Do the Pants daemons ever time themselves out? I just noticed that I have several Python instances running for a few hours, even though I haven't touched anything with Pants in a day

Pants 2.10.0 is out! https://blog.pantsbuild.org/introducing-pants-2-10/ Thank you to all the contributors and early testers 🎉 Posted in #announce

Working with the new resolves, trying to figure out what’s what I have one question and one error report feedback for pex 🧵

Was it a purposeful change in 2.10 to no longer allow targets to be defined with sources outside of where the target is defined?

Hi, I'm running

pytest

with the

pytest_dynamodb

plugin which requires the pytest runner to spin off a subprocess. I'm having an issue where even providing the full

/usr/bin/java

path, Pants' pytest runner can't find the binary

I have questions about segmentation faults in go resolve on M1 machines in ancient pants.

Question: are people using local Python requirements on their

main

branch, where you install from a file on the filesystem? John just landed support for VCS/git requirements with Pex lockfiles 🎉 We're trying to figure out how to prioritize local requirements

Hopefully a quick "quality of life" question but can anyone explain to me how to get

./pants fmt ::

with

black

to run on my BUILD files? I've seen reference to that ability but not sure how to wire it up. 😇

Is there a reason why pants' limit on concurrent runs applies to the

help

command?

Semi-related to Eric’s question above: is there a way for us to swap new versions of PEX into our pants 2.10.0 system, so we can make use of the support for VCS requirements in lockfiles? or do we need to wait for a pants 2.10.1?

👋 Hello everyone. I’m interested in using Pants to install packages from a private Python package index hosted by Azure Artifacts. The recommended solution is to install a plugin for keyring so that pip can use it to discover the credentials: https://pip.pypa.io/en/stable/topics/authentication/#keyring-support. In this case, keyring, artifacts-keyring (the plugin) and pip would all be installed in the same virtualenv so pip can find the plugin. Is this possible to do with Pants?

Re: Python application in a Docker target. I'm attempting to use a PEX file target (Flask web server) with a Docker target and can't seem to get the right files into the Pants sandbox.

Is there a way to explicitly enable

PIP_NO_BINARY

for a single dependency, via an

override

key on

poetry_requirements

etc? I'm currently developing on an M1 and there is currently an issue with the grpc package and the way its binary is shipped (see https://groups.google.com/g/grpc-io/c/9iM4SPpOyC8 for a great summary).

Y'all write really good blogs. They flow so nicely and always have a great mix of technical detail and practicality.

What levers/knobs exist to have pants `stdout`/`stderr` use one log level, and have the log file use another? I want our CI output to be concise for humans, but upload the more verbose log for diagnoses.

This might be useful: https://github.com/pantsbuild/example-python/blob/main/.github/workflows/pants.yaml#L57-L62 I did notice that pants prints to stderr only. I could be wrong about that.

Hi team, sometimes I got the following error while using pants binary. In the pants.toml I have put

pants_version = "2.10.0+gitb27a00dd"

.

BuildConfigurationError: Version mismatch: Requested version was 2.10.0, our version is 2.10.0+gitb27a00dd.

Heyo Pants people. I see that JS is on the roadmap of upcoming languages. Is anyone actively working on it or have a hacked together example to try out?

I think @wide-midnight-78598 likely has the farthest on the hacky approaches 🙂

show me what you got

https://github.com/sureshjoshi/pants-plugins/tree/main/pants-plugins/experimental Currently just Prettier, next eslint. Compilation actually isn't the most difficult, when I look at it, but I'm trying to keep a reasonable order. My (internal) stuff is hacked together shell scripts, random hardcoded plugins, etc. Slowly working on something public and usable there (for a few languages)

Hi, I was trying to run pants in github action and got this when running

./pants --level=debug binary ...

What does this mean and how to fix it?

00:24:57 [DEBUG] urllib3.connectionpool:pid=2668: <https://files.pythonhosted.org:443> "GET /packages/64/de/7e79ee9dd4fad3750f3fd6a1b4f2c7dcece54c486e765747e1593f8df029/googleads-15.0.1.tar.gz HTTP/1.1" 200 53290
00:24:57 [DEBUG] pants.process.lock:pid=2668: releasing lock: <pants.process.lock.OwnerPrintingInterProcessFileLock object at 0x7cf0925dd310>
                     Dumping requirement: PythonRequirement(jenkinsapi==0.3.9)
               Waiting for background workers to finish.
00:24:58 00:17   [complete]

timestamp: 2022-03-29T00:24:58.083088
Exception caught: (pex.archiver.UnpackError) (backtrace omitted)
Exception message: Could not extract /tmp/tmpObxfk2/myproject.common-1.0.tar.gz


               FAILURE
Error: Process completed with exit code 1.

I'm curious about Python PEX in Docker. Is there a way to use

docker compose

in conjunction with PEX. For example, I have a common "logger" package which I want all of my services to use. The best way to deploy "logger" as a dependency of these services is in a PEX file. (Or maybe there's another way?) I want to use

docker compose

to run my entire codebase (front end on port 8080, various microservices, postgresql on port 5432, etc.) Any suggestions, ideas, or otherwise?

On a new machine (x86 Macbook Pro - to make some progress despite some ARM/Apple Silicon issues) I've checked out my repo that has my WIP Pants integration work... I kick off

./pants test

to bootstrap, build the venvs, and run the tests and find I have continuous loop...

14:39:12.61 [INFO] Filesystem changed during run: retrying `Test` in 500ms...
14:39:12.64 [INFO] Canceled: Building requirements.pex with 4 requirements: <requirements elided>
14:40:12.36 [INFO] Filesystem changed during run: retrying `Test` in 500ms...
14:40:12.39 [INFO] Canceled: Building requirements.pex with 4 requirements:
<requirements elided>

Hi, I was upgrading pants all the way from

1.17.0

. Now I am at

1.23.0

, and I ran into this error when running

./pants test

, which I have encountered before and solved by specifying the version of

coverage

under

[pytest]

in

pants.ini

.

No cached artifacts for 254 targets.
                   Invalidated 254 targets.06:44:34 [WARN] /home/cedriczheng/.cache/pants/setup/bootstrap-Linux-x86_64/1.23.0_py36/lib/python3.6/site-packages/pants/backend/python/tasks/pytest_prep.py:92: DeprecationWarning: DEPRECATED: Pants defaulting to a Python 2-compatible Pytest version will be removed in version 1.25.0.dev2.
  Pants will soon start defaulting to Pytest 5.x, which no longer supports running tests with Python 2. In preparation for this change, you should explicitly set what version of Pytest to use in your `pants.ini` under the section `pytest`.

If you need to keep running tests with Python 2, set `version` to `pytest>=4.6.6,<4.7` (the current default). If you don't have any tests with Python 2 and want the newest Pytest, set `version` to `pytest>=5.2.4`.
  return PyTest.global_instance().get_requirement_strings()

06:44:34 [WARN] /home/cedriczheng/.cache/pants/setup/bootstrap-Linux-x86_64/1.23.0_py36/lib/python3.6/site-packages/pants/backend/python/tasks/pytest_prep.py:92: DeprecationWarning: DEPRECATED: Pants defaulting to a Python 2-compatible Pytest version will be removed in version 1.25.0.dev2.
  Pants will soon start defaulting to Pytest 5.x, which no longer supports running tests with Python 2. In preparation for this change, you should explicitly set what version of Pytest to use in your `pants.ini` under the section `pytest`.

If you need to keep running tests with Python 2, set `version` to `pytest>=4.6.6,<4.7` (the current default). If you don't have any tests with Python 2 and want the newest Pytest, set `version` to `pytest>=5.2.4`.
  return PyTest.global_instance().get_requirement_strings()

**** Failed to install coverage-6.3.2 (caused by: NonZeroExit("received exit code 1 during execution of `['/usr/bin/python2.7', '-s', '-', 'bdist_wheel', '--dist-dir=/tmp/tmp8zl5tznp']` while trying to execute `['/usr/bin/python2.7', '-s', '-', 'bdist_wheel', '--dist-dir=/tmp/tmp8zl5tznp']`",)
):
stdout:

stderr:
Traceback (most recent call last):
  File "<stdin>", line 14, in <module>
  File "<string>", line 80
    classifier_list.append(f"Development Status :: {devstat}")
                                                            ^
SyntaxError: invalid syntax



               Waiting for background workers to finish.
23:44:37 00:47   [complete]
               FAILURE

ERROR: Package SourcePackage('file:///home/cedriczheng/myrepo/.pants.d/python-setup/resolved_requirements/CPython-2.7.12/coverage-6.3.2.tar.gz') is not translateable by ChainedTranslator(WheelTranslator, EggTranslator, SourceTranslator)

(Use --print-exception-stacktrace to see more error details.)

But in previous version (1.22 or 1.21), the instructions suggested me to remove those settings. If I put them back, I will get this error instead:

23:51:52 00:04   [complete]
               FAILURE

ERROR: Could not satisfy all requirements for coverage==5.0.3:
    coverage==5.0.3, coverage>=5.2.1(from: pytest-cov<3,>=2.8.1)

I don’t know what to do now…

Is there any further guidance on how to deal with these errors…

[WARN] The target src/xyz.py imports `Abc`, but Pants cannot safely infer a dependency because more than one target owns this module, so it is ambiguous which to use: ['src/abc/__init__.py', 'test/abc/__init__.py'].

test/abc/__init__.py

is blank. The source for

Abc

is in

src/abc/__init__.py

.

src/xyz.py

contains

from abc import Abc

My repo has

src

and

test

roots that largely mirror each other in their hierarchy. Is this a bad pattern? Is there a way to default the dependency inference to consider the

src

root over the

test

root - unless there is an explicit override? Also is there a way to retest the whole codebase for this class of error as it appears to appear for one invocation of

./pants dependencies src/xyz.py

but then disappears (and I think I have other similar issues that I would like to resolve)

OK, so that error is because

coverage-6.3.2

uses f-strings in its setup.py, so it cannot run on python 2.7.

So you're right that you need to pin to an earlier version of coverage

as you're doing with

coverage==5.0.3

it looks like

But this is conflicting with the

coverage>=5.2.1

requirement of the version of

pytest-cov

you're using