Skip to content

Benchmark the speed of different Zsh plugin managers

License

Apache-2.0, MIT licenses found

Licenses found

Apache-2.0
LICENSE-APACHE
MIT
LICENSE-MIT
Notifications You must be signed in to change notification settings

rossmacarthur/zsh-plugin-manager-benchmark

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

32 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

zsh-plugin-manager-benchmark

Benchmark different Zsh plugin managers.

Summary

Introduction

  • The following plugin managers are benchmarked:

    • antibody v6.1.1 (⚠️ archived)
    • antidote v1.8.6
    • antigen v2.2.2 (⚠️ unmaintained)
    • sheldon v0.7.3
    • zgen master @ 0b669d2 (⚠️ unmaintained)
    • zgenom main @ 2b55d9d
    • zinit main @ de85908f
    • zplug master @ ac6c2a3
    • zpm master @ 16d74f8
    • If you would like to add a new plugin manager feel free to open an issue and/or pull request.
  • For each plugin manager the install time and the load time was tested

    • install time is the the time taken on the first time loading ~/.zshrc until a prompt appears.
    • load time is the time taken for each subsequent load of the ~/.zshrc until a prompt appears.
  • 26 of some of the most popular plugins (by GitHub stars) listed in Awesome Zsh Plugins were used as as test case. See plugins.txt. The plugins were extracted using awesome-star-count.

  • hyperfine was used as a benchmarking tool. All benchmarks were run on a quiet cloud VM.

  • No deferred loading of plugins was done. For example: many plugin managers can be used in conjunction with zsh-defer. Or zinit could be run using wait which defers the plugin loading until after a prompt.

Results

Install time

Install time

Lower is better. Although install time is not as important as load time it is probably at least worth doing the install in parallel. From these results its very clear which plugin managers install in parallel vs sequential.

Load time

Load time

Lower is better. This is the metric we care about most because its the time it takes to open a new shell until we get a usable prompt.

Details

Host

  • Vultr.com
  • Ubuntu 22.04
  • 8 vCPUs
  • 16 GB RAM
  • NVMe storage

Usage

To benchmark the 'install' step run the following.

./bench.sh install

To benchmark the 'load' step run the following.

./bench.sh load

These commands will output results to results/. You can then create charts from these results using the following.

First install Python dependencies.

pip install seaborn pandas
./chart.py install

or

./chart.py load

License

This project is distributed under the terms of both the MIT license and the Apache License (Version 2.0).

See LICENSE-APACHE and LICENSE-MIT for details.

About

Benchmark the speed of different Zsh plugin managers

Resources

License

Apache-2.0, MIT licenses found

Licenses found

Apache-2.0
LICENSE-APACHE
MIT
LICENSE-MIT

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published