dougbeal|mb1, jjuran and snarfed joined the channel
#snarfedaaronpk: instagram api vs scraping, whee! :P right, granary REST and UI now only scrape, for exactly that reason. the library itself still supports the API, for users with approved apps.
#[eddie]!tell aaronpk: I’m installing telegraph on my server, and while following the instructions I don’t see a schema.sql as instructed here: https://github.com/aaronpk/Telegraph#developing I only see under ‘migrations’ an init.sql and a couple random id strings .sql files
#Zegnatsknebel if I may make a wiki edit request of you, could you have a look at https://indieweb.org/Let%27s_Encrypt#Renewal ? E.g. the systemd heading talks about setting up a service, but certbot comes with a service all ready these days if I remember correctly from us setting it up? That page might do well with a little updating so people will be les
#ZegnatI don’t actually know enough about what we did to do it myself, I’m afraid.
snarfed and jeremycherfas joined the channel
#ancardaI can confirm the Certbot timer that's installed out of the box does work, at-least for Apache. I haven't tested it with Nginx (I guess wait a few months and see if my website's SSL breaks)
#ancardaThere's a way to add a hook to that to run `systemctl reload nginx.service`, but I never added that
#aaronpksgreger: I think what you're seeing is the fallback API mode. try setting "extended entities" in the API request
#Loqiaaronpk: [eddie] left you a message 13 hours, 36 minutes ago: I’m installing telegraph on my server, and while following the instructions I don’t see a schema.sql as instructed here: https://github.com/aaronpk/Telegraph#developing I only see under ‘migrations’ an init.sql and a couple random id strings .sql files
#aaronpksgreger: in the fallback mode, any tweet longer than 140 chars by the old counting rules will be truncated and you'll see the link to the full tweet
#aaronpkin extended mode you should see everything
#sgregeraaronpk: does that also work with statuses/home_timeline ? it's not mentioned in the parameters there.
#[eddie]aaronpk: I'm working on breaking my monolithic eddiehinkle.com code into more micro-services. I kind of want to make my authorization endpoint completely separate from my website. So that I have a micropub endpoint, an authorization endpoint, and a website that are all separate.
#[eddie]The only issue I'm thinking through is I want to do something similar to your login screen where I can override what channels things get added to during authorization. Is your authorization code base built into your micropub endpoint? Is that how you are setting those values? Or do you somehow communicate that information between your authorization endpoint and the micropub endpoint?
#aaronpkmy micropub endpoint and authorization endpoint are both part of my actual website
#[eddie]ahhh gotcha. I figured they probably were based on those customization options
#sgregeraaronpk: cool, that was the nudge in the right direction! using the Twython library, I had to add tweet_mode='extended' to the api call and now it works :)
#Zegnatancarda: I followed instructions from sknebel to add the nginx reloading hook. Which is one of those little things that I forgot the exact detail about, so would love him to document instead
#[eddie]aaronpk: the only other question I have is, have you ever tried to have them separate and it went badly or you just decided to build it together from the get-go?
#aaronpkI was using indieauth.com for a while but decided to move it into my own site in order to have more customization options like that
#[eddie]Makes sense. Thanks 🙂 Always good to gather lessons learned before embarking down a specific direction 🙂
#sebsel[eddie] to me it doesn't make sense to get the micropub endpoint outside of your site... it's like taking a door out of a house (and adding a window that connects to the door)? Posts have to get inside in some way, right?
#aaronpkhaving a separate micropub endpoint makes sense for things like jekyll sites, where it's a separate service because it can't really be part of your site
#sebselaaa, yeah. You can't separate it from the content, but you can separate it from the part that serves the content.
#[eddie]sebsel: Yeah, currently I have jekyll so I have a micropub endpoint that puts content into a specific folder but the "display" code is currently jekyll. I am planning on changing that, but I do like the idea that if I wanted I could have a PHP website with a node.js micropub endpoint, etc
#[eddie]Basically being able to use the right tool for the right job
#[eddie]and just have them all agree on the content directly
#[eddie]I'm thinking customization is definitely something I want, so maybe integrating my micropub and IndieAuth service is the right direction. That should allow customizations like channels, etc. But since I want my website to be able to allow others to authenticate and view private content. So my website will have to add IndieAuth login flow for other users, so I might as well use it for my login to.
#sebsel(o wow, I just looked at slack because eddie pinged me, but it looks like the whole conversation between aaronpk and me is shown as send by me... is that a known bug?)
#[eddie]That's actually pushed me to start using the chat.indieweb.org for reading on mobile 🙂
#aaronpk[eddie]: being an indieauth server and logging users in is pretty separate, there's nothing wrong with solving those two different ways
#sgregeraaronpk: it still returns quite a few Twitter status-URLs, but those appear to be posts from people i don't follow (i.e. from retweets/likes)
#sgreger...i guess i could at a later stage work on some way to retrieve those links as well
#sgreger(but that would likely have to be on a tweet-by-tweet basis)
#[eddie]aaronpk: Yeah, that's what I was thinking. Seems like authenticating users with IndieAuth and displaying content from my site to them (or me) is fine to be it's own encapsulated thing (except for the content directory, which it and micropub have to agree on).
#[eddie]If anyone else reads the logs and has thoughts how I'm separating my site, feel free to ping me 😉 I'm interesting in thinking this through well before implementing
#aaronpkthe guiding principle I used when deciding how to split up my new site was essentially that the core of my site should not interact with any external URLs
#aaronpkso the first one to split out is obviously sending and receiving webmentions, since in both cases that requires interacting with arbitrary URLs
#aaronpkwebmention.io handles the verification and parsing of received webmentions, then delivers them to my site as a web hook
#sgregerthat is something i have been doing as well (if i understand that correctly) - my site does not load any 3rd party resources or links to functionalities provided under external urls
#[eddie]That's one reason I installed Xray locally. There was parsing I was doing inside of my micropub endpoint, and I'm trying to shift that all to my local xray install
#aaronpksgreger: I think you're a more extreme example of that even. I have 3rd party client-side scripts still.
#sknebelZegnat ancarda the mean issue with renewal is that if you add your own cron/timer with a renewal hook set and let the default one run in parallel, sometimes the default will win and renew the cert without a reload. Got some people confused in the past (especially since they added their own later/switched it from cron to timer on debian at least)
#LoqiHoliday theme is a temporary custom CSS or other change in a website typically meant to visually indicate the celebration of a holiday or special occasion https://indieweb.org/holiday_theme
#aaronpkI forgot the state of that , is it pending some changes ?