Skip to content

Add a Mumbling bitmap reader for v4 manifest deletion vectors #1125

Open

Description

@xndai

Iceberg v4 stores manifest deletion vectors inline on root manifest entries (manifest_info.dv, tracking.deleted_positions and tracking.replaced_positions). They are encoded in the new Mumbling bitmap format, whose spec is still a draft (apache/iceberg#16518). Iceberg Java has a reader (apache/iceberg#16747), but iceberg-cpp has none, and no C/C++ library implements the format. Planning scans of v4 tables needs one (part of #1107).

Proposal: add an internal, read-only MumblingBitmapView. It wraps the serialized bytes without copying them, validates everything that memory safety depends on, and supports Contains() and Cardinality(), the same surface as Java's ManifestBitmap. While the spec is a draft it stays in an *_internal.h header that isn't installed. Writing bitmaps and applying them in the v4 manifest reader would be follow-ups.

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

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions