📚 node [[agora development]]
pull color="#b51f08"> <title>500 Internal Error wtf.
<link rel="stylesheet" href="https://doc.anagora.org/css/center.css"> <button class="pull-url" value="https://doc.anagora.org/css/center.css">">pull</button>
<div class="container-fluid text-center">
    <div class="vertical-center-row">
        <h1>500 Internal Error <small>wtf.</small></h1>
  • #push [[agora development]]
    • set up local dev environment in the train
    • [[better search]]?
      • what's a quantum of better search? it's pretty bad right now; e.g. [[poetry install]] would not bring up [[install poetry]]
        • would re-adding fuzzy matching fix this?
        • should I implement a simple search algorithm myself for the purpose of understanding how to solve this?
          • tokenization -> compute combinations is trivial but expensive
          • -> [[bag of words]] is cheap
            • this actually requires defining what is the source of truth for [[equivalence class]] and for [[proximity]] between nodes
            • I've always wanted to think of Agora search as as an exploration within a certain [[radius]] of an inferred center of gravity; maybe this is distance in some embedding?
              • -> [[edit distance]]
                • seems like an interesting approach to explore next
      • [[auto pull wikipedia]] sort of also tackles this in a totally different direction, in particular if we extend it to [[wikidata integration]]: if wikidata surfaces an entity for us, volunteer its type; if the entity is far from the current node in an interesting way, maybe [[auto pull]] it as well after some time.
      • if I'm adding auto pull maybe it's time to make [[settings]] clearer, in particular have an easy toggle for auto pull to trigger or not (leads to [[agora autopilot]] maybe, which I've always thought could be very fun.)
  • #push [[agora development]]
    • ship one improvement?
      • auto pull social after some seconds?
      • auto pull stoa in empty nodes?
      • bring up twitter bot -> [[fix bot for ivo]]
        • it was up, for some reason Ivo's post is not being seen. might need to write a 'scrape one user' "one-shot" mode -- maybe doing per user timelines is actually a good idea.
        • this is not the first time the bot has issues with Ivo's posts. I have no idea why they're not showing up in the bots' timeline -- according to the Twitter API anyway. The bot is following Ivo, I checked.
        • writing a [[one shot]] mode which takes a timeline to scrape+react to (if we haven't reacted previously) sounds like a reasonable next step at this point; it'd also make [[catch up]] better to have this code. iterating
⥅ node [[channel]] pulled by user empty.
📖 stoas
⥱ context