dill

Version:: 0.3.7
Category:: data
Cluster:: Vali

Author / Distributor

Description

dill extends Python’s pickle module for serializing and de-serializing Python objects.

It supports a wider range of Python object types than pickle, including:

Lambdas and nested functions
Closures and generators
Class instances and interpreter sessions

This makes dill especially useful for parallel and distributed computing applications, including multiprocessing, where it allows passing more complex objects between processes.

Documentation

$ python -m dill --help

usage: python -m dill [-h] [-v] {check,info,diff} ...

options:
  -h, --help            show this help message and exit
  -v, --version         show version information

actions:
  {check,info,diff}
    check               check if pickle files can be unpickled
    info                show the contents of a pickle file
    diff                compare contents of two pickle files

Examples:
  python -m dill check file.pkl
  python -m dill info file.pkl
  python -m dill diff file1.pkl file2.pkl

Examples/Usage

Load Dill:

$ module load data/dill/0.3.7-GCCcore-12.3.0

Unload the module:

$ module unload data/dill/0.3.7-GCCcore-12.3.0

Serialize and restore a Python lambda:

import dill

# Define a lambda function
f = lambda x: x * 2

# Serialize to a file
with open("lambda.pkl", "wb") as f_out:
    dill.dump(f, f_out)

# Deserialize from file
with open("lambda.pkl", "rb") as f_in:
    loaded_f = dill.load(f_in)

print(loaded_f(10))  # Outputs: 20

Installation

Source Code on GitHub