Good question. There were some notorious cases where trolling or pileons made it advisable to shut down a thread altogether, and folks lobbied GH for locking as a solution for that. Definitely mark threads as "Answered". For now, let's reserve locking for situations where rekindling a conversation would be inappropriate for whatever reason. We can always use it more freely later if we want. But a good starting default is openness to the extent good community behavior facilitates it. If someone wants to resume conversation on a stagnant Discussion, it likely isn't more harmful than resuming conversation on a stagnant Issue. And GH gives us tools to move it if need be.

Note: "Answered" status appears to only be available for threads categorized as "Q&A". So one way to deal with this would be to move a concluded brainstorm such as https://github.com/pantsbuild/pants/discussions/17289 over to Q&A and then mark it "Answered". Testing...

Yep. Changing category works. If there isn't one comment that clearly stands out as a thread's "answer", this is a good cue that a wrapup comment would be a helpful addition by summarizing what's been concluded. Then the summary comment can be marked as "Answer".

good idea - thanks!

I like that it's gently coaxing us toward being a little more explicit about decisions. Institutional knowledgebases ftw.

@witty-crayon-22786 I’m in the process of migrating

publish

to be a

USES_ENVIRONMENTS

goal. It looks at the moment like I can’t specify the environment in the

await Effect

line. Is this intentional?

as i debug a cache-related inconsistency in the rust bootstrap, i feel compelled to clarify that it wouldn’t be impossible for Pants to dogfood it’s own docker support + remote exec for all of wheel building and testing. it would just require what the bootstrap problem usually requires (afaik): using a stable or pinned binary version of Pants to build Pants, rather than Pants-from-sources.

@sparse-lifeguard-95737, @bitter-ability-32190, @ancient-vegetable-10556: so! i think that you all have the largest (pending?) changes targeted for 2.15.

As we evaluate GitHub Discussions, please keep an eye out for posts with zero replies in its Q&A and Development categories (channels). Let's show off to the general public the responsiveness that folks here get to enjoy on a regular basis. Thanks, everyone! What you bring to this community is so special. I've never known an open source community as kind and attentive as this one. It's exciting.

@hundreds-father-404: there are twoish more PRs bound for

2.14.x

: i can probably do another release candidate + stable over the weekend if you’ll be out

I’ve hit a weird error running

./pants check

on

pantsbuild/pants

twice today - the error doesn’t repro if I immediately re-run the command, so it could be a race condition?

23:55:38.23 [ERROR] Completed: Typecheck using MyPy - mypy - mypy failed (exit code 1).
src/python/pants/goal/anonymous_telemetry.py:14: note: In module imported here,
src/python/pants/core/register.py:54: note: ... from here,
src/python/pants/core/goals/deploy_test.py:17: note: ... from here:
/Users/dan.moran/.cache/pants/named_caches/pex_root/venvs/07570716880b83f98097b50a2089bac0b0397777/bfaddbeef852a86567e6233311d5ce9c0ceb52e9/lib/python3.7/site-packages/humbug/report.py:20:1: error: Library stubs not installed for "requests" (or incompatible with Python 3.7)  [import]
    import requests
    ^
/Users/dan.moran/.cache/pants/named_caches/pex_root/venvs/07570716880b83f98097b50a2089bac0b0397777/bfaddbeef852a86567e6233311d5ce9c0ceb52e9/lib/python3.7/site-packages/humbug/report.py:20:1: note: Hint: "python3 -m pip install types-requests"
/Users/dan.moran/.cache/pants/named_caches/pex_root/venvs/07570716880b83f98097b50a2089bac0b0397777/bfaddbeef852a86567e6233311d5ce9c0ceb52e9/lib/python3.7/site-packages/humbug/report.py:20:1: note: (or run "mypy --install-types" to install all missing stub packages)
/Users/dan.moran/.cache/pants/named_caches/pex_root/venvs/07570716880b83f98097b50a2089bac0b0397777/bfaddbeef852a86567e6233311d5ce9c0ceb52e9/lib/python3.7/site-packages/humbug/report.py:20:1: note: See <https://mypy.readthedocs.io/en/stable/running_mypy.html#missing-imports>

hey folks, for testing, if I have a rule, I can test it with `rule_runner.request(...)`; is there something similar for a rule_helper?

CI uh oh:

Error: Unable to locate the previous sha: 520aee682f57a9da70ddab51afe8f95f1b1525b5
  Error: You seem to be missing 'fetch-depth: 0' or 'fetch-depth: 2'. See <https://github.com/tj-actions/changed-files#usage>
  Error: Process completed with exit code 1.

https://github.com/pantsbuild/pants/actions/runs/3304088306/jobs/5452778141

Also 👋 I have a few PRs feel lonely of late 🙂 Namely: https://github.com/pantsbuild/pants/pull/17223 and https://github.com/pantsbuild/pants/pull/17240

Forecast: • (I think my 2.15 contributions are about up. Fix/fmt/etc... are shaping up) • Need to deprecate

update-build-files

• Probably another docs pass • I have PRs ready to add

http_sources

and add per-platform

exe

fields to

external_tool

once in • Still working on symlink support. Oof Future: •

pants.toml

fixer

A fun new CI failure: https://github.com/pantsbuild/pants/actions/runs/3313745258/jobs/5472143278

ERROR: No matching distribution found for setuptools<64.0,>=63.1.0

Hey could we turn off Slack announcing new discussions in #general please? That is noisy imo. GitHub now supports very customizable notification settings where you can ask GH to notify you already for that, but ignore other things like new issues

I know we're anti-new Pants deps (for good reason), but the deprecation fixers make me want to use

libcst

for Python and

tomlkit

for

pants.toml

. Thoughts?

I have a slightly odd scenario, but fishing for thoughts (Python + GPU deps) I have

<http://repository.in|repository.in>

that lists most of my deps, minus

mxnet

. -> it gets locked manually (e.g. I call

pex3

) and toss in `mxnet`for the lock. Then I have a

python_requirement

for

mxnet

. Lastly I have a package

mxboard

which has an implicit, undeclared dependency on

mxnet

. So I would normally do:

python_requirements(
    ...,
    overrides={"mxboard": {"dependencies": [":repository#mxnet"]}}",
)

Buuuuuuuuuut In order to support GPU

mxnet

I also have more lockfiles/resolves which swap

mxnet

with

mxnet-cu1XX

for each CUDA version we care about. Plus an additional

python_requirement

for each new resolve. But then my hardcoded dependency breaks and I weep. (All hard-coded deps break, but I despise them. This one is most relevant). So I'm thinking we should make it easy to say "this package relies on this module (name)" that isn't specifically "this

python_requirement

depends on this other `python_requirement`". That way the resolve aspect is delayed to dependency injection plugin time. Or any other way to unblock this 🙂

Does anyone have tips for debugging rule-graph errors coming from a

RuleRunner

in a test? I’m trying to revamp the tests for

test

as part of the batching impl, and I think I’ve reached the limit of my guess-and-check strategy 🤯

It occurs to me that now we're using discussions, the line between "is this a blog post" or "is this a discussion" is blurier. Anyone wanna help me crisp it? I have 2 ideas I'd love to blog about, but nothing stopping me from just making a discussion. Discussion seems less "hey look at me" though 🤔

@happy-kitchen-89482 @bitter-ability-32190 @hundreds-father-404 I’ve finally gotten the batched-test plumbing updated with latest

main

- you’d all reviewed the initial draft, if you want to take another look: https://github.com/pantsbuild/pants/pull/17134

🤔

1145 |           target: target.strip_prefix(name.as_ref()).unwrap(),
     |                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ expected struct `RelativePath`, found struct `Path`
     |
     = note: expected reference `&RelativePath`
                found reference `&Path`

But

target

is of type

&RelativePath

and

strip_prefix

returns `&RelativePath`: https://docs.rs/relative-path/1.5.0/relative_path/struct.RelativePath.html#method.strip_prefix What am I missing?

@ancient-vegetable-10556: release blog so far: https://docs.google.com/document/d/13LS91KBPhOGrJJGUAzqyEoeWYo2GDgGQPxUCw9Mgb_o/edit# … i still need to clean up the bullet points at the end and add a title, but if you want to suggest edits, please do.

some feedback. I’ve not yet got to grips with the new

EnvironmentName

stuff.. so doesn’t understand much of what it does in detail yet, given that, I added a new

union

type bare bones first, I expected it would need the

in_scope_types

thingy, but skipped it first to see if it would work. Now, I can say it did not, but the error was what looks like the entire rule graph spewed out with a lot of the infamous “No installed rules return …” lines for all of the rules/types. Adding in the missing thingy mahjoob on the union solved it, but I would’ve never have been able to arrive at that conclusion based on the error I observed 😛 Just hoping this will be something we can improve a bit… 😉

hiyas, do we have a standard way of benchmarking Pants? I'd like to make sure that my changes haven't caused a slowdown.

haven’t done the PEX release for 2.14.0 yet, but will in about 30 minutes

just opened https://github.com/pantsbuild/pants/issues/17354 to help describe a pitfall around where

EnvironmentNames

are consumed that will likely need to be present in 2.15.x at this point, but that we could maybe fix in 2.16.x.

@flat-zoo-31952: we don’t really have a convention yet, but here is a “stabilization tracking” ticket for Environments: https://github.com/pantsbuild/pants/issues/17355 … which we can reference in help strings, etc