Skip to content

Make Like-burst bookkeeping survive failed writes #39

Closed

Description

@jamesconwaydev

Like-burst detection keeps its state in flags.json, and a few failure paths leave that state wrong:

  • if the flags.json write itself fails, that run's new flags are lost;
  • if updating the aggregate issue fails, its body goes stale;
  • an issue closed while we're editing it can be missed;
  • a single failed issue lookup skips the whole hour.

None of these lets anyone game the list, but each one hides a burst we should have looked at.

Done when each case has a test, a lost write is redone next run, and one bad lookup only skips that one entry.

Activity

  1. added
    priority: mediumNext in line once the high ones are done
    area: botThe Python bots that take submissions, count Likes and check health
    on Nov 6, 2026
  2. updated Effort M, Priority Medium and Area Bot on Nov 6, 2026
  3. moved this from Todo to In Progress in DevStream Roadmapon Nov 6, 2026
  4. jamesconwaydev commented on Nov 6, 2026

    @jamesconwaydev
    CollaboratorAuthor

    Done in #199.

  5. moved this from In Progress to Done in DevStream Roadmapon Nov 6, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    area: botThe Python bots that take submissions, count Likes and check healthpriority: mediumNext in line once the high ones are done

    Type

    Fields

    Priority

    Medium

    Effort

    M

    Area

    Bot

    Projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions