Skip to content

Add an alternative, smaller storage for numpy arrays #9

@mverleg

Description

@mverleg

Some notes:

  1. Would be ideal if it is extensible to other 'raw' data (like images).
  2. Using more than one save file would be a disadvantage as it's easy to lose parts.
  3. Mixing formats in the same file (e.g. json followed by raw binary) might be difficult for humans and other implementations to read.
  4. Base64 encoding loses efficiency (33% and not great for zipping) - or base85, which may be worse.
  5. Using a python specific format (like .npz) would be disadvantage.
  6. Reading and writing efficiency is important (reading is already really slow).

I can't think of a way that satisfies all these preferences, so I'll leave it open for now.

Metadata

Metadata

Assignees

No one assigned

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions