The target is intended to provide metadata for files, so it sits next to them, but you can have

dependencies=[...]

that criss-cross the repo, there are no restrictions

That makes sense. The last missing piece is the syntax.

# src/tests/BUILD
python_sources(dependencies=[":test_csv_files"])

files(name="test_csv_files", sources=["**/*.csv"])

This correctly detects my csv file sources

pants dependencies src/tests
-> 
src/tests/test_data/csv/shape_example.csv:../../test_data_files

I just can't seem to figure out the correct syntax to include in the BUILD file located within my test module directory.

# src/tests/dir/subdir/BUILD

python_tests(
    name="tests",
    dependencies=[":test_csv_files"]
)

:test_csv_files

is out of scope, but I am not sure how to reference the

python_sources

and/or

files

field located within

src/tests/BUILD

. I think I am just lacking a bit of understanding on how paths are resolved between

BUILD

files.

A dependency address is relative to the repo root:

path/to/dir:target_name

Incredible explanation, thank you. This all looks to be working on my end, both in my overly-simple example as well as the more complex, real-life case.

BTW you could consider putting that

files()

target in

src/tests/test_data/csv/BUILD

, but that is not required

I will be trying to only reference the files that I need per

BUILD

file. Most of the heavy handed glob work will be just to increase quality of life. Nonetheless, I will keep this in mind! Thank you again, this was tremendous help. P.S. I integrated pants into more than half of our monorepo. CI time is down from ~50 minutes to ~30 minutes. The final steps will hopefully bring us down to ~15 minutes. Pants is a great tool and its value is tremendous.

Out of curiosity. which Pants features are the CI time savings coming from?

“test” and “lint” are primarily what we have integrated. Remote test caching certainly saves the most time.