black-formatter

[plain atlas]

!pypi pytest-cov

[late dew]

pytest-cov v2.12.1

Pytest plugin for measuring coverage.

[bright glacier]

and?

[plain atlas]

¯_(ツ)_/¯

I just wanted to let you know of it, if you didn't already

[bright glacier]

https://github.com/ichard26/blackbench/blob/e40f3b5497829f877584e8e001ccad7140b37cd9/requirements/tests-cov.txt#L1

[late dew]

requirements/tests-cov.txt line 1

pytest-cov>=2.12.1```

[plain atlas]

lol

[bright glacier]

I started another mini-project so my open source time has been limited

[plain atlas]

awh, you started a closed source project? 😔

[bright glacier]

I basically started a new and improved version of black-primer using ideas from mypy-primer

which originally took ideas from black-primer

so in a way, I'm trying to be even better than mypy-primer

[plain atlas]

lol

[bright glacier]

nope, just another tooling project for psf/black as usual

[plain atlas]

ah lmfao

I have uh

2 repos for one project as it stands right now, soon to be 3 or 4 repos lolol

[bright glacier]

the first major improvement is much prettier output (although if I want to be mostly feature compatible with black-primer this may have to go, but hopefully not)

[bright glacier]

that sounds like a pain to manage

[plain atlas]

its not

one repo is the main project

but due to the nature of the project, default assets will likely on a seperate repo

configurave is a repo too

and then there will project extensions in another repo

[bright glacier]

ah, so this is how you were going to rope me into talking about configurave

[plain atlas]

eh

tl;dr configurave is basically a helper to be able to define a configuration and have it sourced from .env, terminal environment vars, multiple toml files, and other sources

[plain atlas]

no! why would I do that?

configurave is still in alpha and was written in 30 minutes out of annoyance, or so I was told.

[bright glacier]

I am quite slow when it comes to anything

that sort of development speed sounds amazing

[plain atlas]

lol

its still extremely buggy and not exactly usable

[bright glacier]

diff-shades is like 513 lines of somewhat maintainable code

it's basically the core logic of the file that's sorta of a mess

[plain atlas]

😔

diff shades sounds cool

[bright glacier]

it's my take on black-primer

why shades? shades of black

[plain atlas]

ah

lol

huh

you just taught me something lol

i did not realise that http.server could easily just... serve... a directory

[bright glacier]

looking at next-pr-number ?

[plain atlas]

yes

i'm now adding an action to my taskipy tasks

python -m http.server --directory htmlcov -b 127.0.0.1

[bright glacier]

I just use alias wb "python -m webbrowser"

[plain atlas]

lol

reason to have the taskipy tasks is now its for everyone on the project

oh

I've been meaning to ask

how can I make black ignore a folder

current config

[tool.black]
line-length = 110
target-version = ['py38']
include = '\.pyi?$'

[bright glacier]

--exclude=/folder/

[plain atlas]

so

exclude = ["/folder/"]?

[bright glacier]

no, exclude = "/folder/"

it's a regex

[plain atlas]

how to exclude multiple folders, in the toml then?

[bright glacier]

https://github.com/psf/black/blob/b1d060101626aa1c332f52e4bdf0ae5e4cc07990/pyproject.toml#L12-L19

[late dew]

pyproject.toml lines 12 to 19

extend-exclude = '''
/(
  # The following are specific to Black, you probably don't want those.
  | blib2to3
  | tests/data
  | profiling
)/
'''```

[plain atlas]

i uh, need to read more docs because

target-version = ['py36', 'py37', 'py38']

I was setting this to the lowest version supported by the project its formatting

[bright glacier]

honestly, I don't know why black needs all of the versions

[plain atlas]

but you work on it

[bright glacier]

there's an issue on the tracker about it, but we semi agreed that making it a lower bound is probably a good idea

[plain atlas]

I thiiink its so it tests its valid ast on all of those interpreters if it can find one??

i have no idea

that's something that would make sense if that's what it does

[bright glacier]

I'm not 100 sure either

3.6 and beyond is pretty much the same

[plain atlas]

:=

[bright glacier]

other than some syntax changes, nothing special

[plain atlas]

https://github.com/psf/black/blob/91773b89097927a2393bc5223295d37ed26e1632/src/black/parsing.py#L37-L48

[late dew]

src/black/parsing.py lines 37 to 48

def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]:
    if not target_versions:
        # No target_version specified, so try all grammars.
        return [
            # Python 3.7+
            pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords,
            # Python 3.0-3.6
            pygram.python_grammar_no_print_statement_no_exec_statement,
            # Python 2.7 with future print_function import
            pygram.python_grammar_no_print_statement,
            # Python 2.7
            pygram.python_grammar,```

[plain atlas]

seems like its for those

so like

exec()
print()

[bright glacier]

it's also used for Feature.*

so black knows whether it can do a certain thing or not

[plain atlas]

yeah

even though its not really supposed to change the ast

[bright glacier]

3.4 and 3.5 introduced trailing comma support for func signatures and calls

[plain atlas]

but sadly does change the ast in a few places

like

isn't --strict supposed to keep ast identical?

or was that added as a result of that conversation

[bright glacier]

we don't have such a flag right now

[plain atlas]

did it at some point?

[bright glacier]

never did

[plain atlas]

uh

I need to go open 20 tabs to check which of dev tools I remember strict from

[bright glacier]

other than docstrings, black is safe

it's probably mypy

[plain atlas]

i don't use mypy LOL

[tool.poetry.dev-dependencies]
black = "^21.7b0"
codecov = "^2.1.11"
coverage = { extras = ["toml"], version = "^5.5" }
flake8 = "~=3.8"
flake8-annotations = "~=2.3"
flake8-bandit = "^2.1.2"
flake8-bugbear = "~=20.1"
flake8-docstrings = "~=1.5"
flake8-isort = "^4.0.0"
flake8-string-format = "~=0.3"
flake8-tidy-imports = "~=4.1"
flake8-todo = "~=0.7"
isort = "^5.9.2"
pep8-naming = "~=0.11"
pre-commit = "~=2.1"
pytest = "^6.2.4"
pytest-asyncio = "^0.15.1"
pytest-cov = "^2.12.1"
pytest-dependency = "^0.5.1"
pytest-docs = "^0.1.0"
pytest-xdist = { version = "^2.3.0", extras = ["psutil"] }
taskipy = "^1.6.0"

[bright glacier]

we added this docstrings handling code that was supposed to handle the case where black is converting 2-space indentation to 4-space incorrectly

but it turns out it touches a lot of whitespace within docstrings

[plain atlas]

related: i cannot wait for poetry 1.2.0 to release, for the dependency groups
Can move all 7+ test deps above to a second group

[plain atlas]

Smh

[bright glacier]

yeah, you're not the first person to complain about it

[tired shard]

uh uh uh waht, I didn't know about this

[bright glacier]

I personally don't think we should be even touching docstrings (except for maybe the quotes placement) since everything in 'em matter

but I don't feel strongly, I don't really make these sort of decisions here

[plain atlas]

I'll explain in the other server

[plain atlas]

Yeah same

[dense jungle]

@bright glacier I didn't read all of the above but mypy ended up pinning tomli to a version that doesn't have the annoying CR thing, that may make your life easier too with mypyc

[flat kraken]

nipunn1313

[psf/black] Pull request opened: #2414 Present a more user\-friendly error if \.gitignore is invalid

Fixes #2359

JelleZijlstra

[psf/black] Pull request review submitted: #2414 Present a more user\-friendly error if \.gitignore is invalid

Thank you!

[flat kraken]

JelleZijlstra

[psf/black] Pull request review submitted: #2412 Docker image usage description

[flat kraken]

MarcoGorelli

[psf/black] Issue opened: #2415 pre\-commit example shouldn't encourage deprecated practice

From the pre-commit docs

Using a branch name (or HEAD) for the value of rev is not supported and will only represent the state of that mutable ref at the time of hook installation (and will NOT update automatically).

I'd suggest that the docs just use the latest tag in their example

MarcoGorelli

[psf/black] Pull request opened: #2416 document jupyter hook

closes #2415
addresses part of #2413

[stuck vapor]

woohoo, thanks for the merge!

I've made a PR to address the first TODO you've brought up, but I don't understand the second one.

just document the quirk

Which quirk are you referring to?

[stuck vapor]

For the purpose of documenting the new pre-commit hook, what would be the next version of black? 21.8b0?

[red sedge]

(we can always deal with any remaining issues as they get reported).

Yes!

[half hearth]

I really like the code formatter

[sour jay]

Well, there are styling issues in doc-strings you should uniform. Namely:

"""Testing something

This function tests something
"""

Vs ```py
"""
Testing something

This function tests something
"""

And well, anyone using anything else in terms of styling of doc-strings.

[red sedge]

I read docstrings in slow mode so their formatting doesn't really bother me

[sour jay]

I mean that's true, it's not a styling thing that bothers me as much as other stuff. But it's something I definitely notice and should be enforced by an opinionated doc-string.

[red sedge]

There's something enforcing it in most codebases I'm using. Maybe flake8?

[slim olive]

There's pydocstyle which can be plugged into flake8 with flake8-docstrings

[red sedge]

Yeah I'm pretty sure I'm not using it...

I think people are just formatting these by hand

[slim olive]

Yes, pydocstyle only checks the style and complains if it's wrong

it doesn't know how to fix it

I'm not aware of any commonly used docstring formatters... it probably makes sense that this is something humans should do

[spring matrix]

:lem

[plain atlas]

Hey hey black team, does black load the entirety of the pyproject file or can it just load the part that it needs?

[bright glacier]

That sounds more like a feature of the TOML library we're using, and as far as I know tomli does not support that

[austere lava]

here's a progress update on the libcst rewrite: https://github.com/Instagram/LibCST/issues/285#issuecomment-894699168

[bright glacier]

Oooo nice! I wonder how hard that last 1% is going to be 🙂

# HACK: I know this is hacky but the benefit is I don't need to copy and
# paste a bunch of black's argument parsing, file discovery, and
# configuration code. I also get to keep the pretty output since I can
# directly invoke black.format_file_contents :D


def get_project_files_and_mode(
    project: Project, path: Path
) -> Tuple[List[Path], black.FileMode]:
    import black

    files = []
    mode = None

    def shim(sources: List[Path], *args: Any, **kwargs: Any) -> None:
        nonlocal files, mode
        files.extend(sources)
        mode = kwargs["mode"]

    with suppress_output(), patch("black.reformat_many", new=shim):
        black.main([str(path), *project.custom_arguments], standalone_mode=False)

    assert isinstance(mode, black.FileMode)
    return files, mode

One of the more hacky bits of code I've written lately 😛

[austere lava]

I'm hoping not very hard 😁

[plain atlas]

Yes, I was asking what black does lolol

[flat kraken]

ltskv

[psf/black] Issue opened: #2417 Trailing comma after last function argument is occasionally missing

Sometimes black doesn't add a trailing comma after the last function argument. This happens e.g. when there is a star (*) argument in the argument list.

black_trailing_comma_bug_1.py

def a_function_with_a_star_argument(*, aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa, bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb):
    pass


def a_function_without_a_star_argument(aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa, bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb):
    pass

$...

[flat kraken]

michaelmiscanuk

[psf/black] Issue closed: #2402 Nested Tuples are not well formatted

[flat kraken]

JelleZijlstra

[psf/black] Issue closed: #2417 Trailing comma after last function argument is occasionally missing

[mint barn]

what is formatter? what the function of it?

[plain atlas]

yeet

wow

okay

https://github.com/Azure/azureml-examples/blob/2d23f9cb78eca83ab6b6f8d26c229edbf81eafcd/dev-requirements.txt#L2

[late dew]

dev-requirements.txt line 2

black==20.8b1```

[heavy heron]

The goal of a formatter is to change the structure of your source code to conform to certain rules, so you can have a common style everywhere

[flat kraken]

cooperlees

[psf/black] Pull request review submitted: #2412 Docker image usage description

All sounds good to me. Many thanks for this! Once @JelleZijlstra is happy lets merge.

[flat kraken]

spaceone

[psf/black] Issue opened: #2418 Black produced different code on the second pass of the formatter

error: cannot format schoolroom.py: INTERNAL ERROR: Black produced different code on the second pass of the formatter. Please report a bug on https://github.com/psf/black/issues. This diff might be helpful: /tmp/blk_gfkfesa1.log
Oh no! 💥 💔 💥
1 file failed to reformat.

black, version 20.8b1

Mode(target_versions={}, line_length=105, string_normalization=True, experimental_string_processing=False, is_pyi=False)
--- source
+++ first pass
@@ -169,25 +169,29 @@
             {...

[rotund cove]

what is black fromatter

[plain atlas]

!pypi black

[late dew]

black v21.7b0

The uncompromising code formatter.

[flat kraken]

spaceone

[psf/black] Issue closed: #2418 Black produced different code on the second pass of the formatter

[plain atlas]

@dense jungle what was the bug?

[dense jungle]

I think it was the big magic trailing comma thing that Lukasz eventually fixed by just formatting twice

[plain atlas]

👀

[bright glacier]

And this is why I don't run black-primer locally, running black over the whole project selection takes 30+ minutes 😮 (although this is actually diff-shades in this screenshot)

[delicate sky]

just format correctly to begin with 😉 /s

[bright glacier]

I'm confused, what do you mean by that?

It's kinda half the point of diff-shades & black-primer to run on projects that have never interacted with Black so we can catch issues never seen before.

Unless the joke is that Black's style isn't good enough?

[delicate sky]

im just joking lol

[bright glacier]

My point was that to me it wasn't the best joke since it wasn't clear 😉

[delicate sky]

ahh, i didn't really look much at what you said
diff-shades meaning it is only doing a dry-run or what does it mean in this case?

[bright glacier]

so diff-shades / black-primer (the latter is the current utility in use, although I wrote the former to be an much improved version) basically takes a list of open source projects, clones them, runs black over them recording the results

[delicate sky]

ahh interesting

[bright glacier]

we use it to compare differences in behaviour between versions of black, check how big the impact of something will be, etc.

[delicate sky]

is there a black command to see if running black would cause any changes
potentially want to add this as part of a CI/CD pipeline to make sure that black was run for the whole project

ahh nvm i see --check

[swift warren]

-c key, I believe

[delicate sky]

yeah, thanks

[plain atlas]

so having large code bases not blacked is actually good for black, interesting...

[bright glacier]

That's the short form of --code not --check FYI

[dense jungle]

Not necessarily. We want to see how changes in Black itself affect its formatting

[bright glacier]

black-primer is much better suited at gauging the impact of black on projects in general while diff-shades aims to better gauge the differences between two versions of black. The former is helpful to us too but it's not as helpful. Especially since that sort of data is more for when you're (considering) adopting Black BUT not when you're a developer of Black.

[nimble gull]

how can I tell black I would like no line length limit?

tried black -l 0 and black -l -1

[errant barn]

black -l 9999999999999999999999999 shoud get the job done at least :D

But on a more serious note, I don't think that's an option we want to provide to users. Where would that be beneficial?

[lament crow]

if there is no line length limit, what's even there to style, quotes? :)

if someone dared to use multiple lines for something, black would just join it back together in a single line lol

I suggest no magic trailing comma handling for the full hands-on experience of lines with no limits /s

[elfin egret]

😂 😂

[nimble gull]

Where would that be beneficial?

I'd rather scroll to the right in my editor than see the tool try to apply line wrapping. I don't know why my config file should have some large number of nines in a row to get this.

Why not just interpret -l 0 as no length enforcement?

[errant barn]

Fair enough! Can't argue with preference, but I think the overwhelming majority of people would prefer a reasonable line length limit and wrapping lines longer than that. -l -1 or such could encourage using it, and since Black is opinionated, I don't think we should do that. Does that make sense? Out of curiosity, do you prefer no wrapping in every scenario (list items, strings, long nested collections or calls), or a specific one?

[nimble gull]

I generally think not wrapping makes my code more readable and easier to understand how the program flows. I really like being in control of the wrapping by placing , at the end of the thing and I do that in the cases when I think each element on its own line makes understanding easier.

[errant barn]

I can definitely see your point, but I'm afraid you're swimming against the current because we try to not take previous formatting into account 😅

[nimble gull]

I mostly place the a , at the end for function calls where I set a bunch of kwargs

oh well, not the end of the world to have some nines together in my pyproject.toml!

[errant barn]

Black is not for everyone, especially if you like to be the one in control :D

[errant barn]

That was a joke, please don't actually do it

[nimble gull]

haha! too late!

why not remove the line limit config?

[errant barn]

What do you mean?

[nimble gull]

(I actually had it this way to begin with)

[nimble gull]

I was just saying if you think people shouldn't do

[tool.black]
line-length = 999999

Then maybe limit the allowable max line length limit or remove the line-length option all together! I mean Henry Ford only offered one shade of black, right? 😉

Or rename the tool to "black (with a configurable opacity)" 😄

[errant barn]

Touché, but the exact value has been a bit controversial :D see https://black.readthedocs.io/en/stable/the_black_code_style/current_style.html#line-length

[nimble gull]

I love the concept of the tool and I'm sick of wading through diffs, so I'm thinking today that the loss of control is worth it!

[errant barn]

That's a realisation I've come to as well 😄

We are waiting with open arms in the "no formatting worries" land for you

[nimble gull]

\o/

One other black question I have: is it a bug if black doesn't enforce the configured line length limit?

Like, should it break up long strings?

[errant barn]

Pass in --experimental-string-processing and it will! It's being developed.

[nimble gull]

ah cool

[dense jungle]

more generally we do consider it a bug when Black doesn't obey the line length limit. such bugs are tracked in https://github.com/psf/black/issues?q=is%3Aopen+is%3Aissue+label%3A"F%3A+linetoolong"

[flat kraken]

MarcoGorelli

[psf/black] Pull request opened: #2419 Pipfile lock

xref #2413

I added jupyter to Pipfile and then, in a new virtual environment, ran pipenv install --dev, and committed. Is this what's required for that issue?

[bright glacier]

whelp.

[plain atlas]

tl;dr github crashed

[bright glacier]

yeah, was trying to look at a pypa/pip PR and it 500ed twice, looks pretty crashed to me

oh wow, all of the services are yellow

👀

IIRC it wasn't actually super clear whether pip would do this, there was an issue where automated code formatters were being debated. I'm impressed and surprised there isn't as much pushback.

Is it just me or do we all feel a bit more responsible for not messing up black's style now 😅

Oh well, I'm sure we'll be informed very thoroughly if we get it wrong :p

[sour jay]

noooooo not GitHub ;-;

[flat kraken]

MarcoGorelli

[psf/black] Issue closed: #2415 pre\-commit example shouldn't encourage deprecated practice

[bright glacier]

❯ pyperf compare_to fmt-fast-clang.json fmt-fast-gcc-10.json -G
Slower (15):
- fmt-fast-flit/sdist: 84.6 ms +- 5.2 ms -> 95.8 ms +- 7.4 ms: 1.13x slower
- fmt-fast-black/output: 38.4 ms +- 2.7 ms -> 43.1 ms +- 3.8 ms: 1.12x slower
- fmt-fast-black/__init__: 409 ms +- 22 ms -> 453 ms +- 31 ms: 1.11x slower
- fmt-fast-black/lines: 283 ms +- 17 ms -> 313 ms +- 21 ms: 1.10x slower
- fmt-fast-comments: 37.5 ms +- 1.9 ms -> 41.3 ms +- 3.1 ms: 1.10x slower
- fmt-fast-nested: 27.5 ms +- 1.6 ms -> 30.3 ms +- 2.9 ms: 1.10x slower
- fmt-fast-black/linegen: 397 ms +- 24 ms -> 431 ms +- 35 ms: 1.08x slower
- fmt-fast-black/mode: 41.9 ms +- 2.6 ms -> 45.3 ms +- 3.6 ms: 1.08x slower
- fmt-fast-dict-literal: 55.2 ms +- 2.8 ms -> 59.3 ms +- 4.8 ms: 1.07x slower
- fmt-fast-black/strings: 72.5 ms +- 3.7 ms -> 77.5 ms +- 6.2 ms: 1.07x slower
- fmt-fast-flit/install: 182 ms +- 12 ms -> 194 ms +- 16 ms: 1.07x slower
- fmt-fast-list-literal: 33.2 ms +- 1.6 ms -> 35.4 ms +- 3.0 ms: 1.07x slower
- fmt-fast-flit_core/config: 285 ms +- 16 ms -> 304 ms +- 25 ms: 1.07x slower
- fmt-fast-black/nodes: 329 ms +- 16 ms -> 347 ms +- 29 ms: 1.06x slower
- fmt-fast-black/brackets: 120 ms +- 7 ms -> 125 ms +- 11 ms: 1.05x slower

Benchmark hidden because not significant (2): fmt-fast-black/comments, fmt-fast-strings-list

Geometric mean: 1.08x slower

So gcc takes longer to compile while using twice as much memory to do so, produces binaries that are basically no smaller than clang's, is more picky about the C code it's given, and finally is slower ... what was I thinking gcc was viable :>

Anyway it seemed like the last set of optimizations have netted me a ~3-6% additional perf win which is nice! Especially since I got lazy and didn't do as much analysis for this optimization pass 😛

[plain atlas]

hello ichard

[bright glacier]

hey arl

[plain atlas]

is blackbench or diff-shades useful for a end user of black?

[bright glacier]

I mean you could use diff-shades to see the impact of a different revision of black but like it would be easier to just run black with the diff flag

And blackbench is purely focused on benchmarking black's performance, not quite sure how that would be useful to an end user

[plain atlas]

ah

lol

[bright glacier]

Maybe I'll publish it but maybe not (probably not)

[plain atlas]

apparently can't use it

[bright glacier]

It's in the black-mypyc-wheels project

[plain atlas]

ah

contemplating using mypy on my project

like

it caught a few incorrect typings the other day when I tried it for fun

[plain atlas]

hmmmmmm

how does black bump its version?

[torn night]

Hey, working on a project to automate some of my stuff at my job. Do i need to make a channel under open-source projects?

[bright glacier]

via setuptools-scm

[plain atlas]

ah

[silent apex]

Nope, that's specifically for open source projects partnered with this server (also, you shouldn't be able to make a channel, only admins can do that)

[flat kraken]

DetachHead

[psf/black] Issue opened: #2420 two newlines not added before function definition when there's a comment above it

Describe the bug

black does not add the two newlines between these two definitions when there's a comment in the way

a = 1
# adsf
def bar() -> None:
    ...

this causes the following error in flake8:

E302 expected 2 blank lines, found 0

Expected behavior

a = 1


# adsf
def bar() -> None:
    ...

Environment (please complete the following information):

• Version: 21.7b0
• OS and Python version: macos 11.4, python 3.10.0...

[flat kraken]

pfackeldey

[psf/black] Issue opened: #2421 Weird behaviour \(bug?\) of \`\# fmt: skip\`

Describe the bug

Dear black developers,

First of all, thank you for this wonderful package!
I wanted to skip one line using # fmt: skip. Unfortunately for my case this "skip"-statement seems to be "ignored" by black. I think the following example to reproduce makes it more clear.

To Reproduce

Consider the following code:

def somefunction(clusters):
    return bool(x)


if __name__ == "__main__":

    class A:
        foooooobaaar = "foobar"

     ...

[neon loom]

@bright glacier Hey mate - Long time, I've been busy with travelling and life for a change so haven't been around

Did you ever do any benchmarking with black using uvloop and not using uvloop. Especially blackd - I'd expect maybe less memory usage and maybe lower latency on a busier blackd server more so than a reduction - e.g. 5 files vs. 1 being formatted together (if it can even do that - Never used it directly - Pretty sure my vscode does tho)

[bright glacier]

I hope you had as much fun as COVID would let you!

[neon loom]

That's all @austere lava's doing tho (He does that for FB or has helped with it)

[neon loom]

I have - Rubicon off roading, when to chicago to see in laws, wedding in the Bay Area, been fun 🙂

*went

[bright glacier]

Ah no, I've been exclusively focused on benchmarking "how fast can black format predefined input". That sounds interesting tho!

[plain atlas]

uvloop is somewhat buggier, i've heard

!pypi uvloop

[late dew]

uvloop v0.16.0

Fast implementation of asyncio event loop on top of libuv

[bright glacier]

I've never really used it so I have no feel for that, bandersnatch uses it tho and we haven't had any major fires

[neon loom]

@plain atlas I've used it with great success on big busy python services @ Facebook (I work there)

And other open source projects

As ichard alluded to

[plain atlas]

ah

[neon loom]

I've just not seen if it's a win or not for Black

[plain atlas]

I mean

[neon loom]

I've seen memory 1/2 and latency drop on some services @ FB using it

[plain atlas]

pls be an optional dependency for black, if considered at all

[dense jungle]

I'd guess it doesn't matter as much since Black doesn't interact with the event loop all that much

[neon loom]

It is

[dense jungle]

But worth trying

[neon loom]

Already

[bright glacier]

I presume more nerdy drone footage was taken during this adventure 😉

[plain atlas]

ah

[neon loom]

Only off roading + Chicago ... Didn't take to wedding. Feel that's rude.

[plain atlas]

i mean, black takes like 0.64s for me so Ig having the uvloop dependency isn't worth it lolol

hm......

I think I just found a bug.

[bright glacier]

Well OK I thought it was clear drones would be a no-go at a wedding haha

[plain atlas]

black just formatted a file I have gitignored
From docs, and previous experience, I don't think its supposed to do this

[bright glacier]

it's probably mostly the import time being costly

[neon loom]

The photographer had his own to get the shots outside the ceremony

[plain atlas]

is this probably a bug?

[neon loom]

@bright glacier All the drone footage is shared on my Facebook of course 😛

[bright glacier]

Haven't heard of many weddings due to COVID, so it's nice to hear at least one happened 🙂

[neon loom]

I keep complaining to my friends in the video team about how the downsample my 4k video to crappy 1080p. 😦

*they

They live in the 80s

[bright glacier]

What does running black with verbose tell you, is at least one file being ignored via .gitignore or is .gitignore MIA?

[neon loom]

I have an internal task to my mate in the team "Give Cooper 4k"

[dense jungle]

We recently made a change to not listen to gitignore if you also have exclude set up in your pyproject.toml I believe

[bright glacier]

Haha, downsampled 4k is probably better than native 1080p so it could've been worse

is 1080p really the 80s these days? /hj

[dense jungle]

I got married last November with no guests. We plan to hold an actual wedding with our family and friends at some point but I don't know when if ever that will happen

[plain atlas]

my config

[tool.black]
line-length = 110
target-version = ['py38']
include = '\.pyi?$'

its a nested .gitignore file

[neon loom]

Congrats mate!

[neon loom]

Hurts my eyes

(Even tho 99% of what I watch is 1080p) 😛

[plain atlas]

since according to https://git-scm.com/docs/gitignore, you can have multiple .gitignore files and I use that to my benefit

[bright glacier]

I'm pretty someone added support for nested .gitignore support some time back so that shouldn't be a problem hmm

[bright glacier]

Oh nice, congrats! 🎉

[neon loom]

I held my wedding off till May 2022 to try and hope we're sorted by then. And to hope my parents can come from Australia 😦 Does not look good

[errant barn]

Big congrats!

[bright glacier]

The delta strain is really unfortunate, we're probably on our fourth wave at this point :/

[plain atlas]

hm

well i have a different git problem right now so I'll try and fix that and get back to you

also I just deleted those untracked files, FML

[bright glacier]

I did once a week ago and my goodness you don't know how happy I was I had a backup

[dense jungle]

Yeah international travel makes it even harder 😦 The US is still closed to European travelers so my family wouldn't even be able to come

[plain atlas]

i don't have those files backed up

[bright glacier]

fun.

[dense jungle]

And apparently Australia just doesn't let anyone in?

[plain atlas]

thankfully i copied it from a previous project and it took minor changes

and their point was to demonstrate how easy it was to convert other project stuff to this project lolol

so its easy to get back, but annoying

do any of you use visual studio code?

[dense jungle]

I do

[plain atlas]

my vscode isn't set as the default editor for cli actions

[bright glacier]

Even if travel is allowed, I fear delta will spread even among the vaccinated folks which would be not great

[neon loom]

Yup - both at the moment without exemptions

[plain atlas]

and I need to figure out how to set it as said default editor again

[neon loom]

I live in vscode

[bright glacier]

I switched to Vim a long time ago 😛

[plain atlas]

my setup is a device over ssh

the problem is its using vim instead of vscode for all terminal editing actions

specifically git

and github's cli too

$EDITOR is not set to anything

[bright glacier]

I remember setting up a git config in my vscode user settings that sets vscode as the git editor

[plain atlas]

fun thing is editor runs nano, and not vim

[neon loom]

I have vim plugin on vscode and still live in vim

I'm an old Linux BOFH

UNIX even - use to run solaris etc.

I was upset but understand when fedora changed editor default to nano 😦

[bright glacier]

    "terminal.integrated.env.linux": {
        "GIT_EDITOR": "code --wait",
        // this exists so the starship prompt isn't enabled which is broken in VSCODE
        "IN_VSCODE": "indeed!"
    },

that's my solution to using vscode with git relatively painfree

[plain atlas]

thanks

hm

[lament crow]

:O

[plain atlas]

how can i test the git editor is correct now?

[lament crow]

why?

or maybe I should backread fully first

[bright glacier]

I don't remember at all lemme find the issue

[plain atlas]

(basically whats a command that makes git open something to edit without going in to any mode)

eh

[silent apex]

You can probably just do an interactive rebase then abort

[lament crow]

Now that I did backread, I can be more confident it hasn't been answered 😄

[plain atlas]

just make a random edit, commit it, then reset head~1

[bright glacier]

here's the issue https://github.com/psf/black/issues/2164

[dense jungle]

Oh yes, the "gitignore everything" use case, never would have thought of that 🙂

[lament crow]

hmm

does this affect usage of extend/force exclude as well

[plain atlas]

that's an interesting use of gitignore lol

[bright glacier]

extend-exclude still works and force-exclude probably still works

the same*

[lament crow]

ah, then I don't really care

[plain atlas]

just a comment on it: I often look at a gitignore file to see what configuration files the project uses

[lament crow]

personally I'm interested in using existing gitignore but I also want to exclude vendored stubs from black

[bright glacier]

TL;DR:

Changes behavior of how .gitignore is handled. With this change, the rules in .gitignore are only used as a fallback if no exclusion rule is explicitly passed on the command line or in pyproject.toml. Previously they were used regardless if explicit exclusion rules were specified, preventing any overriding of .gitignore rules.

Those that depend only on .gitignore for their exclusion rules will not be affected. Those that use both .gitignore and exclude will find that exclude will act more like actually specifying exclude and not just another extra-excludes. If the previous behavior was desired, they should move their rules from exclude to extra-excludes.

[lament crow]

awesome!

[plain atlas]

IMHO to not have a breaking change, USE_GITIGNORE would've been nice lol

[bright glacier]

Any signs of when those restrictions will be lifted or weakened?

[plain atlas]

I don't want to mess up my local modmail branch lol

not until its reviewed 👀

[bright glacier]

oh, well that doesn't bode well

[austere lava]

So slow

[neon loom]

Australia says not until the new year. They still having 100s of cases a day in the bigger states (that's a lot for us - 25 million people country that's an island)

[austere lava]

Half of that is probably python startup time 😎

[neon loom]

Are those numbers published anywhere? Otherwise I'm gonna Bug report that Python in residence engineer ..

[bright glacier]

IIRC python startup time is measured on speed.python.org

or whatever the benchmarking website URL is

[neon loom]

that's the one

[dense jungle]

Guido's faster cpython project has been working specifically on startup time

[neon loom]

I can't remember too

[dense jungle]

He's been trying to speed up pyc loading

[austere lava]

Nice

[neon loom]

That will be H U G E

[austere lava]

I wish Instagram would/could upstream its runtime improvements

[bright glacier]

And I'm working on better black performance heh (I know, 100% less cool)

[bright glacier]

That sounds awesome, import time is such a big issue

[neon loom]

Me too. And make it available generally internally 😛

[austere lava]

Oh that we're working on

Stay tuned

[bright glacier]

👀

[neon loom]

Make my CLIs great again

[austere lava]

Eggsactly

[neon loom]

@solid adder You got any time for black anymore? Any love for us? 😛

[bright glacier]

I've found the DIR weekly reports fascinating. Oh and I'm particularly excited about the issue and PR stats that are apparently coming next week 😄

[austere lava]

Where can I read them?

I've just seen my 2.5 year old PRs get merged

[neon loom]

Yeah they have been good

[bright glacier]

https://lukasz.langa.pl/7cf5f517-87fb-473c-ae06-72afe095ac72/

[neon loom]

I think out of my few python contributions ambv has merged all but 1. lol

[bright glacier]

I saw some changes get merged into lib2to3 a few days ago lol

I wouldn't be surprised if zsol already implemented the changes (or the functionality of the changes) in blib2to3 a long time ago :p

[dense jungle]

Yes, Lukasz's work has been great, I've been seeing a ton of old PRs getting merged

[austere lava]

Yeah they were just backports of fixes I made in blib2to3

[bright glacier]

Nice! That reminds me that we still need to address the "new parser" issue don't we lol sigh

[austere lava]

This is cool, thanks!

[neon loom]

Shame we have to move out of blib2to3 one day ...

*off of

[austere lava]

Yeah so about that

I've a few people helping me with the rewrite by now, but am gonna go on vacation end of next week

So not sure how much progress there's going to be

[dense jungle]

Your rewrite is for LibCST's parser, right? Or are you also working directly on Black?

[bright glacier]

CPython really gets a lot of PR traffic doesn't it, I would've expected a more noticeable dip from Łukasz's help but nope!

[austere lava]

Nono just libcst

[dense jungle]

So we'd still have to totally rewrite black to work on top of LibCST 🙂

[bright glacier]

how's using rust with python? or have you not reached that stage yet?

[austere lava]

I probably won't have the bandwidth to do that yeah

[bright glacier]

I ask that because someone just said they made a blackd client in rust lol

[austere lava]

But more than happy to dedicate time to support someone who wants to do it 😎

[austere lava]

Oh really?

[neon loom]

We still have some money and if libcst seems best we can tender out spending that money on getting it done or ask for donations etc.

[neon loom]

How will that help?

[bright glacier]

https://github.com/psf/black/issues/2405#issuecomment-897073455

[austere lava]

https://github.com/zsol/blacc

[neon loom]

A client in rust would work ...

[dense jungle]

Less startup cost

[neon loom]

I missed client in the original message from richard and thought someone said they'd make blackd in rust ...

Sure - as long as they add CI and documentation ...

building, unittests + integration testing with blackd

[austere lava]

To anyone reading this: I can't in good conscience recommend this cli as it was my "let's learn rust" project 😁

[frozen badge]

too bad, it seems really neat for whenever im gonna try to turn on format on save

[bright glacier]

I'm a bit worried with how we could integrate it with our build system but apparently PyO3 has a setuptools plugin as I've heard?

The argument we shouldn't distribute native binaries is already out of the window due to my mypyc work tho

unless we decide not to, but I hope not 😅

[flat kraken]

Coronon

[psf/black] Issue opened: #2422 Better Comments

Is your feature request related to a problem? Please describe.

I am using the VisualStudio Code extension Better Comments, which works by coloring comments based on a specific symbol/string right after the comment start. The frustrating part is that Black always adds a space in between, which would turn this:

a = 10  #? Should we change this

into this:

a = 10  # ? Should we change th...

[plain atlas]

that was close

imagine running black / and not black .

temptation got to the best of me

turns out what happens is it ||runs in to a .gitignore file it doesn't have permission to read and fatally errors before it does anything||

[bright glacier]

hmm maybe we need go of the way of the warning rm -rf / does

[plain atlas]

yeah, might be worth it

at the risk of looking like i reformatted all of the pythons on my machine at one point, I'll make an issue on the repo

psf/black#1

[toxic storm]

GitHub

Imports following standalone comments sometimes produce non-deterministic styling

[plain atlas]

^ ignore, just a short link to get my browser open

[flat kraken]

onerandomusername

[psf/black] Pull request opened: #2423 fix: remove unneccessary escape character

When making an issue, there's an extra \ that shouldn't be there, and it doesn't need to be escaped in the .yml file.

[plain atlas]

got sidetracked, sorry

[flat kraken]

onerandomusername

[psf/black] Issue opened: #2424 Have a warning when the user tries to reformat root\.

Is your feature request related to a problem? Please describe.
The . and / characters are very close together, and this can lead to accidentally reformatting from the wrong directory.

Describe the solution you'd like
@ichard26 suggested to have a warning like with rm -rf /, and I agree with this suggestion.

Describe alternatives you've considered

Additional context

ichard26

[psf/black] Pull request review submitted: #2423 fix: remove unneccessary escape character

LGTM, thank you!

[bright glacier]

Made from a phone 😄

[plain atlas]

nice

i'm curious, do you wait for checks to complete and then merge, or do you need someone else to approve it too?

LOL

[bright glacier]

waiting for checks even although it's only the lint one that matters here

[plain atlas]

you don't have automerge enabled?

[bright glacier]

larger changes usually have two reviews but one review is enough for those sort of changes

[bright glacier]

nope.

[plain atlas]

that allows you to turn on automerge and then once all checks pass the branch will be merged :)

[flat kraken]

JelleZijlstra

[psf/black] Pull request review submitted: #2423 fix: remove unneccessary escape character

[plain atlas]

hi

[bright glacier]

salut

[plain atlas]

mrw having problems with windows so I close explorer.exe

:)

[warped hearth]

Why didn't black pick up the CamelCase as an error in https://github.com/Gweno/tutolibro.tech/blob/master/lopy/part9/ReverseRange.py ?

[lament crow]

why would CamelCase be an error?

[plain atlas]

holy shit lol

okay wow

I have to share this project here, its very impressive

https://mdformat.readthedocs.io/

formats all markdown files

now, the reason to share it here, is that it has a plugin system, and there is a plugin for formatting python code within the markdown file

!pypi mdformat-black

[late dew]

mdformat-black v0.1.1

Mdformat plugin to Blacken Python code blocks

[plain atlas]

Now I can format all of the code blocks in my markdown files with black as well!

[lament crow]

have you heard of blacken-docs for reStructuredText?

[plain atlas]

no....

lol

[little parrot]

It works with md jackenmen

I use it

[plain atlas]

!pypi blacken-docs

[late dew]

blacken-docs v1.10.0

Run `black` on python code blocks in documentation files

[lament crow]

I don't, thought it's just rst

but ye, looks like it's rst, md, and tex

[plain atlas]

hmmm

well then

still would use mdformat for formatting md

!pypi mdformat

[late dew]

mdformat v0.7.8

CommonMark compliant Markdown formatter

[lament crow]

yeah, makes sense for auto formatting of whole markdown files

[plain atlas]

yeah

[lament crow]

I know that when it comes to reStructuredText, there's rstfmt but it's in a very, very early stage

[plain atlas]

not sure that using blacken-docs is worth it if I'm already using mdformat

except for one thing

mdformat-black requires the code block to be labeled with python

pydoesn't count

[lament crow]

seems like a bug

[plain atlas]

yeah

[little parrot]

You can just use both

[warped hearth]

@lament crow https://stackoverflow.com/a/160830/4539999

I'd expect it to be identified

[plain atlas]

i mean yeah but also like, if i use blacken-docs I'll cut the mdformat-black from pre-commit

[dense jungle]

Black doesn't enforce this. It's too unsafe. We'd have to have a way to rename the variable everywhere it's referenced, and we don't

[plain atlas]

also uh, dynamic getattr

[warped hearth]

https://www.geeksforgeeks.org/python-program-to-convert-camel-case-string-to-snake-case/

[dense jungle]

yes, there's lots of ways it's unsafe

[plain atlas]

yeah

and it heavily changes the ast of the file

[warped hearth]

imo snake_case is rubbish but that's the way python specs were written. Seems it's often just ignored.

[plain atlas]

i mean, how pep8 was written but yeah

[warped hearth]

has it formally changed since pep8?

[dense jungle]

PEP 8 is just the style guide for the standard library. You can do whatever you want in your own code

If it was feasible, Black would enforce snake_case, but it's not feasible so we don't

[plain atlas]

and even more so, the logging class exists

[dense jungle]

yes, even in the stdlib there is lots of legacy code that doesn't follow the conventions

[warped hearth]

Which is crap when you start mixing code from different sources, the whole point of a style guide.

[plain atlas]

logging.basicConfig smh

[warped hearth]

legacy code is a different issue

[little parrot]

All the names for logging I thought were copied from Java

[plain atlas]

yeah

[warped hearth]

In which case you wouldn't expect them to change

[little parrot]

I feel like they could be changed though they did it with the threading module

[plain atlas]

TBH

blacken-docs looks a lot better than mdformat-black so yeah

ah yes

gotta update it all

[late dew]

@bright glacier

It has arrived!

Here's your reminder: maybe, just maybe ask about configurave
[Jump back to when you created the reminder](#black-formatter message)

[plain atlas]

not sure I summarized it yet but tl;dr take inputs from env, toml, default toml, and put it into models so it can be used.

[flat kraken]

ichard26

[psf/black] Pull request closed: #2423 fix: remove unneccessary escape character

ichard26

[black:main] 1 new commit

355a6b3 fix: remove unneccessary escape character (#2423) - onerandomusername

[plain atlas]

i

did i

noooooooo

I misspelled unnecessary.

wow

@bright glacier I regret my life.

[flat kraken]

ichard26

[psf/black] Pull request review submitted: #2414 Present a more user\-friendly error if \.gitignore is invalid

This seems fine to me, but I will note that code wasn't immediately clear and I was quite scared about "if gitignore is None: sys.exit(1)" bit since I thought it would also exit on cases where a gitignore file didn't exist.

Nevertheless, thank you!

[bright glacier]

it's ok, I'm sure I have my fair share of regrettable commit messages that are now out in public ... forever 😅

[plain atlas]

force pushes tho

[bright glacier]

not an option on a repo like psf/black

[plain atlas]

i mean, you can on pull requests

but yeah, once something gets to main its too late.

[bright glacier]

$ black -c "print("
print(
error: cannot format <string>: ('EOF in multi-line statement', (2, 0))

Haha, I found a unrelated regression while reviewing a PR. Looks simple enough to fix.

The input should definitely be not printed if an error occurs :)

[plain atlas]

should it tho?

I guess with an EOF its worth not printing

but other errors some of it should?

[bright glacier]

Why should it give back the user what it gave quite explicitly?

It's unnecessary and noisy if there's an error blocking black from doing its thing.

[flat kraken]

ichard26

[psf/black] Issue closed: #2420 two newlines not added before function definition when there's a comment above it

[bright glacier]

I haven't closed an issue in a long time haha, feels good to know I'm still reasonably decent at finding duplicates 😄

[flat kraken]

ichard26

[psf/black] Issue closed: #2403 AST error

[flat kraken]

TomFryers

[psf/black] Pull request opened: #2425 Update language server links

Development moved to https://github.com/python-lsp/python-lsp-server.

[flat kraken]

cogat

[psf/black] Issue opened: #2426 INTERNAL ERROR: Black produced different code on the second pass of the formatter

Describe the bug

Black failed to format a valid python file. Appears to be something to do with inline comments on a multi-line statement.

To Reproduce

Diff is below:

--- source
+++ first pass
@@ -118,14 +118,14 @@
 
 
 def _get_rates(product: market_supply_models.Product) -> List[GasRateRow]:
     # Gas rates have stepped consumption and one standing charges for each zone identifier.
     rates_rows: Dict[Tuple, GasRateRow] = dict()
-    for rate in (
-     ...

[stuck vapor]

What's the best way of getting the latest released version of black from within the codebase? (this is for the issue of keeping the pre-commit example's rev in sync with the latest version of black)

One way would be to parse it from the CHANGES.md file, but that feels brittle.
I considered using black.__version__ but that gives e.g. 21.7b1.dev9+gb1d0601.d20210810, rather than the latest released version.
Finally, I considered looking at the latest git tag, but figured that that might not fit in with the release process - if a tag triggers a release, then the check to ensure that the rev in the pre-commit example from the docs is synced with the latest release of black would be triggered too late.

[frozen badge]

have you looked at how the black --version command gets the version? alternatively, could you use the black release rss feed frm pypi?

[stuck vapor]

for the former, it's setuptools-scm, but it only gives the "development" version (e.g. 21.7b1.dev9+gb1d0601.d20210810 as opposed to 217.b0)

as for taking the rss feed - this wouldn't work as a CI check

Thanks anyway for your suggestions!

[bright glacier]

are emojis still not acceptable for a command line tool in certain environments?

[bright glacier]

one idea would be to use that development version, parse it with packaging.versions.Version and then get the main version out from that. Not sure how well that would work tho!

[austere lava]

I would get the latest tag from git. That's what setuptools-scm does too

[stuck vapor]

Thanks - tried that but it doesn't seem to work in CI. Will take another look anyway, I might have done something wrong

[austere lava]

I mean you can always use the github api to get the latest release

git checkout $(curl -sSL api.github.com/repos/psf/black/releases/latest | jq -r .tag_name)

Or something like that

[stuck vapor]

Thanks @austere lava - I tried

        latest_tag = subprocess.run(
            shlex.split(
                "curl -sSL api.github.com/repos/psf/black/releases/latest "
                '| grep \'"tag_name":\' | sed -E \'s/.*"([^"]+)".*/\1/\''
            ),
            universal_newlines=True,
            stdout=subprocess.PIPE,
        ).stdout.rstrip()

but then in CI get

fatal: No names found, cannot describe anything.
curl: (58) could not load PEM client certificate, OpenSSL error error:02001002:system library:fopen:No such file or directory, (no key found, wrong pass phrase, or wrong file format?)

Not sure what this is but I'll investigate

[heavy heron]

There are supported enough to be used I would say, since the new Windows Terminal now adds emoji support natively.

[bright glacier]

I meant this 🙃

[dense jungle]

so you meant acceptable socially, not acceptable technically?

[bright glacier]

yep!

like I get they can look out of place, but it feels a bit silly to add a toggle to disable them

I'm surprised at how much traffic we've gotten over this tho

[stuck vapor]

not managing to get this to work in CI, for now I'll go back to getting the version from the CHANGES.md file

[flat kraken]

AndrewHoos

[psf/black] Issue opened: #2427 INTERNAL ERROR: Black produced different code on the second pass of the formatter

INTERNAL ERROR: Black produced different code on the second pass of the formatter

To Reproduce

Black this file

def black(foo):


    assert pythoncontent.content_from_environment(
        {
            "File": tmpdir / "FilePath.ptest",
            "Environment": {
                "Virtualenv": {
                    "RequiredPackages": {
                        'package': "1.0"
                    },
                    "RequirementsFile": "r.txt"
         ...

[austere lava]

letsgooo https://github.com/pytorch/pytorch/commit/b0043072529b81276a69df29e00555333117646c

If I'm fired tomorrow, then we've learned it's better to ask for permission and not forgiveness

[delicate sky]

you can't be fired, its always better to ask for forgiveness over permission, its the pythonic way

[austere lava]

Well that didn't last long https://github.com/pytorch/pytorch/commit/1022443168b5fad55bbd03d087abf574c9d2e9df

[bright glacier]

not surprised :p

[heavy heron]

I think it is fair to be able to disable them. Some systems just don't like emojis at all and produce garbage data when they are used.

Arguably the system should be updated rather than the app, but who am I to judge 🙃

[drowsy sonnet]

Hello hello. I'm here to be disruptive again: anyone tried Furo with black's documentation yet? 🙃

[dense jungle]

Richard tried it! https://black.readthedocs.io/en/hello-furo/

https://github.com/psf/black/issues/2397

[drowsy sonnet]

Aye. Nice!

Lemme know if y'all see anything disruptive/concerning with Furo.

Most of my Discord channels are muted, but @-mentioning me in this channel (or on the PyPA discord) will reach me.

Anything I can do to help get https://github.com/psf/black/pull/2237 moving again?

[flat kraken]

AndrewHoos

[psf/black] Issue closed: #2427 INTERNAL ERROR: Black produced different code on the second pass of the formatter

[flat kraken]

ichard26

[psf/black] Pull request review submitted: #2425 Update language server links

Thank you very much!

ichard26

[psf/black] Pull request closed: #2425 Update language server links

ichard26

[black:main] 1 new commit

e465acf Update language server links (#2425) - TomFryers

cogat

[psf/black] Issue closed: #2426 INTERNAL ERROR: Black produced different code on the second pass of the formatter

[lament crow]

I've seen furo has recently been added as an example theme to sphinx's tutorial so that's nice

[lament crow]

https://twitter.com/pradyunsg/status/1425942946734555138?s=19

Pradyun Gedam (@pradyunsg)

I think the best-looking uses of Furo (that I've seen so far) has been Black's documentation built with it, from one of the maintainers trying it out.

https://t.co/FZFr1EKens

Twitter · Aug 12, 2021, 10:11 PM

In case someone does not follow Pradyun or just doesn't use Twitter

[delicate sky]

glad we got the update, now we just need to know if you still got your job 😆

[austere lava]

so far so good 😛

[flat kraken]

rethab

[psf/black] Pull request opened: #2428 docs: Remove quotes from file path

[bright glacier]

@dense jungle is there an issue on the mypy issue tracker about renaming the default branch to main or something else? I'm asking rather selfishly because mypy is the only repository I've regularly been developing for that stills uses master but my muscle memory has already learned main so mypy using master is annoying :p

I tried looking but tbh the 1700+ issues were impossible to dig through :/

[dense jungle]

No there isn't. Last I heard the Steering Council would make a call on when to do it for all python repos but that hasn't happened yet

https://github.com/python/peps/issues/1956

[flat kraken]

rethab

[psf/black] Pull request closed: #2428 docs: Remove quotes from file path

[mint barn]

@bot.command()
async def avatar(ctx, *, avamember : discord.Member=None):
userAvatarUrl = avamember.avatar_url
await ctx.send(userAvatarUrl)

[tribal thistle]

Hey! Welcome to the server. We cover most info you need to know in #welcome.

This channel is meant to discuss the package ‘black’.

For general discord.py help, try #❓｜how-to-get-help or #discord-bots

[mint barn]

was

german pls

[mint barn]

why have you the role "muted"

[tribal thistle]

This whole server uses English, please see #rules. If you can’t speak English, it’s fine to use something like Google Translate

[quartz flicker]

hey how's it going, I'm trying to use black on python 3.9.6, for some reason it won't work without the --fast param.
receiving errors when integrating with vim/nvim plugins (Neoformat/Formatter.nvim) as well.

[bright glacier]

What's the error you're getting without the flag @quartz flicker ?

It shouldn't be crashing due to the integration, I suspect black is buggy on a particular input or your input isn't actual valid python code.

In other news, I just finished the last set of changes I wanted before releasing blackbench 21.8a1. This release is special since it's the first one that's going to be published to PyPI. It also will be my first project on PyPI. I don't know if I should be worried or scared, but I am :p

CI is green so I guess it's time to try out my release automation and hope nothing breaks. Afterall, what's the worst that could happen?

Oh wait, step one is first remembering my PyPI password lol, it's been too long since I've last signed in 😅

nox > git commit -m Let's get back to development
nox > Alright, just do a push to GitHub and everything should be done
nox > If you're paranoid, go ahead, verify that things are fine
nox > If not, sit back and relax, you just did a release 🎉
nox > Session do-release was successful.

ahhhhhhhhhhhhhhhhhhhhhhh my blood pressure is still very high right now but I'm excited!

[tribal thistle]

This is what test.pypi is for haha

Grats tho!

[bright glacier]

everything actually does seem to be fine 😄

!pypi blackbench

[late dew]

blackbench v21.8a1

A benchmarking suite for Black, the Python code formatter.

[bright glacier]

I'm too lazy to setup an account for test.pypi.org

Hmm, does readthedocs not create a stable version automatically for you if there's only prereleases

I guess I'll just make a stable branch for the time being then.

Alright, time to benchmark a lot so I can open a PR for my mypyc work soon, but first a break 😄

[plain atlas]

yay

[plain atlas]

how can i delete a project from there so I can test again?

[bright glacier]

well the versions won't be freed, once used they are done forever

[plain atlas]

yeah ;-;

even though it promises its a test pypi

its not actually a "test"

its a second index lol

[bright glacier]

that's meant for testing, so in a way, it serves its purpose, not well tho IMO

[plain atlas]

yeah

i agree its not well

also poetry is buggy rn

and I actually don't have a keychain on my device so I have to delete poetry's config every time since it stores creds in a plaintext file lol

may see about using twine for publishing instead of poetry NGL

[tribal thistle]

If you’re really concerned about messing up the test index, spin up your own copy

[bright glacier]

Eh, I'm just happy I've done a release on PyPI. I'll deal with security later 😉

(disclaimer: I'm not liable for any damages that may occur if you follow my "advice", this is not intended as advice)

[tribal thistle]

Untrue, and OT 🤡

[lament crow]

password managers can help with that :)

[exotic flume]

like keepass...

[stuck vapor]

Sorry to bump this again 😳 - is anything blocking the release?

[neon loom]

My time and people telling me of we’re ready or not. GitHub installs always work too …

I have some time today - where we at?

[dense jungle]

we should be good for a release. I won't have much time to help out though, we're moving tomorrow and I live in a labyrinth of boxes

[neon loom]

haha - That's fine. I'll take a look later. LinkedIn just suckered me into doing a quiz to get a badge - lol

[short forge]

Congratulations! 👏

[neon loom]

lol - thanks. Was like, I hope I can pass this.

[gusty grail]

yeah... I think you passed 😄

[bright glacier]

Oh good luck finding your way out of the labyrinth then! I was going to pester you today since I'm testing and benchmarking the mypyc branch in preparation for a PR ... but I guess Cooper can help out ^^

[silent void]

what is mypyc in the context you used the word

[bright glacier]

a Python-to-C transcompiler

[silent void]

that's the cool thing I've heard

[little parrot]

mypyc is a python to c transpiler that uses mypy to make sure what youre writing is "safe"

[bright glacier]

It also uses the type information to optimize in ways that would be unsafe if the type is unknown at runtime.

Unfortunately mypyc is still alpha software so extensive testing is necessary before deploying it to PyPI, right now I'm doing one of those testing runs

Ah crap, looks like blackbench is buggy with mypyc compiled Black

cool, so mypy didn't catch this bug because one of my type annotations were wrong (should've been Optional[str] not str) and I doubt mypy could ever hope to catch this type annotation issue due to how the function is called.

[little parrot]

do you use --strict?

[neon loom]

I think we should do a release with what we have and do a dedicated no changes for mypyc so we can see if we get any "mypyc" related new bugs etc.

[bright glacier]

The problem is that the type annotation is defined in a function that's called by click's magical callback code

[bright glacier]

I doubt that mypyc will cause any issues at the pure formatting level, and I'm somewhat confident on this because the data just came in from diff-shades which shows no difference in interpreted black vs mypyc compiled to C black.

Although I expect other classes of bugs and edge cases

[neon loom]

Sweet

[bright glacier]

❯ python ../black-mypyc-wheels/diff_shades.py compare main-e465acf.json linux-mypyc-e9834e0.json 
Loading files ... done

aioexabgp (19 files): everything is the same
attrs (46 files): everything is the same
bandersnatch (50 files): everything is the same
blackbench (13 files): everything is the same
channel (44 files): everything is the same
django (2692 files): everything is the same
flake8-bugbear (27 files): everything is the same
hypothesis (384 files): everything is the same
pandas (1390 files): everything is the same
pillow (275 files): everything is the same
poetry (310 files): everything is the same
pyanalyze (48 files): everything is the same
pyramid (178 files): everything is the same
ptr (5 files): everything is the same
pytest (248 files): everything is the same
scikit-lego (124 files): everything is the same
sqlalchemy (590 files): everything is the same
tox (94 files): everything is the same
typeshed (2306 files): everything is the same
virtualenv (145 files): everything is the same
warehouse (504 files): everything is the same

As usual, we're done here, have a wonderful day/night! Cya! <3

[neon loom]

Yeah, I have - experience with all this but it sounds cool

[bright glacier]

I may have made a better version of black-primer 😉

But your work was important still, black-primer spurred the creation of mypy-primer whose ideas I stole for diff-shades haha 😄

Actually @neon loom as I've alluded to before, I'm going to need some help since I don't have an MacOS running machine. Are you up for it?

[neon loom]

I can give you SSH to my laptop if that makes it easier ...

[bright glacier]

Also the M1 compiled wheels are totally untested since we don't have access to M1 CI resources

[neon loom]

Do you have IPv6 enabled on your connection?

[bright glacier]

I'm sorry if this disappoints you but I've never used ssh 😅

[neon loom]

😮

[bright glacier]

Probably not

[neon loom]

Ok. No problems. It's just encrypted remote shell/terminal access basically. Would let you copy files / use my m1 mac to do what you need (I'm using it as we speak too) ... Was planning on making you a non admin account + opening SSH directly to it for your IPv6 address if you had one. I could do IPv4 but it's more effort as I would have to port forward from my main router etc. and I have IPv4 off but default ... # legacy

For all those people who are like you need v4 (unfortunately I do for lame GItHub) I use NAT64 ... https://cooperlees.com/2020/12/nat64-using-jool-on-ubuntu-20-04/

[bright glacier]

I've lived my life on IPv4. Not much of a networking nerd fwiw

[neon loom]

I'm actually ok @ networking, python is a side hobby that I am trying to be good at.

https://www.youtube.com/watch?v=WaCxsVUmHa8 - Done lots of silly stuffs

[bright glacier]

Ugh that's fine. Sounds like a pain to turn on IPv4. I'll do what I can on Linux and Windows and think about what I should do next

[neon loom]

No biggy. You just need me to build + run tests right?

GitHub actions can at least build x86_64 mypyc and test now? lets make an action to test mypyc?

[bright glacier]

I've been running the test suite for all of the environments supported, it's just the M1 wheels that can't be tested

https://github.com/ichard26/black-mypyc-wheels/blob/main/.github/workflows/build.yaml

https://github.com/ichard26/black-mypyc-wheels/blob/a6e773925ad01998be94fc2fcc5e6059c6682ca5/.github/workflows/build.yaml#L83-L87

[late dew]

.github/workflows/build.yaml lines 83 to 87

CIBW_BEFORE_TEST: "pip install -r .mypyc-support/test-requirements.txt"
CIBW_TEST_COMMAND: 'pytest {project} -k "not incompatible_with_mypyc"'
CIBW_TEST_EXTRAS: "python2,d,jupyter"
# Skip trying to test arm64 builds on Intel Macs.
CIBW_TEST_SKIP: "*-macosx_arm64 *-macosx_universal2:arm64"```

[bright glacier]

Linux, Windows and Intel MacOS all run the test suite (excluding some tests that break because mocks and monkeypatching aren't supported by mypyc) during the wheel build

[tribal thistle]

I wonder if I can spin up a couple machines on my host

Worst case scenario, I can write a simple circle CI or smthn

Have it deployed on all machines

[bright glacier]

Yeah it should be that hopefully. We could run benchmarks on Mac too but it's not strictly necessary and can be quite fickle to do properly. Basically setup test environment, install wheel from the workflow. run pytest -k "not incompatible_with_mypyc", and run diff-shades and compare the produced data file to a known good one from main (I'll provide it). Oh do all of that again for M1.

I really need to come up with better instructions

[neon loom]

Can we shell script that so its all repeatable more reliably 😄

[bright glacier]

Good idea, especially since I want to invite members of the community to help out too

[neon loom]

(Y)

Whoops - 👍

(That is a Facebook shortcut)

[left lagoon]

yes

[neon loom]

TIL - 148 files would be left unchanged in the cpython repo 😄

[plain atlas]

nice

[neon loom]

Didn't expect that many

[hybrid island]

probably some really small files, but still impressive

[neon loom]

Just running CI then wull have the PR up for #2407

*will

[bright glacier]

psf/black#2407

[toxic storm]

GitHub

Create/Find a repo that uses every python syntax + Add to Primer

[bright glacier]

So wait, we're adding CPython to primer?

[neon loom]

Ya ya

[bright glacier]

I'm a little bit worried at how long CI is doing to take with that

[neon loom]

4 mins. And only on 3.9 (latest stable cpython so we can support as much syntax as possible) - I think the signal is very valuable. Already found more ESP bugs.

non ESP == 13 failed files - 16 with ESP

[bright glacier]

Ah ok that's good. Why do we run primer across multiple OSes and Python versions btw?

[neon loom]

Well - the 4 mins is on my Mac M1 maxed out. I may well be to slow for Actions. if so, will do it in a dedicated job in 3.9 only

[bright glacier]

To find these bugs?

[neon loom]

We've seen platform specific bugs, so running the same code across different OS's find these things.

[bright glacier]

Fair enough

[flat kraken]

cooperlees

[psf/black] Pull request opened: #2429 Add cpython repository into primer runs

• cpython tests is probably the best repo for black to test on as the stdlib's unittests should use all syntax
  • Limit to running in recent versions of the python runtime - e.g. today >= 3.9
    • This allows us to parse more syntax
• Exclude all failing files for now
  • Definately have bugs to explore there - Refer to #2407 for more details there
  • Some test files on purpose have syntax errors, so we will never be able to parse them
• Add new black command arguments logging in deb...

[neon loom]

Such a small amount of code for the hours of work. I hate we use a regex for excludes personally.

But a cool POC I feel, showing we can format a very high % of cpython. Some files we'll always have to exclude, but ~10 or so there we need to dig into

@bright glacier - Worth a change log? I was torn

I don't feel users care what repos we CI on

3.9 passed but 3.10 timed out:

[2021-08-15 23:08:36,091] ERROR: Running black for /tmp/primer.20210815225821/cpython timed out (['/opt/hostedtoolcache/Python/3.10.0-rc.1/x64/bin/black', '--experimental-string-processing', '--extend-exclude', 'Lib/lib2to3/tests/data/different_encoding.py|Lib/lib2to3/tests/data/false_encoding.py|Lib/lib2to3/tests/data/py2_test_grammar.py|Lib/test/bad_coding.py|Lib/test/bad_coding2.py|Lib/test/badsyntax_3131.py|Lib/test/badsyntax_pep3120.py|Lib/test/test_base64.py|Lib/test/test_exceptions.py|Lib/test/test_grammar.py|Lib/test/test_named_expressions.py|Lib/test/test_patma.py|Lib/test/test_tokenize.py|Lib/test/test_xml_etree.py|Lib/traceback.py|Tools/c-analyzer/c_parser/parser/_delim.py', '--check', '--diff', '.']) (lib.py:190)

[bright glacier]

IMO no

[neon loom]

Seems I read wrong. They both timed out. Will allow 15 mins, if it can't pass reliably in 15 mins I'll disable for now.

[bright glacier]

Whelp I'm never running black-primer locally (or diff-shades once I add CPython) on a whim

it takes around ~30 minutes for me to a run through all of the primer projects

[neon loom]

You can always change the config ... I do when I play with stuff locally

[bright glacier]

Yup yep, for diff-shades I added simple include / exclude options for that

[neon loom]

I don't know what a diff-shared is and why we need so many tools

lol

*shades

[bright glacier]

diff-shades is basically me writing my own black-primer by stealing a bunch of ideas from mypy-primer

it's designed to compare two revisions of black

[neon loom]

Cool

[bright glacier]

I was going to use black-primer instead for my mypyc work but it wasn't ideal because primer can't easily tell you whether two revisions of black behave the same

it does a good job at testing black's stability and impact on projects as a whole BUT it just doesn't tell you a lot at a revision level

[neon loom]

Nope, didn't have that goal in mind when I wrote it. But I'd never be upset if something replaced it and did it all if it's of value etc.

[bright glacier]

Basically this is the sort of tooling I needed and wanted.

Pretty useful to see if a change is too disruptive or changes something it shouldn't ^^

Anyway, diff-shades was thrown together in a few days so the quality of the tool ain't great but I've been using it successfully for my mypyc work which makes me happy

I gotta go for now, but I'll leave with this summary: https://github.com/ichard26/black-mypyc-wheels/blob/a6e773925ad01998be94fc2fcc5e6059c6682ca5/diff_shades.py#L413-L437

[neon loom]

Yeah nice. That's how all good tools get started. primer can be replaced / re-written to do all this if needed.

Give up, going to be no sane way to run cpython. I limited it to Lib/* but still ~3m40s on my laptop. Not gonna happen. But I'd still like to merge the config for others to be able to use etc. if we ever want to debug the bugs it exposes.

Tis a shame. But ya never know unless you try.

Maybe we get rid of django and run cpython as the one large repo - it may work. But I think having the two large repos overloaded the i/o as I think it's running 2 at once (by default)

[bright glacier]

which IMO is not the best idea because black is parallelized anyway ^^

[flat kraken]

ichard26

[psf/black] Pull request review submitted: #2419 Pipfile lock

Yup! Thank you!

P.S. the reason why I wanted the jupyter deps in here is to reduce friction when contributing. You shouldn't have to reinstall black with the jupyter extra just to test your ipynb changes ^^

ichard26

[psf/black] Pull request closed: #2419 Add jupyter deps to Pipfile\.lock

ichard26

[black:main] 1 new commit

b92ec34 Add jupyter deps to Pipfile.lock (#2419) - MarcoGorelli

[bright glacier]

Alright, let's try again at adding the test dependencies to Pipfile.lock, hopefully pipenv behaves this time ^^

Aw crap, the weird pytz issue is still happening, gosh this is annoying

Hmm, looks like this is a known bug: https://github.com/pypa/pipenv/issues/4686

I wonder if my weird environment setup is the cause

I'm now on attempt 4 I think?

[plain atlas]

at some point

I had commits

where my message was just 5

6, 7, 8

etc

because freaking github workflows don't have custom runners to test them

[bright glacier]

a commit history can tell a story ^^

[plain atlas]

...of not being able to count

[bright glacier]

83564ec5c (HEAD -> fix-messages, origin/fix-messages) This is ugly but special-casing this test seems fine
91ac7200b Only sys.exit(1) if an error occurred + fix one warning
0367ebf89 [mypyc] Emit messages regardless if there's no errors

[plain atlas]

attempt 1
attempt 2
attempt 3
attempt 5

[bright glacier]

"attempt idk but i swear this *should* work"

that's how you make a good commit history 😉 /j

[plain atlas]

squash em afterwards

[bright glacier]

HECK YEA

pipenv didn't break the pytz dependency

damn is it sensitive to the environment tho

I had to move a virtual environment out of my CWD before running pipenv lock 😕

ugh, wth now the hashes aren't matching on 3.6

i'mma try again tomorrow, but at least i made progress :D

[flat kraken]

aneeshusa

[psf/black] Pull request opened: #2430 Remove \`language\_version\` for pre\-commit

At my company, we set the Python version in default_language_version
in each repo's .pre-commit-config.yaml,
so that all hooks are running with the same Python version.

However, this currently doesn't work for black,
as the language_version specified here
in the upstream .pre-commit-hooks.yaml takes precedence.
Currently, this requires us to manually set language_version
specifically for black,
duplicating the value from default_language_version.
The failure mode otherw...

[bright glacier]

bloody hell now the hashes work but the lock file isn't complete

Using the lock file on 3.6 for example breaks the documentation

Gosh I dislike pipenv more and more

[little parrot]

Oh I'm actually quite interested about this, black was one of the first projects I ever saw using poetry, why did you guys switch?

[plain atlas]

@bright glacier why are you using pipenv?

[bright glacier]

I have no idea 🙃

The switch happened before I even contributed to black

[tribal thistle]

Join the darkside - switch to poetry

I wonder hmm

Heh

https://test.pypi.org/project/black-poetry/

Y'all probably aren't interested in using a full poetry build and release system, but that's why I have two branches

It's a bit outdated now, but this branch has a solid start for setting up poetry

https://github.com/HassanAbouelela/black/tree/poetry

The one that released the pypi package is poetry-build

[plain atlas]

i'm actually likely going to switch to twine for releases 🙃

[little parrot]

i must be entirely mis remembering cause i cant find poetry in the history anywhere

[flat kraken]

ichard26

[psf/black] Pull request opened: #2431 Implementing mypyc support pt\. 2

It's finally happening :tada:, the initial work was done several years ago in GH-1009 by @ msullivan is now being followed through by yours truly. If this work doesn't find its way to PyPI eventually, I'll be disappointed :slightly_smiling_face:

Performance

I wrote an entire report which can be read here: https://gist.github.com/ichard26/b996ccf410422b44fcd80fb158e05b0d, but the TL;DR is:

• Formatting with safety checks: 1.93x faster
• Formatting without safety checks: ...

[bright glacier]

Finally! I got my mypyc PR up, I still have a lot of work to do, but this is a major milestone

[silent apex]

If you need Windows testing, I can do that any time if you tell me what I'd have to do

[bright glacier]

Uh yeah, I need to come up with better instructions first as noted in the PR description first 😅

[silent apex]

Oh I uh, apparently can't read

[bright glacier]

Haha, no worries, I wrote perhaps too much :p

[neon loom]

I've never liked it

[austere lava]

did pipenv solve the multi-platform problem already?

when going on windows you couldn't pipenv install because the lockfile sometimes had linux specific binaries

[neon loom]

It had a 2021.5.29 release ... and a commit 8 days ago - So getting more action that it use to

Has > 500 issues tho. Makes me feel better about black now

[frozen badge]

tbh i think something like pipenv is inherently more prone to bugs due to its critical and highll complex natuer

[plain atlas]

i mean

poetry also is complex

[flat kraken]

ichard26

[psf/black] Pull request review submitted: #2430 Remove \`language\_version\` for pre\-commit

Please add an changelog entry but other than that this seems fine to me. Thank you!

P.S. for added convience here's a diff you can use:

diff --git a/CHANGES.md b/CHANGES.md
index a678aae..a4b8e01 100644
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -7,6 +7,11 @@
 - Add support for formatting Jupyter Notebook files (#2357)
 - Move from `appdirs` dependency to `platformdirs` (#2375)
 
+### Integrations
+
+- The provided pre-commit hooks no longer specify `language_version` to avoi...

[plain atlas]

@bright glacier how do you push to a pr fork branch?

[bright glacier]

^^

[verbal iron]

import os,time
os.system('cls')
moneygain = 1
money = 0
speed = 0
cost = 10
added = 1
add = 1
ps = True
while True:
    print(f'''
car speed:{speed}mph
cost to upgrade: ${cost} coins
money = ${money}
type (add) to gain ${moneygain}
type (upgrade) to add speed
''')
    commands = input(">")
    if commands == "add":
        money += moneygain
        os.system('cls')
    if commands != "add" or "upgrade":
        os.system('cls')
    if commands == "upgrade" and money >= cost:
        add += 1
        speed += add
        os.system('cls')
        money -= cost
        added += 1
        moneygain += added
        cost += 20

i made this game and here's the code I'm a beginner so please point out any mistakes

[boreal nacelle]

Is there you endlessly entering text?

@verbal iron

[verbal iron]

wdym

[boreal nacelle]

Write in full I'm not a English

[verbal iron]

ok

what do you mean

[boreal nacelle]

Like you have entered text into the console many times at the speed of a car?

@verbal iron

[verbal iron]

yes

[boreal nacelle]

import os,time
os.system('cls')
moneygain = 1
money = 0
speed = 0
cost = 10
added = 1
add = 1
ps = True
def main():
    print(f'''
car speed:{speed}mph
cost to upgrade: ${cost} coins
money = ${money}
type (add) to gain ${moneygain}
type (upgrade) to add speed
''')
    commands = input(">")
    if commands == "add":
        money += moneygain
        os.system('cls')
        main()
    if commands != "add" or "upgrade":
        os.system('cls')
        main()
    if commands == "upgrade":
        if money >= cost:
            add += 1
            speed += add
            os.system('cls')
            money -= cost
            added += 1
            moneygain += added
            cost += 20 
            main()
        else:
            print("Not enough coins")
            main()
main()```

Oh stop

[verbal iron]

that makes alot of sense

thanks

[boreal nacelle]

Stop

It's not working

Sorry idk:(

[verbal iron]

oh it was a good idea

[faint haven]

@verbal iron this isn't the correct channel for that, you should take a look at #❓｜how-to-get-help

[verbal iron]

ok

[graceful iris]

hi i have downloaded python @modern lance ping me when u have a suggestion 🙂

waht do i do now?

i want to learn how to code

oof wrong channel......

[verbal iron]

you can read books sign up for bootcamps and courses i dont recommend videos. reading books are very effective

@graceful iris

[bright glacier]

!resources

[late dew]

Resources

The Resources page on our website contains a list of hand-selected learning resources that we regularly recommend to both beginners and experts.

https://pythondiscord.com/pages/resources

[flat kraken]

JelleZijlstra

[psf/black] Pull request review submitted: #2430 Remove \`language\_version\` for pre\-commit

cooperlees

[psf/black] Pull request review submitted: #2430 Remove \`language\_version\` for pre\-commit

[flat kraken]

JelleZijlstra

[psf/black] Pull request review submitted: #2431 Implementing mypyc support pt\. 2

Thanks for all your work on this! I'm really excited to see this PR.

[flat kraken]

ichard26

[psf/black] Pull request review submitted: #2430 Remove \`language\_version\` for pre\-commit

Ah yeah, didn't realize this was from an organization repository. No worries then, it's not worth the trouble even if it was technically feasible. I also didn't know that only users could permit that, you really do learn something everyday don't you heh.

Congrats on your first contribution to psf/black :tada:

ichard26

[psf/black] Pull request closed: #2430 Remove \`language\_version\` for pre\-commit

ichard26

[black:main] 1 new commit

ef7c45f Remove language_version for pre-commit (#2430) - aneeshusa

[flat kraken]

ichard26

[psf/black] Issue opened: #2432 Avoid looking at \_\_name\_\_ attribute in \`black\.linegen\.run\_transformer\`

Right the transformer code looks at the transformer's __name__ attribute when deciding to return early: https://github.com/psf/black/blob/ef7c45f28132a7704a4549072ce5b272034fa196/src/black/linegen.py#L964-L967

This is kinda hacky and becomes quite a problem when integrating mypyc since nested functions don't keep that attribute due a bug (https://github.com/mypyc/mypyc/issues/884). The current workaround is to dynamtically create a wrapping class at runtime (via type() mind you :upside_d...

[graceful iris]

where do you find these "books" is it like online or something

[potent bone]

https://www.amazon.co.uk/s?k=python

[bright glacier]

I'll note that this entire topic is off topic here: but real python is a good source of free articles. they're more piecemeal like so if you prefer more structured guided learning, they might not be a good bet starting out

I'd suggest moving to something like #python-discussion.

[graceful iris]

ok

[flat kraken]

ichard26

[psf/black] Pull request review submitted: #2414 Present a more user\-friendly error if \.gitignore is invalid

Actually this doesn't go far enough, it didn't occur to me before but we actually support nested gitignores these days, and uhm, this PR doesn't fully fix the ungraceful crash when a nested gitignore is invalid.

❯ tree -a test
test
├── a
│   ├── a.py
│   └── .gitignore
└── a.py

1 directory, 3 files

~/programming/oss/black on gitignore [$?] via v3.8.5 (bm-venv) 
❯ dropblackcache -y ; black test
no caching was found!
Could not parse test/a/.gitignore: Invalid git pat...

[bright glacier]

So wait, how many mini-projects am I running for psf/black again?

• documentation

• developer experience

[plain atlas]

I think its 4?

*40

[bright glacier]

• tooling (blackbench, diff-shades)

• mypyc

• this isn't really a project but I'm trying to get the PR number down too

[plain atlas]

what are the other 35?

[bright glacier]

fortunately most of them don't exist :D

yet

I already have ideas of more stuff I could do, but I really need to limit myself from taking on more

[plain atlas]

hey, I'm working on 7 projects, could you help me with 10 of them? Again, just 12 need your assistance, so join me with working on 17 projects ty

🙃

[plain atlas]

yeah that's a good idea

[flat kraken]

the-wondersmith

[psf/black] Issue closed: #2405 Simple CLI "client" For blackd

john-science

[psf/black] Pull request opened: #2433 Throwing a more descriptive warning when reading a bad \.gitignore

This is in response to the ticket: https://github.com/psf/black/issues/2359

A particularly unhelpful error was thrown before, I feel this will be a lot more helpful to users trying to understand what went wrong.

[neon loom]

Just making sure you've all seen it cause it's pretty rad - got to https://github.com/psf/black (or any github repo) and hit '.' on your keyboard (You must be logged in ...)

VSCode on demand!

And you can even have vim plugin!

No python / black plugins yet - Wonder if we can help make that happen somehow 😄

<3's VSCode

[bright glacier]

!remind 5H "look into this message re. bandersnatch #tools-and-devops message"

[late dew]

You got it!

Your reminder will arrive <t:1629412535:R>!

ID: 2830

[late dew]

@bright glacier

It has arrived!

Here's your reminder: "look into this message re. bandersnatch #tools-and-devops message"
[Jump back to when you created the reminder](#black-formatter message)

[bright glacier]

!remind 16H "Further think through the idea of doing open-house style mentorship via PyDis, xref #ot1-this-regex-is-impossible message"

[late dew]

Sure thing!

Your reminder will arrive <t:1629485233:R>!

ID: 2834

[flat kraken]

glyph

[psf/black] Issue opened: #2434 extremely long left hand side expressions are not reflowed at all

Describe the style change

Sometimes you need to access attributes deep within another object or with extremely long names, and Black helpfully tries to wrap the right hand side of the expression to avoid blowing the length limit, but doesn't touch the right.

Examples in the current Black style

(
    something.something_else.hideously_long_attribute_name.other_property.even_deeper.finally_getting_close.okay
) = 1

Desired style

ichard26

[psf/black] Issue reopened: #2405 Simple CLI "client" For blackd

[plain atlas]

TBH I always thought some of the purpose of blackd was to support configuration rules throughout an organisation without having to put the configuration on every repo

[obsidian gale]

Hello

What project are you guys working on

[plain atlas]

#black-formatter

[mossy light]

SAME ICON

[late dew]

@bright glacier

It has arrived!

Here's your reminder: "Further think through the idea of doing open-house style mentorship via PyDis, xref #ot1-this-regex-is-impossible message"
[Jump back to when you created the reminder](#black-formatter message)

[plain atlas]

👀

hello ichard

[bright glacier]

Ah yeah the mentorship thingy

[bright glacier]

@dense jungle I'm reviewing PR psf/black#2414 for what is hopefully the last time. I'm pretty likely to approve and merge, but I wanted to let you give any feedback first.

[toxic storm]

GitHub

Present a more user-friendly error if .gitignore is invalid

[bright glacier]

Well, actually I guess we do have two PRs to consider. I prefer this one mostly because it hides away the traceback (which can be scary!) but that's just me.

[flat kraken]

ichard26

[psf/black] Pull request review submitted: #2414 Present a more user\-friendly error if \.gitignore is invalid

Thanks for the quick update!

[bright glacier]

Hmm, re psf/black#2412, I suppose the docker image we provide does fall under usage and configuration but it doesn't sit well with me.

[toxic storm]

GitHub

Docker image usage description

[bright glacier]

Thinking about it some more, the current location could work especially since we probably want to document the OS native binaries we also provide. A "Alternative installation" (please bikeshed this -- it's horrible :p) document could easily work.

[flat kraken]

JelleZijlstra

[psf/black] Pull request review submitted: #2414 Present a more user\-friendly error if \.gitignore is invalid

[dense jungle]

let's land it. The other PR seems less thoroough

[bright glacier]

Awesome, I'm happy to see the PR count go down ^^

Also I get to +1 my closed issue count so hooray 🎉

[dense jungle]

oh no you're going to beat me

[bright glacier]

You still have a huge lead IIRC

Yep, ~40 lead still :p

[plain atlas]

...

brb creating 41 spam issues on psf/black (/s ofc) 😼

[flat kraken]

ichard26

[psf/black] Issue closed: #2359 Present a more user\-friendly error if \.gitgignore

ichard26

[psf/black] Pull request closed: #2414 Present a more user\-friendly error if \.gitignore is invalid

ichard26

[black:main] 1 new commit

104aec5 Present a more user-friendly error if .gitignor... - nipunn1313

ichard26

[psf/black] Pull request closed: #2433 Throwing a more descriptive warning when reading a bad \.gitignore

ichard26

[psf/black] Pull request review submitted: #2416 document jupyter hook

Awesome to see more automation come alive -- love it! Thank you!

[bright glacier]

I'm using codespell to check my documentation changes now, so hopefully Jelle won't catch as many dumb typos these days :p

Ooo I should run diff-shades on psf/black#2278. I'm stil working on a markdown output but this PR should be a good experiment into whether it serves its mission as a mypy-primer for black well enough.

[toxic storm]

GitHub

Parenthesize conditional expressions

[bright glacier]

Haha, you know that repository is quite active when CI is maxed out. Some jobs are in queue right now :p

Good timing as I got to go now for the time being 😄

[flat kraken]

john-science

[psf/black] Issue opened: #2435 Clean Trailing Whitespace

Describe the style change

Black does not currently clean trailing whitespace. This is strange, and does not jive well with any major Python linter I can find (say Pylint).

Additional context

The only reason I can think of to not do this now is backwards compatibility.

If ticket isn't immediately closed for being annoying to the dev team, I would happily implement it.

Thanks.

[bright glacier]

Wait what, does black not handle that ???

Eh, I'll triage it later ^^

[plain atlas]

TBH because I have

      - id: trailing-whitespace
``` in my pre-commit config *and* have my editor set to trim trailing whtespace before save I've never noticed it

[bright glacier]

heya, I'd appreciate reviews on the following PRs:

• psf/black#2416 "document jupyter hook" (this one is the lowest priority since it looks OK to me)
• psf/black#2412 "Docker image usage description"
• psf/black#2384 "Stop changing return type annotations to tuples"
• psf/black#2278 "Parenthesize conditional expressions"

[toxic storm]

GitHub

Stop changing return type annotations to tuples
Parenthesize conditional expressions
document jupyter hook
Docker image usage description

[bright glacier]

RE. psf/black#1574 I think the best course of action would be to actually stop suggesting stable as much as possible and instead specify the current latest version. We can extend the script PR 2416 introduces that checks if the pre-commit rev value is correct to cover these cases too.

[toxic storm]

GitHub

docs(gh_actions): fixes #1573

[bright glacier]

Whelp, I was going to run diff-shades against the "Parenthesize conditional expressions" PR but it turns out diff-shades needs a lot more development before it's CI ready

While I like the changes for the most part, the fact we don't hug brackets is really bothering me

just feels like a lot of whitespace

[plain atlas]

hm

side note: anyone know of a toml formatter because my whitespace is wild in those

[bright glacier]

                 param = [
-                    processors[key](compiled_params[key])
-                    if key in processors
-                    else compiled_params[key]
+                    (
+                        processors[key](compiled_params[key])
+                        if key in processors
+                        else compiled_params[key]
+                    )
                     for key in positiontup
                  ]

eg. feels a bit wasteful here

         assert db_request.route_path.calls == [
-            pretend.call("manage.project.roles", project_name=project.name)
-            if desired_role == "Owner"
-            else pretend.call("packaging.project", name=project.name)
+            (
+                pretend.call("manage.project.roles", project_name=project.name)
+                if desired_role == "Owner"
+                else pretend.call("packaging.project", name=project.name)
+            )
          ]

ditto

[plain atlas]

inclined to agree tbh

which pr?

[bright glacier]

• psf/black#2278 "Parenthesize conditional expressions"

[toxic storm]

GitHub

Parenthesize conditional expressions

[bright glacier]

like there's a lot more cases where IMO it improves things, but the above two don't sit well with me, OTOH there's this:

-            lambda tab: tab.c.q.in_(vv)
-            if tab.name == "t1"
-            else tab.c.q.in_(qq),
+            lambda tab: (
+                tab.c.q.in_(vv) if tab.name == "t1" else tab.c.q.in_(qq)
+            ),

which is a massive improvement

[plain atlas]

otoh?

[bright glacier]

on the other hand

[plain atlas]

on the other hand

yesh

huh, pyanalyze

jelle 👀

[bright glacier]

-                    value="singular%d" % ii if ii % 4 == 0 else None
+                    value=("singular%d" % ii) if ii % 4 == 0 else None

ok this one is just cute even tho I don' think it's necessary :p

[plain atlas]

side note: does jelle work for quora?

[bright glacier]

IIRC yes.

[plain atlas]

given that https://github.com/psf/black/pull/2278#pullrequestreview-702596888 refers to https://github.com/quora/pyanalyze which is part of quora, idk. Ofc they also don't have the collab or member badge so idk
edit: but they do have push perms so

[bright glacier]

You can hide your membership in an organization ^^

[plain atlas]

true, but it typically will say collaborator regardless of that

[bright glacier]

It won't actually

[plain atlas]

huh, possible til

[bright glacier]

Take a look here: pypa/bandersnatch#980

[toxic storm]

GitHub

Bump codecov/codecov-action from 1 to 2.0.2

[bright glacier]

Cooper and I are both maintainers, but his membership is private and mine isn't

Cooper probably says "contributor" for you, no?

[plain atlas]

yea

afk

ergh

[plain atlas]

that pull got me on a sidetrack to fix my own ci

https://github.com/pypa/bandersnatch/blob/3b78e040a330e3308389e3d32d8c13e8b8c41af5/.github/workflows/ci.yml#L12 is key

[late dew]

.github/workflows/ci.yml line 12

if: github.event_name == 'push' || github.event.pull_request.head.repo.full_name != github.repository```

[plain atlas]

.bm

i swear

[plain atlas]

yay

my isort and black are fighting

[bright glacier]

Probably need --profile black then

[plain atlas]

already is there--

are you

TIL

in the tox.ini file

if its profile = black it actually means nothing

it has to be profile=black

[bright glacier]

And one more reason why I hate INIs ^^

[plain atlas]

brb making a commit to protected main to fix this

because

yeah

also

king-arthur#2

[toxic storm]

GitHub

Cloudflare Cache purge command

[plain atlas]

was #4

cc @delicate marsh, just sent a pr to fix it, again

whups

[plain atlas]

i am a bad tester; it is still not working

strangely

the line lengths are configured the same

and yet, they're fighting

[bright glacier]

How so?

[plain atlas]

isort:

from modmail.utils.threads import (
    Target, ThreadAlreadyExistsError, ThreadException, ThreadNotFoundError, Ticket, is_modmail_thread,
)

black:```py
from modmail.utils.threads import (
Target,
ThreadAlreadyExistsError,
ThreadException,
ThreadNotFoundError,
Ticket,
is_modmail_thread,
)

[bright glacier]

Oh that looks like the magic trailing comma thingy, IIRC there is an issue on the isort tracker about this, lemme find it.

[plain atlas]

nice thing is if I remove the trailing comma isort puts it back

[bright glacier]

pycqa/isort#1683

[toxic storm]

GitHub

Support for magic trailing comma

[plain atlas]

well

lmao

solution: remove some of these imports since flake8 complains about them anyways

(since they aren't used in the code)

tl;dr currently doing a refactor of this file into multiple files since it deserves them

[bright glacier]

I just finished doing like the 6th? pipenv lock trying to add the test dependencies to the lockfile. I think the lockfile shouldn't be broken this time but I still need to test it.

I literally had to 1) do a clean clone of psf/black, 2) create a Python 3.6.12 virtualenv, and 3) rm Pipfile.lock before locking so it wouldn't misbehave. Lovely.

[plain atlas]

...why are you not trying to migrate to poetry or smth?

[bright glacier]

I might look into that soon, but this is probably still less work than moving to poetry.

Does poetry support custom build scripts like setuptools' setup.py?

If not, it will be a very unpleasant experience packaging up a Rust, C, and Python distribution.

Unless there's good plugins for that, but IDK.

[little parrot]

its in alpha or beta or something