Microsub doesn't actually say anything about feeds or how long ... • Aaron Parecki

Jamey Sharp https://toot.cat/@jamey • Jul 13
@aaronpk Do you have any statistics on storage costs for GUIDs vs post titles vs publication dates vs post contents?
I'm thinking there are multiple sensible levels of cache flushing. It seems nice to keep just enough information to browse the list of posts, while discarding the contents of a post until someone actually opens it. But if nobody scrolls back in the history for a feed for a while, it's probably okay to discard the old post metadata too.
Microsub couldn't do that as-is, can it?

Microsub doesn't actually say anything about feeds or how long data is stored, it's an abstraction for browsing channels of content, where the contents are not specified by Microsub.

I could definitely implement a Microsub server that doesn't actually store entries and just loads from RSS feeds on demand, tho I probably wouldn't do that exactly for performance reasons.

The main things that Microsub cares about are

• Return a list of channels for the client to display
• Return the top posts of a channel (doesn't have to necessarily be date-ordered)
• Indicate whether there are "more" items in that channel that can be loaded

Everything else is an implementation detail, including if/how old content is stored.

Portland, Oregon • 86°F

Fri, Jul 13, 2018 3:35pm -07:00

1 reply
Have you written a response to this? Let me know the URL:
- Jamey Sharp toot.cat/@jamey
  
  @aaronpk But if the Microsub server includes some post in a timeline response, it has to include the full body of that post, right? Even if the client doesn't need the body yet?
  
  Fri, Jul 13, 2018 10:37pm +00:00

Posted in /replies using monocle.p3k.io