Configure resolves/lockfiles via targets(?) Hello, would it make sense to configure resolves via targets, rather than pants.toml? i.e.

resolve(name="res1", lockfile="pants.lock")
python_requirements(resolve=":res1")

I'm curious if there are any blockers before trying to implement this. i.e. is order of parsing build files an issue, given that resolves need to be evaluated before any other targets? It seems a bit similar to

environments()

.

A lot of pants assumes that resolves come from the

PythonSetup

subsystem. I'm not sure if anything requires it. There is more complexity around the options to resolves, but I don't think it's insurmountable. Resolves are also implemented per-language.

Yea this kinda is part of the bigger discussions of "pants configuration per directory" discussion

I suspect this will be a lot easier to do in the new backend than trying to retcon it into the existing one

In my case specifically, I'm trying to create a macro which makes setting up a new module in our monorepo easier. This is all the boilerplate needed just to get close to the default poetry-ish experience 😕:

def python_module(
    name: str, tests: bool = True, test_fixtures: bool = True, typed: bool = True, **kwargs
):
    __defaults__(all=dict(resolve=parametrize(*MODULES), **kwargs))
    resolve(name=name)
    poetry_requirements(name="poetry")
    python_sources(
        name="sources", sources=["**/*.py", "**/*.pyi", "!**/*_test.py", "!**/conftest.py"]
    )
    if typed:
        resources(name="py_typed", sources=["**/py.typed"])
    if tests:
        python_tests(name="tests", resolve=name, sources=["**/*_test.py"])
    if test_fixtures:
        python_test_utils(name="conftest", resolve=name, sources=["**/conftest.py"])

Ah, I see

But speaking more generally, it seems inconsistent that resolves are defined in the pants.toml file, when other "objects" in the pants-world are defined in BUILD files.

Yeah, that dichotomy has always been annoying