. Still had to post manually using http://webmention.app but hopefully will solve that shortly with a little RSS cron job (twitter.com/_/status/1190282236127514634)
#
GWG aaronpk, what was your display solution for posting while on an airplane?
#
aaronpk display solution?
#
[qubyte] RSS cron job is a nice approach. Similar vein: I check for changes to a sitemap.txt.
#
cweiske if you support websub, then the webmention sender could simply subscribe to your blog's feed
#
cweiske and automatically get notified when it updates
#
[qubyte] Yup, that’s the plan in the end.
#
GWG aaronpk: Weren't you improving it recently?
#
[qubyte] I have a few services using the sitemap, and it’d be better to move that ugly to one place. At some point I’ll probably implement a combined hub/publisher/
#
[qubyte] WebSub is a little strange though. The hashing is cool, but the payloads are not quite as I expected them to be.
#
[qubyte] That said, I don’t know what I was really expecting! I suppose JSON because I’m all about SOA and that’s not really the point here.
#
cweiske you can simply ignore the payload
#
cweiske and only see it as "the url has changed" signal
#
[qubyte] That’s true. Use it as a trigger hook instead.
#
aaronpk GWG: the only real offline posting interface I have is Teacup
#
GWG aaronpk, I meant... didn't you change the textual description of your location when posting on a plane?
#
aaronpk ohh
#
aaronpk yes, haven't touched that since building it
#
aaronpk I like it
#
[qubyte] One thing which stood out is the signature and header.
#
[qubyte] It looked familiar, and then I realised it was a generalisation of what GH uses, which I’ve worked with before. https://crates.io/crates/github_webhook_message_validator
#
aaronpk [qubyte]: lol werent you just complaining about multipart uploads? WebSub uploads the exact contents of whatever file is being subscribed to :)
#
aaronpk which means if you subscribed to a jpg URL then you'd get a jpg posted to your websub subscriber
#
[qubyte] But now I need to make a general version of that crate and I don’t remember rust.
#
[qubyte] hmm?
#
[qubyte] Oh really?
#
[qubyte] I must have misread. I interpreted it as sending RSSish snippets.
#
aaronpk were you reading superfeedr docs?
#
[qubyte] Sorry, don’t mean to sound complainey 😅
#
[qubyte] Probably
#
aaronpk that'd be why
#
[qubyte] Lemme bring up the spec.
#
aaronpk superfeedr does a bunch of stuff outside the spec
#
[qubyte] I’ll read the proper thing this time.
#
aaronpk https://www.w3.org/TR/websub/#content-distribution
#
cweiske afaik there was the idea to only send the changed parts of a feed to subscribers
#
aaronpk the one exception is to allow specifically RSS and Atom to remove entries
[Michael_Beckwit joined the channel
#
aaronpk but that's just because it's historical from PubSubHubbub and we wanted to maintain backwards compatibility
#
aaronpk afaik Superfeedr is the only one that does that
#
[qubyte] I see.
#
cweiske (because nobody else has taken the time to implement it)
#
aaronpk in general websub isn't super widely adopted yet either, so we'll see how it shakes out, whether people want to carry that diffing idea forward into other feed formats like jsonfeed or microformats
#
[qubyte] Brilliant! This makes me very happy. It means some little things I’ve been working on aren’t wrong.
#
aaronpk but for now the spec is clear that anything other than RSS or Atom must deliver the full contents of the document
#
[qubyte] This is part of why I proposed a session for it in Brighton, but the lack of adoption meant nobody wanted to chat (or did, but wanted to chat more about something in a parallel session).
#
cweiske i should test and submit phubb to the websub implementations list
#
[qubyte] I tend to use JSON stream quite a lot, so things work out really well for individual payloads.
#
Loqi yea
#
[qubyte] yup
#
aaronpk I was talking with one company about this, they wanted to send only a single item in a feed to subscribers, but their API typically would return like 10 items, so we realized that if you redefine your "document" to mean "only the most current item" then you can get away with sending only one item in the websub payload
#
cweiske which means subscribers may lose some
#
cweiske if they only rely on the data posted by the hub/publisher
#
[qubyte] Where this diverges from the feed use case is that my services don’t care about historic data.
#
[qubyte] (so I can gloss over this)
#
[qubyte] I need to read the subscriber spec more.
#
aaronpk this was in the context of authenticated subscriptions, so the server would be keeping track of whether a particular subscriber has been notified already
#
aaronpk but yeah the risk of sending single items is subscribers may miss notifications
#
[qubyte] This is a problem with webhooks in general though.
#
aaronpk yep
#
[qubyte] The hub could repeat until success with some backoff I suppose.
#
aaronpk also the flip side is the subscriber is then relying on the hub to reliably send every notification
#
aaronpk I have this problem with swarm's web hooks. sometimes they just don't fire
#
[qubyte] And if items have an index, then the subscriber _could_ issue regular requests to a server for indiviual payloads it might have missed, but that’s outside the websub flow.
#
aaronpk so I have to have a separate polling task to check the API to see if there are any checkins I missed
#
aaronpk or rather if there are any checkins that swarm failed to send to my webhook
#
[qubyte] Stretching that index a little, you could poll an endpoint which only responds with the last index.
#
[qubyte] A bit of a hack though.
#
aaronpk frankly this is where the whole webby model breaks down an I start to wonder if what we really need is some sort of proper synchronization protocol
#
[qubyte] client-server (so not so similar) but interesting: EventSource provides just enough in its spec to know when stuff has been missed, and upon reconnection the browser automatically sends the last received event ID (so the server can immediately send stuff missed in the gap).
#
[qubyte] It depends on the server sending events with IDs and keeping a queue though. Those often go unimplemented.
#
@cswordpress Looking forward to the day when mentions of webmentions replaces mentions of pingback/trackback. #IndieWeb #WCUS (twitter.com/_/status/1190291673298931713)
#
[qubyte] Naive question… Does it seem bad to build a combined hub and publisher? Since it’ll only be for my site it seems like a way to avoid some state and code bulk.
#
aaronpk totally fine
#
[qubyte] cool beans
#
aaronpk the reason for separating them in the spec is to let the hub scale independently of the publisher
#
aaronpk if you don't care about that then you can build them together
#
[qubyte] Yup, that’s how I was reading it.
#
[qubyte] This will be private stuff to begin with (webhook standardisation only), so big lumps of stuff like discovery are going to be skipped.
#
[qubyte] It’ll be signature verification and not a whole lot else.
#
[qubyte] Probably some decisions around topics too.
#
[qubyte] Thanks again for the help! 🙂
#
aaronpk 🎉
[tantek] and [snarfed] joined the channel
#
[snarfed] a bit of discussion today on the bridgy facebook email notification experiment. tldr: we're 0 for 3 on facebook backfeed approaches: API, emails, scraping m.facebook.com (logged in). all fully attempted, none really work. sigh. https://github.com/snarfed/bridgy/issues/854#issuecomment-548844715
#
Loqi [snarfed] agreed on all counts. i share your frustration! i actually wrote that crawler you mentioned, a bit ago. www.facebook.com uses JS, so crawling that is hard, but m.facebook.com (with a logged in session cookie) is pure HTML and eminently scrapeable,...
#
aaronpk This is something I hadn't considered https://twitter.com/balloob/status/1190298087748562944?s=21
#
@balloob Now waiting for the Fitbit API shutdown for my privacy… (twitter.com/_/status/1190298087748562944)
#
@schnarfed ↩️ thanks for the feedback! @aaronpk is right: static sites will generally struggle to do anything outside of their build step, ie anything not directly triggered by their author. webmentions are just one of many examples. (twitter.com/_/status/1190303641921474561)
gRegorLove, gxt, KartikPrabhu, [LewisCowles], [chrisbergr] and [tantek] joined the channel
#
[tantek] what is fitbit
#
Loqi FitBit is a hardware device and service for tracking exercise, sleep, and other physical activity https://indieweb.org/FitBit
#
[tantek] wait who bought FitBit? that should go in the dfn
#
[LewisCowles] Google bought fitbit
[Michael_Beckwit joined the channel
#
[Michael_Beckwit i’m now wondering if samsung wearables have any good apis
#
[Michael_Beckwit cause much to my behest, if i can’t get my info easily like i can with Fitbit right now, that’s a blocker 😄
#
[Michael_Beckwit also may just be one that i bite the bullet on and stick with, like i do gmail
[KevinMarks], [dougbeal], [jgmac1106] and McLovinDiscord[m joined the channel
#
@tomcritchlow ↩️ I'm exhausted. So much duct tape. But - finally the beginnings of webmentions appear across the site.... https://t.co/OuxargskaI (twitter.com/_/status/1190369376912003072)
[snarfed] joined the channel
#
@tomcritchlow @keithjgrant Hey - been setting up webmentions all day and your guide was the best guide I found. Thanks! https://keithjgrant.com/posts/2019/02/adding-webmention-support-to-a-static-site/ (twitter.com/_/status/1190376302362202112)
#
@tomcritchlow ↩️ At the bottom of every post there's a "webmention stuff" tab that you can expand that is totally experimental right now but will begin to populate *things*.... More to come next week. (twitter.com/_/status/1190378487477092352)
KartikPrabhu and [KevinMarks] joined the channel
#
@JamieTanna ↩️ What did you end up doing for client-side JS rendering for webmentions? @fluffy has a great project https://github.com/PlaidWeb/webmention.js which I've found makes it super easy to use (twitter.com/_/status/1190390542477283328)
#
@tomcritchlow ↩️ @JamieTanna @fluffy I was adapting this https://keithjgrant.com/posts/2019/02/adding-webmention-support-to-a-static-site/ but that thing you linked to might be better! (twitter.com/_/status/1190392869976195073)
krychu, KartikPrabhu, [tantek], wgirl_, [Michael_Beckwit, strugee and [grantcodes] joined the channel