Skip to content

Clarification of bitmap buffer 'uncompressed length' definition #48883

@kdn36

Description

@kdn36

Describe the bug, including details regarding any error messages, version, and platform.

Ref pola-rs/polars#26126
and https://github.com/apache/arrow/pull/43950/changes#top

Clarification request: is the 'uncompressed length' field for a compressed bitmap buffer defined as (a) exactly the length in bytes rounded up to the nearest byte size (i.e., n.div_ceil(8) for n bits), or (b) any possible length in bytes rounded up to a given word size (e.g. n.div_ceil(32)*4 for n bits).

In the case of 5 rows, (a) would imply exactly 1 byte uncompressed size, while (b) would allow 4 bytes uncompressed size.

Component(s)

Documentation, Format

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions