Re: support for other protocols, it seems like twtxt would be pretty easily adapted to work over the p2p file network DAT, though it'd need client support for DAT or some way to follow people via files and sync in the background, which might be simpler for clients to support but would still require changes to most clients.
@kas I'm not currently using a handrolled client, but I've started a couple.
@nblade I could probably put something together, but I went this route originally because existing generators felt like they took too much control from me over the exact output. I want very small, static pages I can throw behind nginx, and I want to know exactly what the contents are. My current mess of spaghetti Racket is getting hairy since I'm halfway through an unfinished rewrite I started a while ago and forgot about, but at least I know exactly what it's doing.
@pete that's a clever solution; I'll have to set up something like that.
@tdemin good points, though another that I've noticed is that it's difficult to tell who in your network is actually reachable with your tweets. My HTTPS cert went unupdated for a brief while and now I have no idea who is still following me since I got it working again, so it's difficult to tell where I can really have a conversation. A centralized service can tell who's following who, but that's basically impossible in twtxt.
@freemor @kas #twtxt's quietness is actually something I enjoy about it. I care a lot more about signal-to-noise ratio than I care about the regular activity. It's also a really fun thing to write clients for to play around with new libraries or languages.
@kas my bad - I've been developing a twtxt client, hence the frequent requests. I'm switching it over to use a fs cache for testing so I'm not hitting the twtxt files so much.
@quite the static site CMS? It's open source: https://github.com/netlify/netlify-cms/
@phil I actually just started a new job working on a frontend for Hugo (and other static site generators).
Additionally, there's a lot that can be done by a client to reduce the network traffic and UI latency of twtxt without changing the protocol.
@mdom, @kas re: metadata, I'm (obviously) in favor of my suggestion for metadata-in-comments, but I don't think we should have comments in comments.
I'm not working on bussard, I'm just IRC friends with @technomancy and he expressed interest in making scheme available there.
@mdom mostly for fun, but once I have it working more it'll get integrated into https://gitlab.com/technomancy/bussard.
@mdom @kas thanks! I was checking access logs on nginx and noticed a bunch of 404s to my twtxt url, so I decided to pick it back up.
How should metadata about a twtxt feed be stored? Weigh in @ https://github.com/buckket/twtxt/issues/48
twtxt-able twtxt helper scripts, part 1: make 'tt' run 'twtxt timeline -l 1000 | less'
I find the wide variety of ways people are deploying twtxt pretty fascinating
Timeline Sandbox