File: README.rst

package info (click to toggle)
python-ciso8601 2.3.3-2
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid
  • size: 372 kB
  • sloc: python: 1,294; ansic: 939; sh: 22; makefile: 8
file content (105 lines) | stat: -rw-r--r-- 4,011 bytes parent folder | download | duplicates (2)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
=====================
Benchmarking ciso8601
=====================

.. contents:: Contents

Introduction
------------

``ciso8601``'s goal is to be the world's fastest ISO 8601 datetime parser for Python.

In order to see how we compare, we run benchmarks against each other known ISO 8601 parser.

**Note:** We only run benchmarks against open-source parsers that are published as part of Python modules on `PyPI`_.

.. _`PyPI`: https://pypi.org/

Quick start: Running the standard benchmarks
--------------------------------------------

If you just want to run the standard benchmarks we run for each release, there is a convenience script.

.. code:: bash

  % python -m venv env
  % source env/bin/activate
  % pip install -r requirements.txt
  % ./run_benchmarks.sh

This runs the benchmarks and generates reStructuredText files. The contents of these files are then automatically copy-pasted into ciso8601's `README.rst`_.

.. _`README.rst`: https://github.com/closeio/ciso8601/blob/master/README.rst

Running benchmarks for all Python versions
------------------------------------------

To make it easier to run the benchmarks for all supported Python versions, there is a Dockerfile you can build:

.. code:: bash

  % docker build -t ciso8601_benchmarking .
  % docker run -it --rm=true -v $(dirname `pwd`):/ciso8601 ciso8601_benchmarking

Running custom benchmarks
-------------------------

Running a custom benchmark is done by supplying `tox`_ with your custom timestamp:

.. code:: bash

  % python -m venv env
  % source env/bin/activate
  % pip install -r requirements.txt
  % tox '2014-01-09T21:48:00'

It calls `perform_comparison.py`_ in each of the supported Python interpreters on your machine.
This in turn calls `timeit`_ for each of the modules defined in ``ISO_8601_MODULES``.

.. _`tox`: https://tox.readthedocs.io/en/latest/index.html
.. _`timeit`: https://docs.python.org/3/library/timeit.html

Results are dumped into a collection of CSV files (in the ``benchmark_results`` directory by default).

These CSV files can then formatted into reStructuredText tables by `format_results.py`_:

.. _`perform_comparison.py`: https://github.com/closeio/ciso8601/blob/master/benchmarking/perform_comparison.py
.. _`format_results.py`: https://github.com/closeio/ciso8601/blob/master/benchmarking/format_results.py

.. code:: bash

  % cd benchmarking
  % python format_results.py benchmark_results/2014-01-09T214800 benchmark_results/benchmark_with_no_time_zone.rst
  % python format_results.py benchmark_results/2014-01-09T214800-0530 benchmark_results/benchmark_with_time_zone.rst

Disclaimer
-----------

Because of the way that ``tox`` works (and the way the benchmark is structured more generally), it doesn't make sense to compare the results for a given module across different Python versions.
Comparisons between modules within the same Python version are still valid, and indeed, are the goal of the benchmarks.

Caching
-------

`ciso8601` caches the ``tzinfo`` objects it creates, allowing it to reuse those objects for faster creation of subsequent ``datetime`` objects.
For example, for some types of profiling, it makes sense not to have a cache.
Caching can be disabled by modifying the `tox.ini`_ and changing ``CISO8601_CACHING_ENABLED`` to ``0``.

.. _`tox.ini`: https://github.com/closeio/ciso8601/blob/master/benchmarking/tox.ini

FAQs
----

"What about <missing module>?"
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

We only run benchmarks against open-source parsers that are published as part of Python modules on PyPI.

Do you know of a competing module missing from these benchmarks? We made it easy to add additional modules to our benchmarking:

1. Add the dependency to ``tox.ini``
1. Add the import statement and the parse statement for the module to ``ISO_8601_MODULES`` in `perform_comparison.py`_

`Submit a pull request`_ and we'll probably add it to our official benchmarks.

.. _`Submit a pull request`: https://github.com/closeio/ciso8601/blob/master/CONTRIBUTING.md