hello everyone!

Have a question for the room: I am trying to setup a remote caching server to use with pants in a CICD pipeline. In trying to test out Pants (v1.30.4) cache settings I used the following config in pants.toml

[cache]
resolver = 'rest'
read_from = ['<http://localhost:8080>']
write_to = ['<http://localhost:8080>']

I have tried everything I can think of but I cannot seem to get Pants to make requests to a remote cache or fail when passing invalid cache locations...

has anyone else run into this or have a working example of using a remote pants cache?

Another round of voting on new logo font options:

Pants-#13.jpeg,Pants-#14.jpeg

@happy-kitchen-89482 has a polly for you!

think it looked better in all caps 😅

I'm wondering how pants might interact/integrate with two tools: 1) Does anyone here use pants with pre-commit-hooks? Do you have pants run pre-commit? Or do you have pre-commit run pants? https://pre-commit.com/ 2) What about reviewdog? When someone posts a PR in Github, I don't just want to give them a pass/fail; I want the CI to annotate their commits with checks, code suggestions (just click accept/merge), and reviews. Does pants post checks out of the box? Could it use reviewdog somehow to add that feature? https://github.com/reviewdog/reviewdog

We have pre-commit hooks that run Pants. e.g., https://github.com/pantsbuild/pants/blob/main/build-support/githooks/pre-commit

I'm not familiar with reviewdog, let's have a look

Interesting. When you run

./pants lint

pants will emit the various linters' error messages to stderr

(you opt in to the linters you want: https://www.pantsbuild.org/docs/python-linters-and-formatters)

So presumably you let reviewdog run on that output?

i'm not sure from my cursory read how reviewdog gets its paws on linter output, but presumably whatever works for running a linter directly should work for running

./pants lint

I think they have plugins to interpret each tool's output.

Oh wow. Yeah not much is involved to pass black output into reviewdog. https://github.com/reviewdog/action-black/blob/master/entrypoint.sh

And here is the definition for parsing black's output: https://github.com/reviewdog/errorformat/blob/master/fmts/python.go#L26

https://twitter.com/DZoneInc/status/1376262900503961606 Pants Build retweeted: Nice Lost reference 😂 😂 Twitter

According to https://www.pantsbuild.org/v2.4/docs/reference-mypy I got the notion that mypy could be run with the

lint

goal.. but it doesn’t, regardless of if I run it with

--no-mypy-skip

or with

--mypy-skip

.. Is this a doc bug?

https://twitter.com/semaphoreci/status/1372187974432124941 Pants Build retweeted: Choosing a repository strategy is not only about technology but also about work culture and communication. 👀 Learn… https://twitter.com/i/web/status/1372187974432124941 Twitter

Hi folks, I've written up a proposal to streamline this Slack workspace, so there are a lot fewer channels, and it's less confusing. Your thoughts welcome: https://docs.google.com/document/d/1f8eu3NWaOjRCKiqTSGt7zVUdASJMCCkxqQdeUvOp6Iw/edit?usp=sharing

Welcome! 👋

Couldn't find this in the docs... we're advised to use

PANTS_CONFIG_FILES

to use a dedicated CI toml file when running from CI (to disable the dynamic ui among others) but this seems duplicative, and the name

...FILES

seems to imply that we can use multiples. Does this "or" the files together or some such, ie can we

PANTS_CONFIG_FILES=<http://pants.ci|pants.ci>.toml,pants.toml

to not duplicate things, ie have

pants.toml

but overwrite/supplement keys with

<http://pants.ci|pants.ci>.toml

? If so, I assume order matters?

Seems like there was no opposition to this plan to streamline our Slack channels: https://docs.google.com/document/d/1f8eu3NWaOjRCKiqTSGt7zVUdASJMCCkxqQdeUvOp6Iw/edit?usp=sharing

So I will start working on this later - some channels will be renamed, others will be archived, I will post in each archived channel which channel replaces/subsumes it.

Hi, Perhaps something for the docs. I’m using namespace packages, and had to put in some custom config in order for mypy to correctly detect my package names. Configuration I’m hoping pants should be able to add, but as it works with just a few lines of configuration, adding it to the documentation might be good enough to save others the trouble of finding this out themselves.. In

pants.toml

:

[mypy]
config = "mypy.ini"
args = "--namespace-packages --explicit-package-bases"

and in

mypy.ini

:

[mypy]
mypy_path = src:plugins

Any paths under which you have sources, but that aren’t part of the package name. So in this case, there are sources in

src/<namespaceroot-pkgs>/…

and

plugins/<other-pkgs>/…

To aid in debugging this, the

-v

flag to mypy was really helpful, so you can inspect the output from mypy, on the form:

LOG:  Found source:           BuildSource(path='plugins/xx/flask/rest/routes/generic.py', module='xx.flask.rest.routes.generic', has_text=False, base_dir='/private/var/folders/8j/c8jf_msj009947wyw82xvdkw0000gn/T/process-executionLIFk6M/plugins')

Here, the

BuildSource(…, module=)

will show which module mypy thinks the source is in. Relevant (and helpful) mypy docs: https://mypy.readthedocs.io/en/stable/running_mypy.html#mapping-file-paths-to-modules

Anyone ever ran into this before? Using pants 1.30.x we occasionally see

ERROR: No matching distribution found for <some dep>==x.x.x

That dependency clearly exists in pypi but it was updated in the last x minutes. It looks like pip is hanging on to its cache or not reaching out to pypi due to maybe the cache-control header being set too high. Anyone have a solution for this? We have tried clearing the pip cache, pants caches, or just waiting. sometimes it works… others it does not. What am I missing here?

Very minor question--any timeline for adding Python 3.9 support (in terms of just pants itself; I notice the

pants

executable only lists up to 3.8)? Not a dealbreaker at the moment, but our app is built with 3.9 and it would lightly simplify some infrastructure.

Getting a new error on Mac but not on Linux, anyone else seeing this?

11:30:12 00:02       [run]
                     Traceback (most recent call last):
                       File "/Users/amckenna/code/python-commons/.pants.d/test/pytest-prep/CPython-3.8.2/dbace314d7260e4bab850ba0bdd823530b504bb3/.bootstrap/pex/pex.py", line 394, in execute
                         working_set = self._activate()
                       File "/Users/amckenna/code/python-commons/.pants.d/test/pytest-prep/CPython-3.8.2/dbace314d7260e4bab850ba0bdd823530b504bb3/.bootstrap/pex/pex.py", line 92, in _activate
                         for dist in env.activate():
                       File "/Users/amckenna/code/python-commons/.pants.d/test/pytest-prep/CPython-3.8.2/dbace314d7260e4bab850ba0bdd823530b504bb3/.bootstrap/pex/environment.py", line 239, in activate
                         self._working_set = self._activate()
                       File "/Users/amckenna/code/python-commons/.pants.d/test/pytest-prep/CPython-3.8.2/dbace314d7260e4bab850ba0bdd823530b504bb3/.bootstrap/pex/environment.py", line 408, in _activate
                         resolved = self._resolve(working_set, all_reqs)
                       File "/Users/amckenna/code/python-commons/.pants.d/test/pytest-prep/CPython-3.8.2/dbace314d7260e4bab850ba0bdd823530b504bb3/.bootstrap/pex/environment.py", line 290, in _resolve
                         distributions_by_key[distribution.as_requirement().key].append(distribution)
                     AttributeError: 'NoneType' object has no attribute 'as_requirement'

Also wondering how we can avoid pulling watchman from pantsbuild.org, this is not allowed in our prod or build environments