Hi all. I’ve got pants set up on my project, but have some questions. If I run

./pant fmt ::

multiple times in a row, it reports that changes were made each time. Why is that?

Your tools might be disagreeing on formatting

I’ve got black configured in pyproject.toml, and I thought in that one place.

Yes, sorry, I’ve got isort in there. But I did take steps to get them to “agree”

black runs a bit differently in pants. you can use --keep-sandboxes=on_failure for pants to see what is being run. also in pants sandbox, all the source files is passed to black via commandline args, so you’ll need to use something like force-exclude instead of exclude, just to highlight one of the differences.

Did you set this? https://pycqa.github.io/isort/docs/configuration/black_compatibility.html

Yep, did that.

to be fair, I haven’t reconciled my isort configuration as well, it produces different results in pants sandbox and outside

Any context you can provide would help. What tools are you using? What configs do you have, and where? etc...

Here’s the contents of my

pants.toml

file:

[GLOBAL]
pants_version = "2.14.0"

backend_packages = [
#    "pants.backend.docker",
#    "pants.backend.docker.lint.hadolint",
    "pants.backend.python",
    "pants.backend.build_files.fmt.black",
    "pants.backend.python.lint.black",
    "pants.backend.python.lint.isort",
    "pants.backend.python.lint.pylint",
    "pants.backend.python.typecheck.mypy",
]


[python]
tailor_pex_binary_targets = false
interpreter_constraints = ["CPython==3.9.*"]

[source]
root_patterns = [
    "project/app",
    "tests",
]

[anonymous-telemetry]
enabled = false

[pylint]
config = "config/.pylintrc"

[test]
output = "all"

[pytest]
extra_requirements.add = []

extend-exclude will probably need to be force-exclude

Changing to

force-excluded

generated a lot more “Pants cannot infer owners for the following imports…” messages.

Looks like I need to change the order of execution of black and isort. Is there a way to make isort run first and then black?

I think the order in the

backend_packages

controls the order that they run for

fmt

. --- With

isort

specifically, there's https://github.com/pantsbuild/pants/issues/15069 that can result in isort's order changing, but that may not be what you're hitting. Our configuration (that is stable), is: • `pants.toml`: black first, then isort • `pyproject.toml`: (I suspect the

force_...

etc. aren't relevant, but maybe

default_section

is)

[tool.black]
line-length = 88
target-version = ['py39']
skip-magic-trailing-comma = true

[tool.isort]
profile = "black"
py_version = 39
line_length = 88
# attempts to do something similar to zimports
force_single_line = true
force_sort_within_sections = true
order_by_type = false
# FIXME: workaround for <https://github.com/pantsbuild/pants/issues/15069>
default_section = "FIRSTPARTY"

--- I've also noticed that a re-run can report changes even if it doesn't actually make any. I suspect this is if the source code is the same as the last run's output and the invocation is somehow cached/not rerun, but am not sure, although this may not be what you're hitting with your follow-up message about

lint

vs

fmt

.

not sure why nobody else is running into this issue. I got black to work using force_exclude since pants pass all files as args so if you just do exclude or extend-exclude it would format all the files you excluded because whatever you pass in using cmd line args takes precedence. similar with isort, needed filter_files = true flag in order for skip_blob to take affect

@broad-processor-92400 and @plain-night-51324, thanks very much for the feedback, helpful! Here is the relevant section of my pyproject.toml file:

[tool.isort]
profile = "black"
py_version = 39
line_length = 100
force_single_line = true
force_sort_within_sections = true
order_by_type = true
default_section = "FIRSTPARTY"
filter_files = true
skip_glob = "tests"

[tool.black]
line-length = 100
target-version = ['py39']
skip-magic-trailing-comma = true
force_exclude = "tests"

Using this

./pants fmt ::

and

./pants lint ::

behave in predictable ways, and don’t seem to “undo” each other. Again, thanks very much for your help and insight!

@chilly-tailor-75063 can you share what changed? Previously you all addressed black butting heads with isort but the claim was that was not the issue / you had that sorted IIUC. What did solve it?