A question on the poetry_requirements macro. Is there a way for pants to ignore dev-dependencies in project.toml?

Any suggestion on how to increase permanently the

ulimit

with

WSL2

I can't make it work correctly on Windows, every time I get, I also tried to modify different files following many tutorial, I still have the warning at the beginning...

ulimit -n 65536
-bash: ulimit: open files: cannot modify limit: Operation not permitted

Hey folks, been finding that if I remove a BUILD file and run

./pants tailor <path-to-removed-file>

, the BUILD file is not regenerated even if there are still unmapped targets in that directory. Is this expected behaviour? This also happens if I run

./pants --changed-since=<commit-with-removed-file> tailor

While I can edit sub pages of https://www.pantsbuild.org/ I don't seem able to edit the main page which needs some tweaking. I am the impatient sort and just want to start using pants and could care less about how great it is, though it does sound pretty special! The 'Getting Started' button would better be renamed 'Welcome' and a new button added called 'Getting Started' that would link to https://www.pantsbuild.org/docs/getting-started. Y

Hi all, just a quick question regarding the python interpreter path. I'm using

pyenv

to manage the python version and then dynamically setting

.vscode/settings.json

to

{"python.defaultInterpreterPath": sys.executable}

in order to get autocomplete functionality etc. This path typically points to the venv in the cache directory. Is this right approach or do you recommend another setup for VSCode? Thanks.

in Pants 2.9.0, the

--dependencies-type

option was deprecated. The docs say:

This option is misleading and not very useful. In the future there will be a more robust way of querying and filtering dependencies.

suggesting doing this in the meanwhile

./pants dependencies ::
| xargs ./pants filter --target-type=python_requirement
| xargs ./pants peek | jq -r '.[]["requirements"][]'

Have I missed a clever way to list only 3rd party or only 1st party dependencies in Pants now? 😕

We've put a lot of effort recently into getting most of our python repo (and all of our default resolve) compatible with native apple m1 development. There are a couple places where it is not feasible to use only packages that are compatible with m1, but we have isolated them into their own resolve. This has worked well for being able to manage packaging the different artifacts, but resulted in a state where developers who running in a native m1 environment are now unable to run commands like

./pants lint ::

. Is there anything we can do? An env var we could set on these systems to always filter out targets on the "bad" resolve?

Anyone have a sense for when

2.15.0rc1

might release? I’m trying to decide if I should wait for it or just use

2.16.0.dev3

which contains the bugfix I’m waiting for already. How buggy are dev releases normally for sort of day-to-day use where the build being blocked for a couple hours is no big deal?

Is there any support for setting

GOOS

and

GOARCH

env vars on a

go_binary

target? I’m not seeing anything documented or in the source but maybe there’s some way I’m missing? Edit: nevermind, found

PANTS_GOLANG_SUBPROCESS_ENV_VARS

as well as Environments.

A (quick) question, a couple of my colleagues encounter the following issue when they use the pants binary at their first time. ❯ ./pants --version Traceback (most recent call last): File "/Users/386162564/miniconda3/envs/brain_toy/lib/python3.8/runpy.py", line 194, in _run_module_as_main return _run_code(code, main_globals, None, File "/Users/386162564/miniconda3/envs/brain_toy/lib/python3.8/runpy.py", line 87, in _run_code exec(code, run_globals) File "./pants/__main__.py", line 89, in <module> File "./pants/__main__.py", line 51, in __maybe_run_venv__ File "./pants/__main__.py", line 11, in __re_exec__ FileNotFoundError: [Errno 2] No such file or directory Any idea?

Hello all! I am trying to use the new environments feature to run a test in a docker container. The problem is that my function code is relying on a python module that is not part of the pants context, i.e. not added as a third party package that is inferred by pants, but is only part of that docker image. If I run the test it fails due to an import error. I guess that pants is not looking for over python modules in that environment but runs in its sandbox. How can I incorporate those modules during the test run?

Hi. So I was messing about with the pants.toml and ran into a very cryptic error message:

➜ ./pants --no-pantsd -l=debug  generate-lockfiles
06:23:00.28 [WARN] /Users/vish/.cache/pants/setup/bootstrap-Darwin-x86_64/pants.Z3K0Pi/install/lib/python3.9/site-packages/pkg_resources/_vendor/packaging/specifiers.py:255: DeprecationWarning: Creating a LegacyVersion has been deprecated and will be removed in the next major release
  warnings.warn(

06:23:00.28 [ERROR] 1 Exception encountered:

  InvalidRequirement: Parse error at "'='": Expected string_end

Wasn’t helping. Anyways, after messing about the toml file a bit more and checking git diffs, I was able to resolve it. What I had done was this:

[python.resolves_to_interpreter_constraints]
airflow-default = "CPython>=3.7.10, <3.8"

The correct way is to set it as an array:

[python.resolves_to_interpreter_constraints]
airflow-default = ["CPython>=3.7.10, <3.8"]

I think these kinds of mistakes can occur. if the error message pointed to line or even say the error was in toml file, I would have been able to resolve it more easily. It isn’t critical, but a nice usability improvement 🙂

Hmm, one way to filter out targets is to use tags. You could have an

m1_incompatible

tag and set it on all the relevant targets, and then have m1 users set the PANTS_TAG env var to exclude those targets. This is how Pants selects targets to run on macs in its own CI, for example.

This doesn't relate to resolves, directly, though

But you could perhaps use a macro to generate a target that has both the tag and the resolve set

That was my one idea so thank you for validating it! Going to give this a try tomorrow

Yeah, probably a

m1_incompatible_python_sources

macro could save you some typing

Related - we are close to being able to cross-build for linux on a mac by automatically executing relevant processes either in a local docker container or on a remote executor. At that point your code may not need to run on an m1 in order to develop natively on an m1

Look forward to trying that, but the m1s are still a problem. We're still dealing with some dependencies that don't publish arm wheels. Remote could work though That would be a larger change to integrate into everyone's setup

We'll definitely use local docker builder right away. Currently, even when developing on Linux, the PEXes don't match our target docker images and we have a whole complicated script wrapping the pants packaging in a docker image that would be nice to be able to do natively

Hi, I'm trying to run a script with

pants run ...

. When the script is a single file, it works perfectly. However, I can't figure out how to configure it to work with relative imports. Will appreciate any help. Details in 🧵.

Playing with this and I don't think i can do it with tags. The m1 incompatible resolve is not disjoint with our default resolve and we have a lot of

'resolve': parametrize('default', 'cnvkit')

. I am not aware of any way to apply the

m1-incompatible

tag only to the

@resolve=cnvkit

targets.

Hey Folks, I'm trying to setup a scala 3 project using pants. I've added the config to set the global scala version:

[scala.version_for_resolve]
jvm-default = "3.2.1"

When I run

./pants check ::

it errors with:

17:18:37.48 [ERROR] 1 Exception encountered:

  MissingRequiredJvmArtifactsInResolve: The JVM resolve `jvm-default` is missing one or more requirements for the Scala runtime library. Since at least one JVM target type in this repository consumes a `scala_sources` target in this resolve, this resolve must contain `jvm_artifact` targets for each requirement of the Scala runtime library.

Please add the following `jvm_artifact` target(s) somewhere in the repository and re-run `./pants generate-lockfiles --resolve=jvm-default`:

jvm_artifact(
  name="org.scala-lang_scala-library",
  group="org.scala-lang",
  artifact="scala-library",
  version="3.2.1",
  resolve="jvm-default",
)

Which is expect from what I've seen in the docs. I created a

BUILD

file in the root with this config.

jvm_artifact(
  name="org.scala-lang_scala-library",
  group="org.scala-lang",
  artifact="scala-library",
  version="3.2.1",
  resolve="jvm-default",
)

The problem is that this artifact doesn't exist, the scala-library for scala 3 is named scala3-library. If I try to change the artifact name to the correct one it errors with the same message as above.

This is some admittedly vague onboarding feedback. I'm trying to get pants working with an existing repo that has a tangled

requirements.txt

file that is installed in a managed environment with some pre-existing packages that ought to be treated as constraints but are not yet. It is fair to see that this at best accidentally works and the dependencies are messy. Pants/pex keeps telling me -- correctly! -- variants of "this doesn't work! neither does that! these dependencies are incompatible you fool!", This is good in the sense that part of the reason pants is attractive is to have dependencies that more than accidentally resolve, but it is really hard to get to "first light" for this repo where the unit tests run. Right now it would feel nice if there was a

--yolo

flag to have things "work" like they do today.

I’m sure I’m missing something here, but I’ve been staring at the docs for too long and can’t see the obvious answer - I’m using

fastapi

which vendors

starlette

as a transitive dependency with a different package name - I’m using a poetry

pyproject.toml

which I guess may be the issue as

starlette

is listed in my lockfile but not in the

pyproject.toml

?

17:14:25.85 [WARN] Pants cannot infer owners for the following imports in the target projects/api/main.py:api:

  * starlette.middleware.base.RequestResponseEndpoint (line: 13)

but the

fastapi

dependency is being pulled in:

./pants dependencies  projects/api/main.py:api                                                                                    
//:poetry0#fastapi
//:poetry0#loguru
//:poetry0#mangum
projects/api/api_metadata.py:api
projects/api/router/all_routers.py

so this feels close but not quite there 🙂

We've just hit 2400 github stars! Onward and upward! (PS If you aren't already starring https://github.com/pantsbuild/pants, we'd appreciate it!)

And we're already at 2401! Thanks to whoever started the push to 2500 🙂

Hitting an issue trying to set up a docker build on Mac OS ARM (pants

2.16.0.dev3

):

stderr:
There was 1 error downloading required artifacts:
1. dockerfile 3.2 from <https://files.pythonhosted.org/packages/6e/75/1395fac29bd2dcccb51ffae11ba9a19fe159eabf9596bca4cf49c1a2b191/dockerfile-3.2.0.tar.gz>
    ERROR: Command errored out with exit status 1: python setup.py egg_info Check the logs for full command output.

This seems to be coming from bootstrapping the docker backend. Where can I find the logs of this process to keep troubleshooting? I’ve got

--keep-sandboxes=on_failure

set.