Also it looks like

./pants peek

isn’t getting the

resolve

of the target

Curious if anyone can suggest a tool for visualization of a Python project dependency graph? Looking to have a deeper understanding on how

./pants dependees --transitive <file_path>

finds links to files. The current dump shows all dependees for a file but not how they are connected.

Can I have resources on a in-repo plugin? it seems in-repo plugin ignore the BUILD files. I have a plugin at

pants-plugin/jvm_stage

, BUILD file:

python_sources(resolve="pants-plugins", dependencies=[":java_resources"])
resources(name="java_resources", sources=["*.java", "strip_jar.lock"])

java_sources(name="strip_jar", resolve="strip_jar_dev")

jvm_artifact(
    name="reproducible-builds-jvm-stripper",
    group="net.bzzt",
    artifact="reproducible-builds-jvm-stripper",    
    version="0.9",
    resolve="strip_jar_dev",
    packages=["io.github.zlika.reproducible.**"]
)

then in the plugin I'm trying:

@rule
def generate_strip_jar_lockfile_request(
    _: StripJarToolLockfileSentinel,
) -> GenerateJvmLockfileFromTool:
    return GenerateJvmLockfileFromTool(
        artifact_inputs=FrozenOrderedSet(
            {
                "net.bzzt:reproducible-builds-jvm-stripper:0.9",                
            }
        ),
        artifact_option_name="n/a",
        lockfile_option_name="n/a",
        resolve_name=StripJarToolLockfileSentinel.resolve_name,
        read_lockfile_dest=DEFAULT_TOOL_LOCKFILE,
        write_lockfile_dest="pants-plugins/jvm_stage/strip_jar.lock",
        default_lockfile_resource=(
            "jvm_stage",
            "strip_jar.lock",
        ),
    )

When I'm trying the plugin it fails with:

Engine traceback:
  in select
  in pants.core.goals.package.package_asset
  in jvm_stage.rules.package_jvm_stage (modules/maitred-node/src/unit/maitred/node:main)
  in jvm_stage.strip_jar.resolve_fallible_result_to_strip_jar
  in jvm_stage.strip_jar.strip_jar
  in jvm_stage.strip_jar.build_processors
  in pants.jvm.resolve.coursier_fetch.materialize_classpath_for_tool
Traceback (most recent call last):
  File "/opt/homebrew/Cellar/python@3.9/3.9.12_1/Frameworks/Python.framework/Versions/3.9/lib/python3.9/importlib/resources.py", line 97, in open_binary
    return open(full_path, mode='rb')
FileNotFoundError: [Errno 2] No such file or directory: '/Users/somdoron/git/unit-finance/strip_jar.lock'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/somdoron/.cache/pants/setup/bootstrap-Darwin-arm64/pants.sVxb8A/install/lib/python3.9/site-packages/pants/engine/internals/selectors.py", line 705, in native_engine_generator_send
    res = func.send(arg)
  File "/Users/somdoron/.cache/pants/setup/bootstrap-Darwin-arm64/pants.sVxb8A/install/lib/python3.9/site-packages/pants/jvm/resolve/coursier_fetch.py", line 744, in materialize_classpath_for_tool
    lockfile_bytes = importlib.resources.read_binary(
  File "/opt/homebrew/Cellar/python@3.9/3.9.12_1/Frameworks/Python.framework/Versions/3.9/lib/python3.9/importlib/resources.py", line 126, in read_binary
    with open_binary(package, resource) as fp:
  File "/opt/homebrew/Cellar/python@3.9/3.9.12_1/Frameworks/Python.framework/Versions/3.9/lib/python3.9/importlib/resources.py", line 111, in open_binary
    raise FileNotFoundError(message)
FileNotFoundError: 'strip_jar.lock' resource not found in 'jvm_stage'

What am I missing?

I’m looking at

vcs_version

https://www.pantsbuild.org/docs/generating-version-tags#the-vcs_version-target So, how do I populate a

python_distribution

version from

vcs_version

?

python_distribution(provides=python_artifact(version=???))

I’m using setup.py, but it looks like the target definition is the same for PEP-517 w/ pyproject.toml

Is there any way to pay for enterprise support for pants?

What’s the recommended way to convert the pants-generated

.lockfile

to something like a

requirements.txt

so that I can

pip instsall

it so that my VSCode can figure out the dependency?

Hello Pants Community, I'm fairly new to pants and want to see if anyone has a suggestion for how I can improve the below. I have a project with a few dozen pants targets. I want to check if a given target has changed since a given git tag. My current approach is to call the below once per target and tag combination:

f"./pants --changed-since='{tag}' --changed-dependees=transitive filter --address-regex='{target}'"

This seems to give me what I want but it takes a while to run and, when multiplied by the number of combinations that I have, when i run all of my combinations serially, it adds ~10-15m to my deploy step in my CI, which is much longer than I'd prefer. So, my question is, is there another way to achieve this result which is quicker? My current solution is to run this parallel in separate jobs for each combination but, for various reasons, this isn't really how I'd hoped to solve this.

IS there access to Python built ins in pants macros, or would I need to create a plugin for that?

For example, I want to do something like this:

def myorg_docker_image(**kwargs):
   # If someone specifies a source and does not embed instructions into BUILD file, fail fast
   if kwargs['source'] or not kwargs['instructions']:
       raise ValueError("Cannot use Dockerfiles")
    if not are_dependencies_encoded_in_instructions(kwargs['instructions']):
        raise ValueError("Must encode dependencies of docker image into instructions kwarg")
    docker_image(**kwargs)

unfortunately, raising exceptions isn’t currently supported, although it really should be: https://github.com/pantsbuild/pants/issues/15420

That issue is on my shortlist to look into

in the meantime, you could define an

object

which raises the relevant exception type… trying to find an example

@bitter-ability-32190: yea, that would be helpful.

I guess in a broader sense then, what capabilities do I have in a macro? Can I define functions? Access python built-ins? Or is what I'm trying to do in the realm of "extending an existing target via a plugin" rather than "write a macro"

the example you’ve given should definitely be in scope for a macro: it’s an oversight that you can’t raise exceptions… the builtins in scope are intentionally quite limited, but we should definitely punch holes through for this kind of case.

Can I make the target fail in some other way then?

Does

--build-file-prelude-globs

with a file that just does `from builtins import *`work?

so: here is an example of defining an `object`: https://github.com/pantsbuild/pants/blob/9ca84b5eb6ff66e224ee4f4e0a5a7aec0c922843/src/python/pants/backend/python/register.py#L57-L58 … the alias you give will be in scope in all

BUILD

files. so as a workaround for https://github.com/pantsbuild/pants/issues/15420, you could define an

object

for an exception type

@enough-analyst-54434: i don’t think so, because

import

is banned

Even in a prelude?!

maybe not... i don’t see enforcement there.

can

--python-requirement-constraints

be used with

--no-enable-resolves

?

enable_resolves

is set in

pants.toml

but i want to override and set it to false when running a specific CI command that uses

--python-requirement-constraints

Just tried

from builtins import *

and it doesn't appear to work. Little confused by the example (first foray into customizing pants) but I'll see if I can poke at it for a bit and make something work

does the

objects

example make sense? it would involve creating a very tiny plugin, with a

register.py

file. see https://www.pantsbuild.org/docs/plugins-overview#enabling-plugins-with-registerpy (in particular, the second tab in that code block), and defining a

def build_file_aliases

function in it.

Don't try this at home:

$ cat expose-builtins 
exec(f"from builtins {'IMPORT'.lower()} *")

$ cat macro-uses-builtins 
def foo(name=None):
  if not name:
     raise Exception("A name is required.")
  python_sources(name=name)
   
$ cat BUILD.example 
foo(name="jake")
foo()

$ ./pants --build-file-prelude-globs='*-builtins' list :
15:12:53.20 [ERROR] 1 Exception encountered:

Engine traceback:
  in select
  in pants.backend.project_info.list_targets.list_targets
  in pants.engine.internals.specs_rules.resolve_addresses_from_specs
  in pants.engine.internals.specs_rules.resolve_addresses_from_raw_specs
  in pants.engine.internals.specs_rules.addresses_from_raw_specs_without_file_owners
  in pants.engine.internals.build_files.ensure_address_family ()
  in pants.engine.internals.build_files.parse_address_family ()
Traceback (most recent call last):
  File "/home/jsirois/dev/pantsbuild/pants/src/python/pants/engine/internals/mapper.py", line 47, in parse
    target_adaptors = parser.parse(filepath, build_file_content, extra_symbols, defaults)
  File "/home/jsirois/dev/pantsbuild/pants/src/python/pants/engine/internals/parser.py", line 167, in parse
    exec(build_file_content, global_symbols)
  File "<string>", line 2, in <module>
  File "macro-uses-builtins", line 3, in foo
    raise Exception("A name is required.")
Exception: A name is required.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/jsirois/dev/pantsbuild/pants/src/python/pants/engine/internals/selectors.py", line 705, in native_engine_generator_send
    res = func.send(arg)
  File "/home/jsirois/dev/pantsbuild/pants/src/python/pants/engine/internals/build_files.py", line 181, in parse_address_family
    for fc in digest_contents
  File "/home/jsirois/dev/pantsbuild/pants/src/python/pants/engine/internals/build_files.py", line 181, in <listcomp>
    for fc in digest_contents
  File "/home/jsirois/dev/pantsbuild/pants/src/python/pants/engine/internals/mapper.py", line 49, in parse
    raise MappingError(f"Failed to parse ./{filepath}:\n{e}")
pants.base.exceptions.MappingError: Failed to parse ./BUILD.example:
A name is required.

Ah, just got it to work! THanks for the help here, the example does make sense. I had some files in the wrong place for a minute but just got it all sorted

So tis is nasty, but backwards compatible. When we support builtins, just remove the

expose-builtins

thing.

Also, you'll want to ensure that is the 1st prelude glob. It must be parsed before those others that will use it.

I think.

I guess not if the raises are all inside functions.