Skip to main content

hammett is a fast python test runner

Project description

https://travis-ci.org/boxed/hammett.svg?branch=master https://codecov.io/gh/boxed/hammett/branch/master/graph/badge.svg

Hammett is a fast python test runner that aims to be compatible with the parts of pytest most people use (unless that conflicts with the goal of being fast). It also adds some features that can dramatically improve your testing experience if you adapt more fully to hammett.

How much faster is hammett?

This will depend on how fast your test suite is. Hammett isn’t magic, it’s just a runner made by someone who cares about performance a lot. It aims to have minimal startup overhead and minimal overhead for each test. I have written some benchmarks for measuring this overhead which you can find at https://github.com/boxed/test-benchmarks (TL;DR if your test suite is < 1s in pytest you will see VAST improvements, if it’s minutes or hours, not so much).

A real world example is running the test suite of tri.declarative:

  • pytest: ~860 ms

  • hammett: ~160 ms

Or iommi:

  • pytest: ~10 s

  • hammett: ~8 s

But even if your test suite is big and slow you can still get some big improvements out of hammett if you often run just one file or one test:

In iommi, running -k test_render_attrs_none

  • pytest: ~1.3 s

  • hammett: ~0.6 s

All of this is from a full and clean run. Hammett has features to avoid that!

Tests locked to a module

In hammett you can optionally name your test files module__tests.py (that’s two _). This tells hammett that if you change module only the tests in module__tests.py needs to be run. You can place those files either in the tests directory or right next to the module you’re testing.

Run just relevant tests

Hammett keeps track of what modules and what tests have changed and runs only the tests it needs to. Assuming you lock your tests to a module like above.

If hammett gets confused you can delete the .hammett-db file and it will start from scratch.

Pytest features that work in hammett

  • pytest.mark

  • parametrized tests

  • fixtures

  • with pytest.raises

Some plugins work, but you have to specify to load them in setup.cfg:

[hammett]
plugins=
    pytest_django

Usage

First install: pip install hammett

Then run hammett: python -m hammett

Hopefully it will run your entire test suite!

Hammett works with some pytest plugins, most notably pytest-django, at least for some projects. You need to specify what plugins hammett loads manually in setup.cfg though.

Keeping pytest compatibility

If you want to use the hammett specific feature of module__tests.py-style test files, you can still keep compatibility with pytest by specifying the module of your project and the tests file pattern, like this:

testpaths=
    tests
    my_project
python_files=
    *__tests.py

Also keep importing pytest instead of hammett obviously.

License

BSD

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

hammett-0.10.0.tar.gz (23.5 kB view details)

Uploaded Source

File details

Details for the file hammett-0.10.0.tar.gz.

File metadata

  • Download URL: hammett-0.10.0.tar.gz
  • Upload date:
  • Size: 23.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.10.15

File hashes

Hashes for hammett-0.10.0.tar.gz
Algorithm Hash digest
SHA256 e825aa7ac3e1b2c743fc69e348d709fd072017c34bf223441c969486192abc08
MD5 cac44f9cf2fe9bdca404c2ee6b81b08e
BLAKE2b-256 2d1b9e6e564edf69f3d1c85fa436026490304a39ff8f444c66dba52b29476c82

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page