I want to make some .mo translation files available to my program at runtime so I added a target

resources(name="locale", sources=["**/*.mo"])

but Pants seems to ignore them? They don't show up when I run

pants dependencies

Is Pants filtering binary files or something?

There a way to publish packages (

python_distribution

) where I get the version from the current git tag? Is there a way to do this without writing my own plugin?

Hi everyone! 👋 Our startup is looking to switching our whole code base to use Pants. One issue we’ve been having is that we’ve been unable to figure out the correct way to get the debugger to work for FastAPI. I created a sample project here to give an example of what we’re trying to do. When I run

./pants run --debug-adapter src/python/greeter/main.py

, I can successfully connect through

Python: Remote Attach

in Visual Studio Code. However, when I set a breakpoint on line 9 of

/src/python/greeter/main.py

and try to visit

localhost:8000/

in my browser, the breakpoint is never activated. I see output in both my terminal and the debug console but the breakpoint is just never activated. Any help would be greatly appreciated, thanks!

👋 Hi everyone! We are implementing pants in our new project, one doubt I had during this how we can manage versions of package inside pants. I need to maintain multiple version for a shared package. Different versions will be used by different service. Is it possible to have multiple versions based on git tag for a local package.

I’m looking into a lockfile issue that’s more annoying than anything else, but any insights appreciated. 🧵

Regarding

mypy

config, what’s the reason for having both an

extra_requirements

option as well as

extra_type_stubs

?

Is is possible to augment some existing type hint with my own? For example, pytorch comes with some type hints but it's missing some. So I created a

torch.pyi

file and added some type info for the

torch.load

method. I was hoping mypy would use the def of

torch.load

from my

.pyi

file and then fall back to the ones that ship with pytorch for everything else. But that doesn't appear to be the case: now it only pays attention to my .pyi file ignoring all else. Maybe this is more of a question for the mypy folks??

here’s another lockfiles question.. 🧵

Is there a way to change the CWD of the

docker_image

rule? I have a use case of running Docker on a git submodule but I'm trying to avoid modifying the code in that submodule. So my setup is in the parent directory of the submodule and looks like

docker_image(
    name="prom-aggregation-gateway",
    source="feat-configurable-gauge-agg/cmd/prom-aggregation-gateway/Dockerfile",
    dependencies=[
        ":submodule"
    ]
)

files(
    name = "submodule",
    sources = ["./feat-configurable-gauge-agg/**"]
)

Unfortunately, the Docker build isn't able to find

/prom-aggregation-gateway

FROM alpine:3.10
COPY prom-aggregation-gateway /
ENTRYPOINT ["/prom-aggregation-gateway"]

Hi, what would be the way to use env variable for generating docker repository name? During our CI/CD on AWS we fetch the region and the account id to push the image to different deployment env (staging/production) ECR repository. So it can’t be a static name. https://www.pantsbuild.org/docs/tagging-docker-images#setting-a-repository-name

is anyone using a workflow engine/language on top of pants for complicated jobs? — in the context of github action workflows, pants seems to be natural for running individual actions — but not for “orchestrating” workflows — am i missing something?

Hello, We've started to use the publish goal for pushing docker images and I just noticed the following warning during the run:

13:08:28.42 [WARN] DEPRECATED: InteractiveProcessRequest is scheduled to be removed in version 2.15.0.dev1.

 

Instead, use `InteractiveProcess.from_process`.

Is there something I should change?

It seems like Pants isn't able to run a Docker target that hasn't been published yet. It keeps trying to reach the Docker registry and complains because the image can't be found. Is this intentional?

I may have found a bug with multi-stage docker builds. I converted a single stage docker build into a multistage build, and at that point Pants failed to resolve file dependencies that I explicitly included.

Is there a way using the pants cli to see a docker_image target's registry.repo:tag ?

Hey @wide-midnight-78598, thanks for the progress on

pyright

! I'm coming with an additional request 😉 I've tried pinning `pyright`'s version as follows in `pants.toml`:

[pyright]
version = "pyright==1.1.258"
lockfile = "pants_dir/3rdparty/pyright_lockfile.lock"

But this isn't recognized:

→ PANTS_SHA=11b4fd412631c6315b474e49fe482dec1767cf29 ./pants check libs/geometry::
...
10:00:25.71 [ERROR] Invalid option 'lockfile' under [pyright] in /home/churlin/dev/kaiko-eng/pants.toml
10:00:25.71 [ERROR] Invalid option 'version' under [pyright] in /home/churlin/dev/kaiko-eng/pants.toml
10:00:25.71 [ERROR] Invalid config entries detected. See log for details on which entries to update or remove.

Am I doing something wrong or is the locking support missing? I need it because right now

pants

pulls a different version of

pyright

than my regular CI and I get discrepancies 😕 I can create an issue if that would help you, let me know.

Hi, my

generate-lockfiles

was taking forever (>=10 mins) and won’t even finish. I figured it out that it’s because of adding s3fs which depends on aiobotocore. It’s the aiobotocore which has some very restrictive dependencies. How long is pants/pex supposed to keep trying? I tried it with pip and pip throws an error about this.

Share my project: The Hard Way to Go Through All Models in Sklearn It is managed by pants https://github.com/da-tubi/rikai-sklearn

Heyo again! I seem to have found what might be a bug?

❯ ./pants 
./pants: line 478: /Users/user@company.com/.cache/pants/setup/bootstrap-Darwin-arm64/2.14.0_py39/bin/python: No such file or directory

❯ ls -lah /Users/user@company.com/.cache/pants/setup/bootstrap-Darwin-arm64/2.14.0_py39/bin/python
lrwxr-xr-x  1 <mailto:user@company.com|user@company.com>  staff    70B Nov 10 12:43 /Users/user@company.com/.cache/pants/setup/bootstrap-Darwin-arm64/2.14.0_py39/bin/python -> /Users/user@company.com/.pyenv/versions/3.9.13/bin/python3.9

❯ file /Users/user@company.com/.cache/pants/setup/bootstrap-Darwin-arm64/2.14.0_py39/bin/python
/Users/user@company.com/.cache/pants/setup/bootstrap-Darwin-arm64/2.14.0_py39/bin/python: broken symbolic link to /Users/user@company.com/.pyenv/versions/3.9.13/bin/python3.9

This is what I get when I try to run pants on my M1 Mac. I believe the issue might have to do with the “@” in my username since I log in using a federated account?

📣 Pants team invites you to add your organization or company to the Who Uses Pants page. We'd love to feature your team on this page! It helps showcase the diverse set of organizations that Pants is helping, and it demonstrates to your own users, employees and candidates that you're on the cutting-edge of the latest and greatest in build system technology. Click here to get started.

Hello, I have a problem with the plugins and how the dependencies to those are resolved. I have a plugin requiring a package that is available only for python3.9. Pants seems to run the pip command with python3.7, so pip tells me there is no package matching the requirements. I tried to add interpreter_contraints to differents places but I'm not able to make that work. How can I force pants to use python3.9 when resolving dependencies for pants plugins?

I know this isn’t the right place to ask this, but I can’t find a good VSCode resource to ask these questions. After I’ve done a

./pants export ::

I have a dist folder with an isort binary, how can I configure VSCode to use that isort when formatting my code? I have it using the correct black binary, but I can’t get it to use the isort binary. This is with both the default python extension and the microsoft official isort plugin. Any ideas how to get pants and vscode to share the same isort binary?

Hey Pants Team, We are upgrading from 2.13 -> 2.14 and running into this issue:

[2022-11-15T22:41:38Z] ModuleNotFoundError: No module named '_sqlite3'

this is when running lint in CI. Will paste some logs in the thread.

I'm trying to create a macro for building a docker image from Python code following https://blog.pantsbuild.org/optimizing-python-docker-deploys-using-pants/ (thanks for publishing it!), and I'm not sure of the best way to get PEX files into the docker image. In particular, if the pex file is at

abc/def/ghi:target-name

then dep inference will know that

COPY abc.def.ghi/target-name.pex ...

in the docker instructions refers to it... but how can a macro automatically compute the output path? (particularly the

abc.def.ghi

part) One option would be to set

output_path

to have it fixed, but that risks collisions when the macro is used multiple times. (Sketch of the macro in the 🧵 )

Hey all, new to pants and have been playing around with it over the last few days along with pyoxidizer. Very excited to see where this can go, my first interest is autocompletion scripts (which obviously require very fast initialisation speeds) I've created this post playing around with a very mini project. Steep learning curve moving from setup.py to tomls. Hope this post is useful to others out there

This message was deleted.

I'm having some trouble running a goal after I added sources related to type checking. My

src/services/app/BUILD

currently looks like this:

python_sources(
    name="app_sources",
    sources=[
        "mypy-stubs/stub1.pyi",
        "mypy-stubs/stub2.pyi",
    ]
)

pex_binary(
    name="app",
    entry_point="main.py",
    dependencies=[
        "src/vendor/python:reqs#setuptools",
    ],
    restartable=True,
)

With this definition,

./pants run src/services/app:app

, I get

ModuleNotFoundError: No module named 'services'

. When I remove the stubs from

python_sources

, the run command works fine, except then type checking fails. With this definition,

./pants dependencies

also doesn't show anything - it's as if adding the stubs completely negates any other dependencies. Clearly I've defined things wrong, but I can't seem to find how.

🚲 🏚️ on naming the

NodeToolBase

subsystem:

NodeToolBase

? <-- generalizing to NodeJS

NpxToolBase

? <--

npx

is used under the hood

NpmToolBase

? <--

npx

is part of the

npm cli

Thoughts on adding Git integration to introspection goals like

./peek

? Might be a bit scope creepy but very useful for CI + there's already subsystem logic for Git. One use case I have in mind is raising a CI lint error if a Docker image target changes but its version number hasn't changed.

Hi folks! Periodic reminder that two really useful things you can do to show appreciation for all the help and support provided in this community Slack are: 1) Star the pantsbuild/pants repo and encourage your teammates to do the same 2) Add your company/org to the Who Uses Pants page Thanks in advance! ~ Benjy, on behalf of the Pants team