@witty-crayon-22786 maybe a juicy one?: https://github.com/pantsbuild/pants/issues/15653 Let me know if you have experiments I should run.

Recently I’ve started to get issues with the pre-commit hook, throwing this error:

Exception: Unmatched glob from CLI arguments: ".github/workflows/*"

but I can’t find where that argument is coming from.. and what’s more is that I do have files there, so it’s somthing fishy going on:

$ tree .github/workflows/
.github/workflows/
├── audit.yaml
├── cache_comparison.yaml
├── cancel.yaml
├── test-cron.yaml
└── test.yaml

0 directories, 5 files

omg epic. now that

update-build-files

works with the new CLI specs, you can incrementally migrate things. Which gives me a related idea, allow you to turn off individual fixers, even if we default to True. So you can still keep running

update-build-files

to format BUILD files etc, but only use the deprecation fixer when ready

Proposal: Elide

file

and just use

resource

everywhere. Google Doc: https://docs.google.com/document/d/1pGib2ZJiyltfij0aZMm81S2XbATa8s6_ghuLnf2Xr_k/edit?usp=sharing

hashable datastructures as rule inputs and outputs 🧵

as it stands,

--lockfile-generator=pex

is the default in

2.12.x

, and so will gain some more exposure in

2.12.0

. since the deprecation required the flag to be set explicitly, this will only really affect new users, or those who opt in. but we should likely still be cautious about timing to ensure that capable folks (read: @enough-analyst-54434) are likely to be around for the potential uptick in new issues. since

2.12

doesn’t have a whole lot else in it, we probably have a bit more flexibility around timing… and so could maybe agree on a rough date now (subject to bumps for blockers). cc @hundreds-father-404, @bitter-ability-32190, @happy-kitchen-89482

i’ve added a very simplistic Maintainer onboarding section to the

2.12

docsite… let me know if i’ve missed anything, and i’ll add it.

May I have an approval to run CI please: https://github.com/pantsbuild/pants/pull/15684 hmm. Looks like I need labels before CI will run. 🤷

Hmm. I'm trying to run

./pants

in a checkout of the pants repo, but I'm getting an ugly

pex

file not found error. 🧵

Interesting thread giving a lot of varied perspectives on "developer experience" and what expectations the term conjures up. https://twitter.com/kelseyhightower/status/1527672023370985478?s=20&t=kiLzlRRlTRuTO01fH9RzGQ

The

name=

proposal is ~rewritten to no longer propose the intermediate step of always requiring

name=

. Instead, deprecate the 5 ambiguous cases, then once that is over we can safely switch to new semantics. https://docs.google.com/document/d/1jSCJ3pwvIMg5duafsr8KmiLOO_2ARMaYCSrb8K85JQc/edit#

Does anyone disagree with this change, to make

--no-process-cleanup

also preserve the tmpdir that the

run

goal runs your process in?

Also, https://github.com/pantsbuild/pants/pull/15689 is a long-requested feature to allow

run

to use the sources directly from the repo instead of sandboxing them. This is necessary for, e.g., Django's

makemigrations

, and other processes that operate on the repo itself and compute output locations relative to the sources it's running on

what is your platform? we have this clause

if is_macos_arm; then
  echo "* Running \`./cargo clippy\`"
  ./cargo clippy || exit 1
else
  echo "* Running \`./cargo clippy --all\`"
  ./cargo clippy --all || exit 1
fi

macos x86 - I didn't touch rust code, it was just a new python backend

Will wipe the working directory, re-clone and try again tonight, but this was the same working dir Ive been using for weeks, and committing to without problems. Pulled from upstream that day (only a handful of commits behind), and then my next git commit got bodychecked by the pre-commit script

Wiped everything, restarted, and didn't get hassled about this 🤷

I’ve been playing with the POC Explorer UI a bit again https://github.com/pantsbuild/pants/pull/14348 🧵

Initial attempts at trying out multiple resolves… 🧵

I'm running into an error with the coursier fetch script (in the pants repo), because my venv path has special characters in it.

__coursier/coursier_fetch_wrapper_script.sh: line 11: syntax error near unexpected token `R'

Line 11 in the script (from

--no-process-cleanup

) is:

/home/cognifloyd/.cache/pants/pants_dev_deps/Linux.x86_64.Intel(R).Core(TM).i7-3610QM.CPU.@.2.30GHz.py37.venv/bin/python __coursier/coursier_post_processing_script.py "$json_output_file"

So, bash doesn't like the

()

chars it seems. I'm not sure why I have all of those chars in the venv name, does anyone know where that comes from? The naive fix is to surround

{python_path}

with quotes

'{python_path}'

here: https://github.com/pantsbuild/pants/blob/main/src/python/pants/jvm/resolve/coursier_setup.py#L83 Is there a better way to escape that? eg what happens if a path has a

'

in it (why I don't know. But I don't know why I have

()

chars either)? Also, is there anything else in those scripts that might need shell escaping?

FYI the

mypy/rules_integration_test

fails on py2 tests for me with:

The typed_ast package is not installed.\nFor Python 2 support, install mypy using `python3 -m pip install "mypy[python2]"`Alternatively, you can install typed_ast with `python3 -m pip install typed-ast`

Informal proposal: rename "process-execution" prefix for sandboxes to "pants-sandbox". It's much clearer in intent, and I think is less chars

I got a weird

sphinx/rules_integration_test

error in CI https://github.com/pantsbuild/pants/runs/6663224290?check_suite_focus=true#step:11:203

looks like we missed releases on Friday: i’m on duty this week, so can probably run them today/tomorrow.

Please get anything you want in 2.13.0.dev4, 2.12.0rc2 or 2.11.1rc2 in the relevant branches ASAP, as I will be kicking off release soon. If there's anything I should hold for, let me know now.

https://pantsbuild.slack.com/archives/C046T6T9U/p1654020278507039 is a useful anecdote i think about why untyped dependencies can be confusing. dependencies of Python targets and a bunch of other positions (docker in this case) are unchecked in general, so you don’t get an error if you have a dependency that doesn’t make sense.

Release PRs:

Please no merging into main until I land the release PRs, thanks!!

Re releases: 2.11.1rc2 and 2.13.0.dev4 are out (@hundreds-father-404 when you have a moment can you do the M1 releases?), 2.12.0rc2 is waiting on wheel builds, and I have to get on a plane soon, so it may be delayed until tomorrow morning PT. I'll do all the announcements together.

I opened an RFC in the form of a PR. It is to allow

./pants run ...

on a

python_source

, freeing users from requiring the declaration of a

pex_binary

for scripts, and frees us to implement

pex_binary

running as

package + run

.