Skip to content

Every commit rewrites the entire Arrow storage file #243

Description

@ccmdi

bake_and_save rewrites the entire Arrow IPC file on every commit, which takes about 15 seconds at 10 million rows, versus about 17ms for a delta autosave. The goal is to accumulate deltas and bake lazily instead: stop baking on every commit, write chained delta files, replay deltas on open, background-compact past a threshold (roughly 50 deltas or 1MB), dedupe during replay, track a bake version somewhere for crash detection, consider moving delta storage into SQLite, and use thin undo entries (IDs, not full clones) for large imports.

Related: #236

Activity

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

    effort: largeBig or open-ended, needs breaking downenhancementNew feature or request

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions