Skip to main content

Pants 2.18.0 is released!

· 6 min read
Alexey Tereshenkov
Alexey Tereshenkov
Pants Maintainer

We're pleased to announce Pants 2.18.0, the latest release of Pantsbuild, the scalable and ergonomic build system. To update, set pants_version = "2.18.0" in your pants.toml. If you're not using Pants to build, format, test (and more) your codebase, get started now.

Highlights in 2.18.0 include:

Even more improvements to AWS Lambda and Google Cloud Function support, including simpler configuration and support for AWS Lambda Layers. Many more options can now be controlled when building Docker images. Significant feature work and bug fixes to "deploy" backends like Helm and Terraform.

There's much more, too: performance optimizations, new backends, bug fixes and features!

Starting from the previous 2.17.0 release, you can now support Pants financially.

Pants is an open-source project that is not owned nor controlled by any one company or organization, and does incur some expenses. These expenses are managed by Pants Build, a non-profit that was established for this purpose. This non-profit's only source of revenue is sponsorship by individuals and companies that support Pants.

We offer formal sponsorship tiers for companies, as well as individual sponsorships via GitHub.

Experimental backends!​

Did you know that Pants has a lot of "experimental" backends? These include support for languages, tools and ecosystems that are still in preview. The documentation of backends now lists all these backends. Please go check it out, try backends relevant to you and let us know how they go for your use cases, report bugs and/or submit fixes! People like you trying them is how we can perfect the features and squash the bugs to stabilize them.

Functions-as-a-service improvements for Python​

Pants includes support for building function-as-a-service artifacts for Python, via the pants.backend.awslambda.python and pants.backend.google_cloud_function.python backends. These backends make it easy to generate zip artifacts appropriate for uploading to cloud services.

These backends had significant improvements to their start-up performance and size in 2.17, and the improvements have continued in 2.18:

  • The new python_aws_lambda_layer target can be used to build Layers for AWS Lambda. For instance, this allows putting third-party requirements from PyPI into a layer, and only first party-sources from your repository into the function code, making for faster builds and uploads. (For consistency, the python_awslambda target has been renamed to python_aws_lambda_function.) The runtime field is no longer necessary when the interpreter constraints unambiguously refer to an entire single major version. For instance, with [python].interpreter_constraints = ["==3.11.*"] in pants.toml, one doesn't need to specify runtime="python3.11" to python_aws_lambda_function or runtime="python311" to python_google_cloud_function, as it will be inferred.
  • For AWS Lambda targets, the complete_platforms argument now has default values provided by Pants, which gives more reliable builds, especially when building on a different platform than the AWS Lambda itself. This change also means that only one of the complete_platforms or platforms fields should be provided.

Simplification of Python tool lockfiles​

You now install custom versions of Python-based tools (such as Pytest, MyPy and so on) by pointing the tool's install_from_resolve option at a regular Python resolve. You create the lockfiles for these resolves just as you would for your own code's resolves. You can have one resolve per tool, or one resolve for multiple tools, and you can even install tools from the same resolve you use for your code. This is useful when you have tools that also provide runtime libraries, such as Pytest. See the documentation for more details.

Note that this means that you can no longer export the built-in, default lockfile for a tool, as that is not represented by a named resolve. If you need to export Python tool virtualenvs for consumption outside of Pants, e.g., by an IDE, you must first create a resolve for them.

Infrastructure-as-code iteration​

Pants has nascent support for describing and deploying infrastructure via the pants.backend.experimental.helm and pants.backend.experimental.terraform backends. These have seen several improvements in 2.18.0:

Docker finesse​

The pants.backend.docker backend now supports many more options to control how images are built:

  • on the [docker] subsystem: build_hosts, build_no_cache
  • on [the docker_image target]: build_network, build_platform, extra_build_hosts

These new options allow, for instance, more easily building a x86_64 image while running on an arm64 host.

More visible visibility​

The pants.backend.experimental.visibility backend now also runs as a linter, such as when running pants lint ::, to report all visibility violations at one time.

This backend allows applying rules and restrictions to which targets can depend on other targets, for example, a repository with two applications and some shared library code can use the visibility backend to enforce that code in one application cannot depend on code in the other application.

Finding paths between addresses that expand to multiple targets​

Pants previously allowed finding all paths in the dependency graph between two individual targets, like a python_source and a python_test. You can now find all paths that exist between multiple targets, for instance, by passing a directory address.This will help you to answer various questions such as "How is this Python module connected to a Python package, if at all?" or "How tightly are two Java packages connected?".

This also enables sophisticated dependency graph analytics. By listing all the paths in the repo, for example, between sources and tests, one can find the source modules that have most paths leading to tests, identify very long paths, and many more.

For example, this command will count all paths between all source and all test targets in a repository:

$ pants paths --from=project/tests:: --to=project/src:: | jq length
42

Much more!​

Pants 2.18 involved several months of work from dozens of contributors, thank you everyone! See the release notes for all of the new features and fixes.

To update, set pants_version = "2.18.0" in your pants.toml. If you're not using Pants to build, format, test (and more) your codebase, get started now.