note: this was an earlier draft version of this post, which is unfortunately missing some sections, many links, and some edits that i had added before publishing! atproto growing pains for publishing, <3 leaflet, content sync is hard!

What I thought this week would be:

  • Finish the microcosm website update

  • Sketch out the atproto linked-data post and spec

  • Review a big exciting UFOs client data handling contribution (deferred: she's still working on it!)

  • Constellation API updates

  • Get slingshot running under process supervision 😬

  • Chase some potential funding

  • Hopefully work a bit on at://advent

But someone cut the fibre internet cable to my house on Monday, so this week was more like:

  • Incident response

  • Resilience work and planning

Which, hey! Also important things 🙂

~

Building on microcosm

Conversations

~

below are the rough notes from the week as i went (super ignore-able)


monday

  • home internet died. got constellation onto cellular backup after a few hiccoughs, kept status timely public status updates. nothing to do until the ISP tech comes tuesday (earliest they had).

tuesday

  • internet restored

  • rerouted fibre so modem is next to router

    • bunch of homelab physical stuff, wall mounting etc

  • rented a small vultr vps for a secondary gateway (executing on previous resilience planning)

  • purchased EasyDNS's DNS Pro for microcosm.blue

    • found out that namecheap doesn't have a zone export booo

wednesday

  • week reassessment :D

  • UFOs client data model contribution quick review (PR still pending)

  • website progress!

    • little logos for constellation, spacedust, and slingshot

    • quickstart code snippets for each, in python, bash, and javascript respectively

      • runnable! (with a little js magic)

thursday

  • fix the whats-hot feed on UFOs

    • no prefetched post rip
  • website progresssssss

  • debugged sync problems for 3rd party pds on relays (microcosm and bsky both!)

    • no prefetched post rip

friday

  • debugged more deactivated state stuff, ran a large survey of the network

  • no prefetched post rip