Shulou(Shulou.com)06/03 Report--

This article will explain in detail how to speed up the startup time of Python applications. The editor thinks it is very practical, so I share it with you as a reference. I hope you can get something after reading this article.

I heard that pipenv9.0.2 has been released and the startup time has been greatly improved.

I tried it quickly, but I didn't feel happy. So I used the new features of Python3.7 to study it.

In this article, I'll introduce this feature and how to use it.

Startup time ≒ import time

For example, pipenv-h takes much longer to execute than it takes to display a help message.

In general, when an application starts, there are some startup processes, such as loading environment variables or configuration files.

For Python applications, the import module takes up most of the startup time. For example, pipenv-- version costs about 800ms, while import pipenv costs 700ms.

Display the import time of the module

Python 3.7has a new feature to show when the module was imported.

This feature is enabled through the-X importtime option or the PYTHONPROFILEIMPORTTIM environment variable.

For example, you can test the import time of pipenv with the following command:

Or

Here is an example of a pipenv-version output.

Research introduction time

At the end of the output, you will see these lines:

On the last line, 579479 indicates that import pipenv needs 579479us.

While importing pipenv, many other modules have been imported as well. From the example above, you can see that pipenv imports pipenv.cli. The sub-import is indented by 2 spaces.

Let's look at the last line. 507 indicates that only 507us is required to run the pipenv module. 579479-507=578972us is used for child imports.

Find the slow part.

Let's find the slow subtree from the output. I selected a few lines.

Pkg_resources

As you can see, importing pkg_resources is slow.

But surprisingly, pkg_resources is not indented; it is not a child import of pipenv.

This means that the pkg_resources is imported by the pipenv script rather than the module.

The bad news: importing pkg_resources is slow. This is a known problem and is difficult to fix without breaking backward compatibility.

The good news: you can avoid importing pkg_resources!

After wheel is installed, pip builds it and installs it from the build package.

Installing from wheel (.whl) and source packages (.tar.gz) is a different process.

When installing from wheel, pkg_resources is not used in the script:

IPython

Look at the following section.

Pipenv imports dotenv, dotenv imports dotenv.ipython, and it imports IPython.

This is why pipenv starts slowly in my environment; I have installed IPython.

But why import IPython? I read the dotenv source code and found that it is used for IPython extensions.

Of course, pipenv and many dotenv users do not use IPython extensions.

I submitted a push request to dotenv, advising dotenv to import IPython as needed.

Because pipenv has its own copy of dotenv, I submitted a push request to pipenv to completely delete dotenv.ipython.

Conclusion

I can reduce the time of pipenv--version from 800ms to 500ms.

Importing time analysis is a good way to study and optimize application startup time.

This is the end of the article on "how to speed up the startup time of Python applications". I hope the above content can be of some help to you, so that you can learn more knowledge. if you think the article is good, please share it for more people to see.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.