this seems like a real bug
#black-formatter
dense jungle
flat krakenBOT
mint barn
Hi, i hope i can get some help here. I'm a bit clueless right now. Our Black setup changed with the new release but because we don't want to deal with it now, we try to pin the version to the one we used all the time: v22.3.0 in this PR:
https://github.com/cryptoadvance/specter-desktop/pull/2121/files
But now the formatting seem to differ a lot with my local black installation and i can't explain that. Also the debugging doesn't seem to be helpful.
ANy hints what i can do? We're happy black user since more than a year.
dense jungle
Hi, i hope i can get some help here. I'm a bit clueless right now. Our Black set...
the version of the action doesn't determine the version of the formatting it uses. see https://black.readthedocs.io/en/stable/integrations/github_actions.html for how to set the version with the action
and yes this is highly confusing and we will change it
mint barn
and yes this is highly confusing and we will change it
Ok, but how can it differ if i haven't specified a specific version, neither in my pre-commit-config not in the github-action config?
$ cat .pre-commit-config.yaml
repos:
- repo: https://github.com/ambv/black
rev: 22.3.0
hooks:
- id: black
language_version: python3.10
$ black --version
black, 22.3.0 (compiled: yes)
the action will use the latest version (23.1.0) by default
and your local config is using 22.3.0
mint barn
the action will use the latest version (23.1.0) by default
Ahh, ok, got it. that's indeed ... counterintuitive.
mint barn
and your local config is using 22.3.0
Thanks!
dense jungle
yeah I think the original motivation was so that you could use the action to format with Black versions before the one that we released the action with
lament crow
and yes this is highly confusing and we will change it
Anything concrete or is it up for grabs?
dense jungle
https://github.com/psf/black/issues/3382 don't think there's anything blocking this
from @bright glacier's comment seems like this might not be trivial, but I definitely think this is something we should fix soon
lament crow
I'll look into it today
dense jungle
thanks!
lament crow
Got me interested
flat krakenBOT
glossy oak
Any devs and/or cooperlees here ? https://github.com/psf/black/pull/3531#issuecomment-1414144301 got me thinking if i understand the comment correctly (context; im trying to help with github actions integration, my pull request async communications via pr comments is a bit "meh" 🙂 )
neon loom
@neon loom comes from time to time 🙂
Please always tag me as I don't read the channel a lot
What's confusing? (Thanks for taking the feedback tho)
bright glacier
Regardless of however this turns out, I think it's worth adding a guide to the docs about handling Black upgrades, mentioning pinning and --required-version, etc. etc.
I do wish --required-version worked nicely with pre-commit but of course it doesn't.
https://github.com/psf/black/issues/2493 yeah this makes --required-version rather unusable. It might be worth patching Black to only warn if version information is clearly unavailable because we're running under pre-commit).
bright glacier
Heh, my Black 23.1.0 blog post ranks pretty highly on Google!
I was curious to why I was getting so much traffic from Google since usually Google isn't a major source of traffic for these posts.
ah of course, my mypyc stuff is still the most popular
glossy oak
@neon loom so if i got it correctly, the "optimal" way now would be to write the json from black itself and then action just utilizes the json to report it to $GITHUB_OUTPUT ?
glossy oak
<@813813914181697566> so if i got it correctly, the "optimal" way now would be ...
thats at least how i understood your comment on the pr .. eg, if black would create a json report directly, that sort of functionality would also support possible other task runners besides github actions and then the action/main.yaml would just use the generated json .. If that is the case, im wondering a bit about maybe there should be a possibility to set the location where the json would go via command line arg ..
(and i btw, truely like this approach better)
glossy oak
@neon loom that said, if the user provides "version" for black, and the version is old enough to not support generating the json, that would cause some issues and worrying part regarding that is that action.yaml still would have reference to outputs but there would be none in case of older black ..
ivory quest
ah of course, my mypyc stuff is still the most popular
I'd say it's the only good source on applying mypyc to production project with some real-world problems
neon loom
<@813813914181697566> that said, if the user provides "version" for black, and t...
Yeah, I feel you just error there and say only supported >= 23.2.0 if it makes it to that version.
neon loom
I'd say it's the only good source on applying mypyc to production project with s...
mypy itself was a pretty good win to …
ivory quest
mypy itself was a pretty good win to …
but was the whole process covered like it was in the case of black?
neon loom
I remember reading something. The detail was no where like Richard’s tho.
flat krakenBOT
bright glacier
I got fed up with seeing T: design. Sometimes stepping from a project helps with perspective :)
bright glacier
I also updated https://ichard26.github.io/ghstats/ because why not
unique remnant
I Hi everyone, I will like to contribute to open source project but need a mentor so as to guide me. I don't know where to start. There is so much information and most of the newbie friendly open source I find are from two years ago.
If someone have already done contributions to open source I will highly appreciate the guidance
Thanks
ivory quest
I also updated <https://ichard26.github.io/ghstats/> because why not
what is the license on that (and the source)? I would like to copy this for other projects 😄
bright glacier
I might just make it configurable. I know @lament crow also wanted to use it for their own projects. I want to use it for mypyc anyway so it'd be good to remove all of the hardcoding for Black.
Currently all of the front-end and helper scripts assume Black.
or at least I'll make it a lot easier to bootstrap your own copy.
ivory quest
I Hi everyone, I will like to contribute to open source project but need a ment...
my suggestion would be to find the project you enjoy and look into open issues, join discord or other form of forum the project has (if they do) and start contributing. Contributions are both issue triage (testing older issues if they are still valid, pointing invalid issues as closing candidates to maintainers etc) and actually working on code and PRs. There is not much more philosophy to open source. Have fun at it!
ivory quest
I might just make it configurable. I know <@176070082584248320> also wanted to u...
Would be nice. If you were to add license, I might put some work into making it easier to configure and PR it
(as if I didn't have too much work already lol)
bright glacier
I'll try to take a look later today. It's actually split between two repos right now which probably isn't particularly friendly.
bright glacier
(as if I didn't have too much work already lol)
You tell me, I'm split between black and mypyc :p
ivory quest
I'll try to take a look later today. It's actually split between two repos right...
yeah, I noticed. didn't understand the why, just blindly accepted the choice
ivory quest
You tell me, I'm split between black and mypyc :p
poetry (3 projects that come into it), blinker, pendulum and few other projects I want to get into
bright glacier
meh, I'll just get started right now. I was going to try and fix one more mypyc bug but this seems to be harder than I previously thought 😅
ivory quest
meh, I'll just get started right now. I was going to try and fix one more mypyc ...
nice 😄
bright glacier
gosh the python code involved here sucks
def oldest(self) -> Issue:
lowest = min(i for i in self._issues.keys())
return self._issues[lowest]
def newest(self) -> Issue:
newest = max(i for i in self._issues.keys())
return self._issues[newest]
wtf
unique remnant
@ivory quest thanks!
flat krakenBOT
Is your feature request related to a problem? Please describe.
Previously black allowed to have a newline after a code opening black. This can indeed improve readability.
Compare for instance this
with
There are two reasons, why a single newline at the ...
drowsy sonnet
my suggestion would be to find the project you enjoy and look into open issues, ...
Well... There's a lot of philosophy + political background to the open source movement at large, but that's off topic. I mainly wanna flag that "not much philosophy" isn't correct. 😅
ivory quest
Well... There's a lot of philosophy + political background to the open source mo...
I meant more in terms of "how to start contributing", but I guess some phrases aren't too good when translated to English. Of course there is a lot of philosophy and politics in open source, but for someone just joining the scene, it doesn't really matter most of the time
flat krakenBOT
flat krakenBOT
flat krakenBOT
green mulch
```python
def oldest(self) -> Issue:
lowest = min(i for i in self._i...
O.o
where is it?
glossy oak
O.o
where is it?
almost makes me think that there's Story involved with code like that .. It could be a bad story but story never the less 😄
green mulch
def apply(self, func: Callable) -> Sequence[Issue]:
return list(map(self._issues.get, func(self._issues)))
def oldest(self) -> Issue:
return self.apply(min)[0]
def newest(self) -> Issue:
return self.apply(max)[0]
It's been refactored, but here's the file on an older commit. It makes up the GitHub Issues library I wrote a few years ago for issue data visualization. https://github.com/ichard26/ghstats/blob/7a026bf669f4263bb54a3993b3046f28181e7403/scripts/lib.py
green mulch
👍
flat krakenBOT
flat krakenBOT
flat krakenBOT
flat krakenBOT
neon loom
Wow - TIL mypy 1.0
drowsy sonnet
Wow.
gritty totem
hello everyone! 👋 I've noticed that there is an increasing number of lint/format/build tools that get great performance wins by using Go or Rust under the hood -- esbuild, ruff, and rome come to mind here. I'm curious how you all have thought about using these languages or others for performance-critical code or if there's any ongoing work in this vein already.
Apologies in advance if this is asked a lot -- I did a Discord search in this channel but didn't see much relating to this.
mint barn
hey
dense jungle
hello everyone! 👋 I've noticed that there is an increasing number of lint/forma...
no current plans. We are using mypyc for performance and we'll hopefully make more performance optimizations in the future. A rewrite in a totally different language would be a huge undertaking, beyond the resources available to the team
gritty totem
no current plans. We are using mypyc for performance and we'll hopefully make mo...
Thanks for the reply and context! Makes sense, I certainly understand that a team can't pursue everything. Thanks for everything you do on Black, it's a lovely tool.
bright glacier
We're lagging behind on regular maintenance of the project already. Maintainer time is extremely limited so even the rewrite of a small part of the project would likely stall and never see the light of day.
I also don't know Go or Rust. I know some C but I don't think any of us would want to rewrite Black into C 😅
cyan panther
Is there a way to have black format Protocol class definitions like they're stubbed pyi files?
ivory quest
We're lagging behind on regular maintenance of the project already. Maintainer t...
I feel like the question of "rewrite in <insert compiled language>" is going to appear on every runnable project soon. we have such questions in Poetry on a regular basis and the answer is pretty much the same as yours...
bright glacier
"If it's not written in <compiled language>, it must be too slow!!"
silent apex
hello everyone! 👋 I've noticed that there is an increasing number of lint/forma...
out of curiosity, is there "performance-critical code" that you're thinking of in particular?
usually black is rather fast, and i can't imagine what would require it to be faster than it is now
bright glacier
people using it on save with somewhat big files in IDEs complain black's too slow
little parrot
black in C would certainly be a sight
silent apex
people using it on save with somewhat big files in IDEs complain black's too slo...
huh.
bright glacier
there's a relatively easy optimization that we can do to improve formatting performance by 50%, but it means dealing with a lot of edge cases
may look into that eventually
ivory quest
people using it on save with somewhat big files in IDEs complain black's too slo...
I guess that when they have autosave enabled and it runs every X seconds, it might add up to a delay
dense jungle
there's a relatively easy optimization that we can do to improve formatting perf...
which one is that?
little parrot
turning off --safe?
bright glacier
wait actually I might be wrong, but the forced second pass seems unideal
I just remembered that we probably cut the third pass in the safety check though so it shouldn't have affected performance much.
ivory quest
wait, so black does more than 2 passes?
bright glacier
Hmm, so format_str does two passes if changes are made, and then assert_stable will do one more pass. A long time ago we only did at most two passes.
ivory quest
it's always fascinating to learn how stuff like this works...
bright glacier
I guess that when they have autosave enabled and it runs every X seconds, it mig...
According to an user report, Black takes 200ms per 1K LOC
With Rust they claim it would be basically instantaneous.
bright glacier
wait, so black does more than 2 passes?
Yeah we forced one more pass to work around a nasty stability bug with trailing commas and optional parentheses.
dense jungle
Hmm, so `format_str` does two passes if changes are made, and then `assert_stabl...
I thought we made it so we didn't do the third pass if the second pass didn't change anything? that would be a nice optimization
ivory quest
people then: *computer does stuff* haha, it's thinking!
people now: why do I have to blink to see the result?!
bright glacier
We did not. https://github.com/psf/black/blob/d9b8a6407e2f46304a8d36b18e4a73d8e0613519/src/black/__init__.py#L1079-L1085 format_str does not communicate whether it did one or two passes.
late dewBOT
src/black/__init__.py lines 1079 to 1085
dst_contents = _format_str_once(src_contents, mode=mode)
# Forced second pass to work around optional trailing commas (becoming
# forced trailing commas on pass 2) interacting differently with optional
# parentheses. Admittedly ugly.
if src_contents != dst_contents:
return _format_str_once(dst_contents, mode=mode)
return dst_contents```Seems like a mostly safe optimization though? Not sure how we'd implement it though without breaking API compat.
dense jungle
Seems like a mostly safe optimization though? Not sure how we'd implement it tho...
add a new _format_str_internal function that returns (data, was_changed)
format_str calls it; in format_file_contents we call _format_str_internal
bright glacier
So format_file_contents calls _format_str_internal directly so it can decide whether to call assert_stable?
I guess since we never had safety or stability guarantees for format_str that's fine.
gritty totem
out of curiosity, is there "performance-critical code" that you're thinking of i...
I was referring abstractly to hot code in Black, but I'm not deeply familiar with the project internals or where the hot code might be.
As for the context of my original question, I recognize there's a pretty wearisome crowd of drive-by commenters asking their favorite projects to "Rewrite it in Rust" [https://transitiontech.ca/random/RIIR] and I could have made it clearer that I was asking out of curiosity and interest in contributing to such efforts. I've maintained enough projects to know how frustrating drive-by comments suggesting massive work items can be, and I hope that my question wasn't taken in that spirit
neon loom
I believe a lot of libcst is a python parser in rust …
And is our likely successor once we stop flogging blib2to3
@austere lava would know more 🙂
marble basalt
<@&831776746206265384> 🤷♂️
civic bane
!mute 960981532309340180
late dewBOT
:ok_hand: applied mute to @mint barn until <t:1676738298:f> (1 hour).
stone bough
Hi guys, how to make black format tuple and list like this?
CLASSES = (
'person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus', 'train',
'truck', 'boat', 'traffic light', 'fire hydrant', 'stop sign',
'parking meter', 'bench', 'bird', 'cat', 'dog', 'horse', 'sheep',
'cow', 'elephant', 'bear', 'zebra', 'giraffe', 'backpack', 'umbrella',
'handbag', 'tie', 'suitcase', 'frisbee', 'skis', 'snowboard',
'sports ball', 'kite', 'baseball bat', 'baseball glove', 'skateboard',
'surfboard', 'tennis racket', 'bottle', 'wine glass', 'cup', 'fork',
'knife', 'spoon', 'bowl', 'banana', 'apple', 'sandwich', 'orange',
'broccoli', 'carrot', 'hot dog', 'pizza', 'donut', 'cake', 'chair',
'couch', 'potted plant', 'bed', 'dining table', 'toilet', 'tv',
'laptop', 'mouse', 'remote', 'keyboard', 'cell phone', 'microwave',
'oven', 'toaster', 'sink', 'refrigerator', 'book', 'clock', 'vase',
'scissors', 'teddy bear', 'hair drier', 'toothbrush', 'banner',
'blanket', 'branch', 'bridge', 'building-other', 'bush', 'cabinet',
'cage', 'cardboard', 'carpet', 'ceiling-other', 'ceiling-tile',
'cloth', 'clothes', 'clouds', 'counter', 'cupboard', 'curtain',
'window-blind', 'window-other', 'wood')
Black makes long tuple one item per line.
tropic verge
Hi! I want to apply black formatting on my Python project, including build/ directory which is excluded by default. I tried using --exclude but then .gitignore is ignored and other bad things happen. Is there an elegant solution to my use case?
lament crow
Hi guys, how to make black format tuple and list like this?
```python
CLASSES = ...
You can't, Black purposefully does not have many formatting options. You can make Black not format part of the code by using # fmt: off and # fmt: on comments but that means you'll have to format that part manually
lament crow
Hi! I want to apply black formatting on my Python project, including `build/` di...
exclude has a default value which you can override but you have to ensure that you keep the existing excludes that you care about while removing build/ which you don't care about
The default value of exclude can be found in black --help
The output of that command can also be found here: https://black.readthedocs.io/en/stable/usage_and_configuration/the_basics.html#command-line-options
I'd suggest specifying such options in a configuration file, not as cli arguments so that anyone working on the project gets them when they use Black
exclude is a regex btw
tropic verge
@lament crow , when overriding excludes, .gitignore is ignored. According to the docs it is an expected (and the observed) behavior, but I don't understand why and how can it be unignored (https://black.readthedocs.io/en/stable/usage_and_configuration/file_collection_and_discovery.html#gitignore)
The suggested solution, to use --extend-exclude isn't a good fit for my problem, as I want to remove a value from the default exclusion list
bright glacier
There's no recourse. That's as designed as per https://github.com/psf/black/issues/2164.
Might be better to allow .gitignore to be configured using a flag. Not sure how common either of these use-cases are though.
tropic verge
thanks @bright glacier, I can't say how common it is, but I can say that if you happen to have a build directory that you would want to apply Black on, your life becomes black 😄
If having this feature controlled by an argument is an option, I would be happy to open a pull request for it
bright glacier
You can always pass the files in the build directory to black explicitly.
plain atlas
are there any plans to make something like https://github.com/charliermarsh/ruff-pre-commit to allow usage of the mypyc wheels in pre-commit?
pre-commit builds the package when the hook is in the same repository, but when using a mirror it can download from pypi and instead use prebuilt wheels
lament crow
<@176070082584248320> , when overriding excludes, .gitignore is ignored. Accordi...
Huh, that's weird
bright glacier
are there any plans to make something like https://github.com/charliermarsh/ruff...
I want to, but I haven't pushed for it. It's a bit unfortunate that we have to create a mirror.
stone bough
You can't, Black purposefully does not have many formatting options. You *can* m...
Got it. Thank you.
plain atlas
I believe a lot of libcst is a python parser in rust …
ye, and its quite nice
flat krakenBOT
flat krakenBOT
flat krakenBOT
half notch
Hello I am trying to use black jetbrains integration. I followed the instructions.
pip install black[d]
Requirement already satisfied: black[d] in c:\users\\desktop\statwolfbot\test\venv\lib\site-packages (23.1.0)
Requirement already satisfied: aiohttp>=3.7.4 in c:\users\\desktop\statwolfbot\test\venv\lib\site-packages (from black[d]) (3.8.3)
Requirement already satisfied: charset-normalizer<3.0,>=2.0 in c:\users\\desktop\statwolfbot\test\venv\lib\site-packages (from aiohttp>=3.7.4->black[d]) (2.1.1)
Requirement already satisfied: yarl<2.0,>=1.0 in c:\users\\desktop\statwolfbot\test\venv\lib\site-packages (from aiohttp>=3.7.4->black[d]) (1.8.1)
Requirement already satisfied: async-timeout<5.0,>=4.0.0a3 in c:\users\\desktop\statwolfbot\test\venv\lib\site-packages (from aiohttp>=3.7.4->black[d]) (4.0.2)
Requirement already satisfied: frozenlist>=1.1.1 in c:\users\\desktop\statwolfbot\test\venv\lib\site-packages (from aiohttp>=3.7.4->black[d]) (1.3.1)
Requirement already satisfied: multidict<7.0,>=4.5 in c:\users\\desktop\statwolfbot\test\venv\lib\site-packages (from aiohttp>=3.7.4->black[d]) (6.0.2)
Requirement already satisfied: attrs>=17.3.0 in c:\users\\desktop\statwolfbot\test\venv\lib\site-packages (from aiohttp>=3.7.4->black[d]) (22.1.0)
Requirement already satisfied: aiosignal>=1.1.2 in c:\users\\desktop\statwolfbot\test\venv\lib\site-packages (from aiohttp>=3.7.4->black[d]) (1.2.0)
Requirement already satisfied: colorama in c:\users\\desktop\statwolfbot\test\venv\lib\site-packages (from yarl<2.0,>=1.0->aiohttp>=3.7.4->black[d]) (3.4)
Clicking start gives
return _run_code(code, main_globals, None,
File "C:\Users\\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 86, in _run_code
exec(code, run_globals)
File "C:\Users\\AppData\Local\Programs\Python\Python310\Scripts\blackd.exe\__main__.py", line 4, in <module>
File "C:\Users\\AppData\Local\Programs\Python\Python310\lib\site-packages\blackd\__init__.py", line 13, in <module>
raise ImportError(
ImportError: aiohttp dependency is not installed: No module named 'aiohttp'. Please re-install black with the '[d]' extra install to obtain aiohttp_cors: `pip install black[d]`
have you followed instructions in black's documentation?
https://black.readthedocs.io/en/stable/integrations/editors.html#pycharm-intellij-idea
half notch
have you followed instructions in black's documentation?
https://black.readthedo...
Yes, uninstalled both black and the extension. Than started from scratch on the instructions. The only difference is I checked "Show changes on potential destructive format changes" or similar.
lament crow
It's most likely that the path to black you have is not the same black installation that you installed while following the instructions
assuming you ran the pip install command exactly as shown in the instructions, can you show the output of pip --version
half notch
I am not at my PC rn. This is likely true. I remember that when running pip list black was shown but no black[d] is this normal?
half notch
assuming you ran the `pip install` command exactly as shown in the instructions,...
My pip version is 22.1 but I think you meant blacks version?
lament crow
I wanted the output
aside from version it also shows the path
which is what I cared about here
half notch
Would it make sense to install black[d] globally instead of on the venv? Than it would be easy to get the right path.
But I'm not at my PC rn so I will have to try it later.
lament crow
You're not pointing at the executable in a venv here
half notch
Ok, I will uninstall and reinstall globally that should fix it. Or at least make it so I can manually tweak the path a bit.
Thank you
bright glacier
Huh.
flat krakenBOT
drowsy sonnet
https://github.com/Sawbez/brown
I'm gonna guess it's a $corp wanting to maintain a fork for their needs?
dense jungle
I'm gonna guess it's a $corp wanting to maintain a fork for their needs?
maybe? I didn't look too deeply but it didn't sound too corporate, e.g. there are Discord links for the authors
and they deleted all the docs
drowsy sonnet
Very weird.
flat krakenBOT
haughty summit
I'm trying to use Black with vim but I'm getting this error...I see that the vim script is trying to use a different python...anyway to make black use my pyenv python instead?
haughty summit
I'm trying to use Black with vim but I'm getting this error...I see that the vim...
nvm neoformat can handle it without the black vim plugin
flat krakenBOT
flat krakenBOT
flat krakenBOT
flat krakenBOT
flat krakenBOT
flat krakenBOT
slow ibex
black is formatting -2**3 as -(2**3)
why?
>black -c "-2**3"
-(2**3)
>black --version
black, 23.1.0 (compiled: yes)
Python (CPython) 3.11.0
it is a bit weird IMO
dense jungle
black is formatting `-2**3` as `-(2**3)`
why?
so you don't think it's (-2)**3
slow ibex
in math when you write -a^b it always mean -(a^b)
i think it is obvious enough that -a**b mean -(a**b) and not (-a)**b
i dont like this behaviour 😔
lament crow
I don't think it's universally true across languages though it is common. But either way, my first instinct is that unary operators bind most tightly (and in some languages aren't treated as operator during parsing but rather a part of the number) so even if mathematically it makes sense, I wouldn't be sure that this is how Python behaves. It all comes down to avoiding a potential confusion point.
bright glacier
I hate GHA sometimes..
plain atlas
what happened?
bright glacier
go ahead
half meteor
Looks like there is an Actions outage: https://www.githubstatus.com/
plain atlas
<https://github.com/psf/black/actions/runs/4300301186>
@tight bone btw i know you do a lot of ci troubleshooting, any ideas?
bright glacier
That explains why everything GHA seems to be broken.
plain atlas
thanks, apparently its an outage @tight bone
tight bone
<@321581335942529024> btw i know you do a lot of ci troubleshooting, any ideas?
Had the same issue a couple minutes ago, so yea :/
flat krakenBOT
slow ibex
Why type comments are still supported?
I thought they are useless since annotations exists
Mypy is also raising SyntaxError if there is invalid #type:smth inside code.
It happens very often to me when i comment big parts of code, because annotation type: smth becomes #type: smth and it is invalid type comment
dense jungle
Why type comments are still supported?
I thought they are useless since annotati...
yeah on mypy's side we have talked about deprecating them. note though that # type: ignore counts as a type comment and that's still important
slow ibex
def f(
type: type[int],
) -> int: ...
# after commenting out it becomes this:
# def f(
# type: type[int], ## <- this is a syntax error, very annoying
# ) -> int: ...
I never thought about # type: ignore like a type-comment, that's interesting
plain atlas
#black-formatter message
after nearly two years, I can now say this codebase is formatted with black.
flat krakenBOT
flat krakenBOT
late dewBOT
:incoming_envelope: :ok_hand: applied mute to @mint barn until <t:1678059817:f> (10 minutes) (reason: duplicates rule: sent 4 duplicated messages in 10s).
The <@&831776746206265384> have been alerted for review.
flat krakenBOT
late dewBOT
@vale heron Per Rule 6, your invite link has been removed. If you believe this was a mistake, please let staff know!
Our server rules can be found here: https://pythondiscord.com/pages/rules
flat krakenBOT
slim stirrup
!e [3.10] <print("You Are Hacked!")>
<@&267628507062992896> How To Check Code?
!e
late dewBOT
slim stirrup
!e code
late dewBOT
@slim stirrup :x: Your 3.11 eval job has completed with return code 1.
001 | Traceback (most recent call last):
002 | File "/home/main.py", line 1, in <module>
003 | code
004 | NameError: name 'code' is not defined
spring ether
@slim stirrupthis channel is for the black formatter project; please go to #bot-commands
soft stag
oh wow
That moment when you throw your code into black, and it just spits out the exact same thing!
feels nice
flat krakenBOT
flat krakenBOT
restive creek
hello ?
restive creek
what is this
lament crow
Read the channel topic
restive creek
i now but what is the project about
restive creek
ok thanks
flat krakenBOT
civic ore
I'm curious what are black devs thoughts on changes that https://github.com/google/pyink makes. I know some changes that have been upstreamed already, but the point 1 and 3 in the readme are ones that have pushed me away from black historically.
(I guess maybe also 2, but I suspect that one will be very controversial :)
dense jungle
I'm curious what are black devs thoughts on changes that https://github.com/goog...
2 and 3 are not things we're willing to change. 1 would be good to have though
lament crow
1 makes me think of https://lukasz.langa.pl/36380f86-6d28-4a55-962e-91c2c959db7a/
flat krakenBOT
half meteor
2 and 3 are not things we're willing to change. 1 would be good to have though
Pyink maintainer here. Adding to what Jelle said, I plan starting to upstream (1) in the coming weeks. No plans for (2) & (3).
flat krakenBOT
flat krakenBOT
flat krakenBOT
Hello, I just hear of this project so I installed to test it, following the usage indication on some file I got this
I installed it like pip install ruff
ruff version: 0.0.256
Python version 3.11.2
OS: Windows 10
silent apex
i wonder how they found their way to the black issue tracker from ruff
flat krakenBOT
mint barn
Guyssss, how do I skip check for one line in Black?
bright glacier
Guyssss, how do I skip check for one line in Black?
# fmt: skip on the end of the line
dense jungle
hm tests are broken on main
flat krakenBOT
silent apex
i dunno, i don't work on black
bright glacier
I'll take a look at that PR tomorrow Jelle. I need sleep.
half meteor
This should fix the bug introduced for blank lines
dense jungle
This should fix the bug introduced for blank lines
thanks, merging it now
plain atlas
hey where does black store its cache?
bright glacier
hey where does black store its cache?
flat krakenBOT
flat krakenBOT
arctic goblet
Hey I don't have single knowledge of git hub can you tell me how and where I can learn about GitHub
flat krakenBOT
flat krakenBOT
flat krakenBOT
flat krakenBOT
half meteor
Just checking, any chance there will be a new Black release soon?
dense jungle
Just checking, any chance there will be a new Black release soon?
yes I should do that. going to aim for end of this week
half meteor
yes I should do that. going to aim for end of this week
Thanks! Let me know if there is anything I could help.
half meteor
Thanks! Let me know if there is anything I could help.
(We need a new version for a change in blib2to3)
bright glacier
For anyone who is paying attention here, just an FYI that I've stepped away from maintainership of Black for the time being. I'll be around to help out with anything that may be within my domain (CI/automation/docs), but I'm not actively following Black's development any longer.
dense jungle
For anyone who is paying attention here, just an FYI that I've stepped away from...
thanks for all you work on Black! Your mypyc work made a huge difference to performance, and diff-shades made it so much easier to evaluate proposed style changes
flat krakenBOT
dense jungle
I should probably get https://github.com/psf/black/pull/3508 in too
flat krakenBOT
bright glacier
Congrats! 🎉
flat krakenBOT
flat krakenBOT
tough sage
it just reads the file and has a list of mistakes, then if it sees one of those in your code it will re-format it i'm guessing
solid adder
Yes
exotic stratus
Does installing black[d] implicitly install black as well? I've always installed both via pip so far, but my IDE is now stating that black is redundant with black[d] present
dense jungle
Does installing `black[d]` implicitly install `black` as well? I've always insta...
yes
exotic stratus
Thank you
Also found out that I can merge package extras as psycopg[binary,pool], nifty stuff
barren brook
Is psf/black#2787 still needed?
toxic stormBOT
flat krakenBOT
turbid dagger
The black formatter just formatted the pygame argument in this way. Maybe I am biased but I find this style of formatting less readable…
I am open to discussion
slow ibex
What do you think is more readable?
vague pewter
None of the above?
flat krakenBOT
mint barn
Short question about black: I use the option --no-color but black's output is still colorized. Especially the blue characters are difficult to read on the black background of my terminal. How can I switch off colorized output of black altogether?
Looks great ofcourse, but I don't need this.
And I can't read it
lament crow
Based on the documentation of that option, it only determines whether Black returns a colored diff or not (when --diff is used).
If this is something that Black maintainers want to change, this should, presumably, be as easy as setting ctx.color (ctx being click.Context) to the value of color flag
mint barn
If this is something that Black maintainers want to change, this should, presuma...
So, no option to have black suppress colorization. 👍 I'll see what I can do in the CLI to strip ANSI-codes.😞
flat krakenBOT
flat krakenBOT
nova tinsel
that seems.. very random
dense jungle
some things we are not meant to understand
flat krakenBOT
pearl spruce
I made a tool that makes Python CLI tools start much faster. With it, I can run black on a few files in 10-20ms on my laptop, vs. at least 120ms for the mypyc-compiled version. Especially great when combined with pre-commit in git pre-commit hooks.
Happy to get some early adopters checking it out!
https://github.com/taleinat/jumpthegun
dense jungle
I made a tool that makes Python CLI tools start much faster. With it, I can run ...
That sounds awesome, thanks! Do you have any resources on how it works? I assume it starts some sort of server that caches the process so you don't pay import costs every time. Some concerns I would have are: does it work on all OSes? what if memory usage from the cached process grows too large?
pearl spruce
That sounds awesome, thanks! Do you have any resources on how it works? I assume...
Hi Jelle! I haven't written anything about how it works yet. I'll add more documentation at some point.
Indeed, for each CLI tool, JumpTheGun runs a background daemon which loads Python and does the imports in advance. This is run the first time the tool is run via JumpTheGun. It stays running until the tool hasn't been run for a while (4 hours by default).
To stop the daemon, run jumpthegun stop <tool-name>, or just kill the daemon process.
bright glacier
FYI please feel free to CC me for anything mypyc related. I'm on the mypy(c) core team and I'd be happy to help!
half meteor
FYI please feel free to CC me for anything mypyc related. I'm on the mypy(c) cor...
Since you are already here, how can I build the Black mypyc wheels locally in the same way as CI?
pearl spruce
That sounds awesome, thanks! Do you have any resources on how it works? I assume...
Just realized I hadn't answered all of your questions...
- JumpTheGun works on Linux (tested on Ubuntu 20.04) and macOS.
- Memory usage should not balloon over time, since each invocation forks a new dedicated sub-process. (If the processes never terminate, that would cause memory usage to grow over time regardless of JumpTheGun.)
pearl spruce
Since you are already here, how can I build the Black mypyc wheels locally in th...
I can help with that, I did it recently. IIRC it basically consists of pip install cibuildwheel and then running it. Let me look up the details.
pearl spruce
Since you are already here, how can I build the Black mypyc wheels locally in th...
So looks like I did the following to build a mypyc-based wheel only for my local Linux machine:
0. Have Docker installed, the Docker daemon running, and permission to run the docker cli.
- Create a venv and run
pip install cibuildwheel cibuildwheel --platform linux --helpcibuildwheel --platform linux --print-build-identifierscibuildwheel --platform linux --only cp310-manylinux_x86_64pip install ./wheelhouse/black-22.12.1.dev37+g18fb884-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
half meteor
So looks like I did the following to build a mypyc-based wheel only for my local...
Thanks! This worked.
bright glacier
Thanks Tal. I've never actually used CIBW locally. (anyway CIBW is amazing <3)
slow ibex
Just realized I hadn't answered all of your questions...
* JumpTheGun works on ...
Are you planning to support Windows?
pearl spruce
Are you planning to support Windows?
Not initially, since both forking and bash don't work on Windows.
flat krakenBOT
gilded scarab
Hello everyone, is it possible to somehow start Black immediately automatically, that is, when I start PyCharm, does it also start in a separate terminal?
flat krakenBOT
flat krakenBOT
flat krakenBOT
pearl spruce
Hello everyone, is it possible to somehow start Black immediately automatically,...
What are you trying to achieve? Do you want black to be run (repeately, automatically) on your code as you edit files?
pearl spruce
I'll just share this here, here's my current preferred way to have black (via pre-commit) run automatically on code changes in PyCharm Professional via its "File Watchers" feature:
gilded scarab
Thanks
flat krakenBOT
flat krakenBOT
exotic stratus
black doesn't enforce some sort of an opinionated docstrings format?
exotic stratus
Hello everyone, is it possible to somehow start Black immediately automatically,...
You can run black as a daemon. First install black[d] via pip and install the BlackConnect plugin via pycharm. Access it under Tools > BlackConnect and set your preferences as well as run it
spring ether
import os, json, math, csv, fasttext, nltk, spacy, subprocess, copy, random
import pandas as pd
import numpy as np
import os, json, csv, fasttext, nltk, spacy, subprocess, copy
why didn't black put each of these on different lines?
and remove duplicates?
fossil scaffold
Does anyone use Black on Airflow DAG files?
fossil scaffold
and remove duplicates?
I'm guessing the duplicates are there because while an import should be idempotent it doesn't have to be?
spring ether
I'm guessing the duplicates are there because while an import should be idempote...
I guess
fossil scaffold
For an Airflow DAG I'm non-plussed by the many lines of graph setup that went from:
[
dataset_service_tablename_A_ready_sensor,
dataset_service_tablename_B_ready_sensor,
dataset_service_tablename_C_ready_sensor,
] >> prepare_cluster >> [
parallel_job_on_A,
parallel_job_on_B,
parallel_job_on_C,
parallel_job_on_A_and_B,
parallel_job_on_B_and_C,
]
to
(
[
dataset_service_tablename_A_ready_sensor,
dataset_service_tablename_B_ready_sensor,
dataset_service_tablename_C_ready_sensor,
]
>> prepare_cluster
>> [
parallel_job_on_A,
parallel_job_on_B,
parallel_job_on_C,
parallel_job_on_A_and_B,
parallel_job_on_B_and_C,
]
)```It's mentioned that --preview has some cases where parenthesis aren't added… but this doesn't seem to be one of them
fossil scaffold
The extra indentation makes it stand out next to lines like:
cleanup_cluster >> send_message
cleanup_cluster >> [validate_A, validate_B] >> update_status
cerulean shore
Is there any way to influence how Black wraps long boolean expressions (chains of and/or)?
I think it formats this extremely awkwardly here:
# Before
return (
self.file >= other.file
and listify(self.version) > listify(other.version)
)
# After Black:
return self.file >= other.file and listify(self.version) > listify(
other.version
)
Why would Black line-break inside a parameter list when it could instead break on the and?
If I add a third condition to this expression (like and fooz == baz) it will format it how I originally did. Is there any way I can influence Black to do this sooner, like a trailing comma can do for dicts and argument lists?
fossil scaffold
Maybe it should keep input expression parens. IDK there's probably cases were removing excess parens also looks better.
flat krakenBOT
late dewBOT
:x: failed to apply.
ivory quest
```py
import os, json, math, csv, fasttext, nltk, spacy, subprocess, copy, rando...
black doesn't touch imports. use isort or ruff for import management
flat krakenBOT
flat krakenBOT
flat krakenBOT
fossil scaffold
It seems I misinterpreted this channel as discussion on style choices or unexpected changes
dense jungle
It seems I misinterpreted this channel as discussion on style choices or unexpec...
That's a reasonable use for the channel. But discussion of style choices can be kind of endless, so our answer might just be "Black will keep doing what it's doing"
fossil scaffold
It currently looks like a place to catchup on issue updates… or maybe get some help in usage? I just want to be sure I understand the meaning of "the development and usage of the tool"; whatever works for the primary users.
lament crow
You can talk about pretty much everything related to Black so that would include things like why Black chooses some style, development, usage, issues people encounter, features they want. Some of these things often make more sense as a trackable Issue on GitHub but there's no a solid line between them
You only have seen issue updates here recently probably because there wasn't much to talk about. I think Jelle is now the main active maintainer so he doesn't have to talk with say, ichard, about Black that often
fossil scaffold
Does anyone have an argument script for winning space-over-tabs by arguing tabs are anachronistic metal things on a teletype and have little place left in files (other than formats like tsv)?
lament crow
Does anyone have an argument script for winning space-over-tabs by arguing tabs ...
I don't because in an ideal world I think that tabs actually are actually better about a lot of things, it's just that this doesn't matter because it's more important to use the usual code style for your language of choice. And as said, tabs are better in an ideal world and we don't get to live in one. For example, some of the advantages of tabs can also be disadvantages - the ability for every developer to configure tab width as they wish is a very good thing but the problem with that is that this means that different programs may render tabs differently (on single user's computer) and they don't all allow you to configure them to render things differently. For example, it's popular for some software to assume that one tab is 8 spaces, i.e. tabs in terminals were always typically 8 spaces and some of the command-line applications depend on it which makes it hard to set them to render as 4 spaces even if your terminal applications gives you the choice. Spaces don't have such problems since 1 space always has a width of a single character.
tabs are better in a vacuum, spaces are better in practice is basically my point
fossil scaffold
Its true that if tabs rendered the same in the terminal, an editor or ide etc they'd be a non-issue and a byte-savings. But they were intended to move the print head to the next bit of movable metal tabstop, so they're intended to be configurable.
lament crow
if they actually were configurable then that would be fine
my point is that they aren't
not everywhere
dense jungle
I think that's enough closed issues for today
bright glacier
That explains why my email is blowing up (yeah I'm still watching the repository).
that's one more month where the issue count went down!
dense jungle
that's kind of depressing, I only made up for one month of growth?
elder tusk
every now and then i close like a dozen mypy issues. but mypy’s net issue growth is still like 30/month
bright glacier
that's kind of depressing, I only made up for one month of growth?
sounds like I should join in for an evening of issue tracker spring cleaning :)
anyway, you also dealt with this month's growth too so you've closed more issues than it may appear on that chart
cedar bluff
Hello! Our team have the diff black formatting result on each developer env. How we can debug it? And consider what Black rules we should add to pyproject.toml for got the identical result
dense jungle
Hello! Our team have the diff black formatting result on each developer env. How...
you likely have different people running different versions
consider setting required_version in pyproject.toml to standardize on a version
cedar bluff
you likely have different people running different versions
we use the latest, also I got different result format at ubuntu host and in docker conteiner
dense jungle
if you really have different formatting with the same version, that's a bug and you should report it. But my experience is that when people report this, they're almost certainly not running the version they think they are running. Setting required_version = "23.3.0" in your pyproject.toml should help ensure that, at least for versions new enough to support that option
cedar bluff
if you really have different formatting with the same version, that's a bug and ...
Will try in minutes and return with result, looks like you are right
flat krakenBOT
cedar bluff
if you really have different formatting with the same version, that's a bug and ...
Thansk @dense jungle a lot. Your recommendation works perfect.
Where I can got all list of available settings for pyproject.toml ?
I can't find it in doc. ESLint has a perfect description for each rule.
dense jungle
I have a PR to expand that list but it had some weird sphinx errors and I haven't gotten around to finishing it 😦
dense jungle
https://github.com/psf/black/pull/2839/, let me try to get it ready this time
dense jungle
finally got CI green on that, please review!
late dewBOT
:incoming_envelope: :ok_hand: applied timeout to @rugged ridge until <t:1683139543:f> (10 minutes) (reason: duplicates spam - sent 4 duplicate messages).
The <@&831776746206265384> have been alerted for review.
flat krakenBOT
brave orchid
Any low hanging issues I could help fix over the weekend?
flat krakenBOT
flat krakenBOT
flat krakenBOT
placid heart
why in thonny black format is not working?
clever peak
when black converts py print(weighted_randint(1, 10, [1, 1, 1, 1, 4,5,6, 1, 1, 1])) to ```py
print(
weighted_randint(
1,
10,
[
1,
1,
1,
1,
4,
5,
6,
1,
1,
1,
1,
],
)
)
fallow lance
Hi friends! Question: my reading of https://black.readthedocs.io/en/stable/the_black_code_style/current_style.html#how-black-wraps-lines
Black ignores previous formatting and applies uniform horizontal and vertical whitespace to your code.
// The emphasis on "ignores" is mine.
Is that I should be able to do the following tickery:
- Run black with my default
line-length = 99. This is a no-op. - Run black with
line-length = 30. This will change a bunch of lines. - Run black with
line-length = 99again. That should revert all the changes.
This is NOT the behavior I'm observing. (3) doesn't revert anything.
Any thoughts on this?
The bigger context is that I'm auto-adding a bunch of pyright diagnostic supressions into a codebase with black formatting and trying to make this 2 tools work together.
dense jungle
Hi friends! Question: my reading of https://black.readthedocs.io/en/stable/the_b...
Yes, this is unfortunately not 100% true. The biggest exception is the "magic trailing comma", where we keep lists that end in a trailing comma exploded on multiple lines.
If you run black -C, you may get closer to your original formatting
fallow lance
Yes, this is unfortunately not 100% true. The biggest exception is the "magic tr...
Oh gotcha, makes sense. What is -C ?
fallow lance
🙏 thank you so much Jelle, you are always helpful
@dense jungle is there a similar option by any chance for the () adding?
I have this diffs now all around
- run, _, _, expected_side_artifacts = run_with_artifacts
+ (run, _, _, expected_side_artifacts) = run_with_artifacts
no, unfortunately
ideally we should remove parens in cases like that when we don't need them, but it's a slow process
fallow lance
yeah this pyright + black interactions could be better
Thinking about it, it's actually not possible to improve it I think
You will always have the problem that you don't know which line you need to attribute the supression after the re-formatting
fallow lance
@dense jungle one thing I noticed is that -C doesn't respect the [] type comment syntax. I.e. it was not able to undo this
dense jungle
<@783088578363523104> one thing I noticed is that `-C` doesn't respect the `[]` ...
we might have a bug for that already, if not please report it 🙂
late dewBOT
@hollow valley :white_check_mark: Your 3.11 eval job has completed with return code 0.
001 | 0 0 RESUME 0
002 |
003 | 1 2 LOAD_NAME 0 (x)
004 | 4 POP_JUMP_FORWARD_IF_NONE 4 (to 14)
005 | 6 LOAD_NAME 0 (x)
006 | 8 STORE_NAME 1 (y)
007 | 10 LOAD_CONST 0 (None)
008 | 12 RETURN_VALUE
009 | >> 14 LOAD_CONST 0 (None)
010 | 16 RETURN_VALUE
011 | None
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/okuqalinis.txt?noredirect
@hollow valley :white_check_mark: Your 3.11 eval job has completed with return code 0.
001 | 0 0 RESUME 0
002 |
003 | 1 2 LOAD_NAME 0 (x)
004 | 4 POP_JUMP_FORWARD_IF_FALSE 4 (to 14)
005 | 6 LOAD_NAME 0 (x)
006 | 8 STORE_NAME 1 (y)
007 | 10 LOAD_CONST 0 (None)
008 | 12 RETURN_VALUE
009 | >> 14 LOAD_CONST 0 (None)
010 | 16 RETURN_VALUE
011 | None
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/unosodedik.txt?noredirect
hollow valley
quaint sleet
Is blib2to3 considered stable?
I’m working on adding structured pattern matching and parenthesized-context-manager support and as part of that effort have proposed forking lib2to3 with some features of bllib2to3 backported to it.
There are questions around just using blib2to3 instead.
See https://github.com/google/yapf/issues/1067 for more info.
dense jungle
Is blib2to3 considered stable?
I’m working on adding structured pattern matching...
it's not a public UI, and we may remove it if we come up with some other parser
I wouldn't be averse to making it into a separate installable project though, if there's interest
quaint sleet
@dense jungle thank you for the update. I’ve added your comments to the referenced PR.
dense jungle
I wouldn't be averse to making it into a separate installable project though, if...
though thinking more about this, it may cause complexities with packaging/mypyc compilation, so not sure if it would be worth the effort for us
flat krakenBOT
flat krakenBOT
fallow lance
ok so another edge case of pyright+black interactions https://black.vercel.app/?version=stable&state=_Td6WFoAAATm1rRGAgAhARYAAAB0L-Wj4AEvAMpdAD2IimZxl1N_WmufDs0LpH-m0u1CYeUiV-mNdcgl9W4yVi-Y_0m71dk3gORdBorNlMpLUC2Q8sVZg-xGpsLzLZpzNyG2KSPeLQvxK9dMpScAOTol1exTdsQ9Xq0zWFnQEA8r_LsvFhxTkkAByrR1yAMeeoxRUjwZ1Q3Pa4ZpzKKZFUb2pwsIUdKXMcuadKspposliL5QS6Kv9zv-XHyH-7shHi3-82uorWknEnXxRFC-GmGItgxbn1Z--qDS1bHaX943n9ikSzmGczAAAACMnFawvW14VgAB5gGwAgAABp_eILHEZ_sCAAAAAARZWg==
with patch(
"xxx"
) as get_firmware_function:
# return_value is unknown param, whatever
get_firmware_function.return_value = "/dev/null" # pyright: ignore[reportGeneralTypeIssues]
This supression is getting messed up by black
flat krakenBOT
flat krakenBOT
tawny heron
Is it possible to override default max line length just for a module?
flat krakenBOT
dense jungle
haven't gotten one of those in a while
slow ibex
no
I think you can format all except one module with one line length and then format only one module with other line length
dense jungle
I think you can format all except one module with one line length and then forma...
true. but you can't do it with one invocation. We got a PR for that but I think we decided against it
pearl spruce
Would there be interest in an alternative to blackd that doesn't require modifying black at all, uses a generic protocol and client, and can be used for other CLI tools as well?
dense jungle
Would there be interest in an alternative to blackd that doesn't require modifyi...
yes
pearl spruce
Cool! I've been working on something like this for a while, and have recently released it. It works perfectly with black :).
In terms of maturity, it's still early stage, but I already have it integrated into my git pre-commit hooks and IDE file watchers and use it all of the time.
JumpTheGun: https://github.com/taleinat/jumpthegun
I'm super-curious to hear what you think about it @dense jungle !
flat krakenBOT
flat krakenBOT
sweet tendon
What is black formatter?
silent apex
black is an automatic code formatter for python
you run it on a python file, and it will convert everything to a standard style
see https://black.readthedocs.io for more info
ivory quest
while it's not an official formatter, it became sort of a standard in the community
ornate hedge
you run it on a python file, and it will convert everything to a standard style
What about Pycharm auto styling?
silk dust
Failed to run Black. Whenever I try to run Black from within vscode (ctrl-shift-P > Black: Run Black), I get this error. I can run it from the command line, so I am pretty sure it isn't a Black issue per se, but this is the only plugin where I have problems like this. Hoping some kind person can show me what I am doing wrong. Thanks in advance!
ivory quest
Failed to run Black. Whenever I try to run Black from within vscode (ctrl-shift-...
what plugin do you have in mind? AFAIK black doesn't offer vscode plugin
rancid abyss
is the accidential newline convention change in https://github.com/pytest-dev/pytest/pull/11036 something black ought to undo?
dense jungle
is the accidential newline convention change in https://github.com/pytest-dev/py...
I think we're supposed to preserve existing newlines
So if someone changes the whole file, that will be preserved
rancid abyss
Thanks for the clarification
hexed furnace
AWith current latest black, 23.3.0 (compiled: yes):
import black
code = """
if True:
# fmt: off
pass
# fmt: on
pass
"""
black.format_str(code, mode=black.FileMode())
Traceback (most recent call last):
File "repro.py", line 10, in
black.format_str(code, mode=black.FileMode())
File "src/black/__init__.py", line 1084, in format_str
File "src/black/__init__.py", line 1089, in _format_str_once
File "src/black/parsing.py", ...
pulsar kayak
!d round
late dewBOT
pulsar kayak
!e
round(1/1024, 6)
@pulsar kayak :warning: Your 3.11 eval job has completed with return code 0.
[No output]
!e
print(round(1/1024, 6))
@pulsar kayak :white_check_mark: Your 3.11 eval job has completed with return code 0.
0.000977
zenith jacinth
!d round
late dewBOT
zenith jacinth
oops
flat krakenBOT
flat krakenBOT
flat krakenBOT
flat krakenBOT
solid adder
@austere lava libcst's native parser is default now? Noice. Congratulations on 1.0.0
austere lava
Yup! Thanks
solid adder
@austere lava is the parser recoverable?
austere lava
No unfortunately
solid adder
any chance it will or no?
austere lava
We have a project called errpy that's an error recoverable parser based on tree sitter
It's currently powering pyre and we'll migrate libcst to it too
That will take a while though
solid adder
Ocaml, eh?
austere lava
Rust will be the lowest common denominator and there will be python/ocaml/rust apis
solid adder
I'm asking because there's drama a-brewin' because Python 3.12 changes its tokenizer to raise SyntaxErrors
and it turns out IPython was using the ability for the tokenizer to gulp invalid code to support some of their magic
austere lava
Aha
solid adder
and now they are upset as they would have to migrate to something and there doesn't seem to be anything to migrate to
so you could come in on a white horse and save that shit
austere lava
Do you have a link to the drama?
Sidenote: it's very hard to make a peg parser error recoverable while also keeping it usable
solid adder
austere lava
Oh of course the blue bird of toxicity
solid adder
Mastodon isn't better. Like soylent green, it's people!
austere lava
I find that CPython's implementation of these things have conflicting goals with dev tools (performance vs UX)
replied on the thread, let's see what they say
solid adder
by the way
Black will be now met with new grammar shenanigans
f"hey {"zsol"}!"
this will be fun for Batuhan to work around, eh? if even possible
dense jungle
this will be fun for Batuhan to work around, eh? if even possible
yes, I already implemented and merged support for PEP 695 in blib2to3, but I'm not touching this f-string stuff
austere lava
Understandable
flat krakenBOT
pastel breach
black fails for me when run via precommit in pypy3.8 ci
Traceback: https://paste.pythondiscord.com/kibejapasa
flat krakenBOT
flat krakenBOT
flat krakenBOT
flat krakenBOT
dense jungle
this one. black.
flat krakenBOT
elder tusk
this one. black.
formatters should make your code easier to read. black is a little shy about preserving the AST... i find "rm" does a really good job of making my code more readable
vocal escarp
In this PyCon talk, Lukasz said that black uses asyncio to parallelize formatting: https://youtu.be/oVhJAVS3g00 . But the talk is mostly about ProcessPoolExecutor and shared memory. I didn't think asyncio would help with compute-intensive work.
elder tusk
it doesn't; it's purely used to orchestrate things: https://github.com/psf/black/blob/b60b85b234d6a575f636d0a125478115f993c90c/src/black/concurrency.py#L119 is literally the only async function in black proper
late dewBOT
src/black/concurrency.py line 119
async def schedule_formatting(```the ProcessPoolExecutor is started a couple lines above that
vocal escarp
the `ProcessPoolExecutor` is started a couple lines above that
thanks. that's similar to what Łukasz showed in the talk.
bright glacier
frankly I don't understand why we use asyncio.
Is it used as a better abstraction over using multiprocessing directly?
flat krakenBOT
zenith plover
whats this??
silent apex
black is an automatic code formatter for python code
check out the link in the channel description for more info
flat krakenBOT
flat krakenBOT
lament crow
Does mypy have its own codec?
dense jungle
Does mypy have its own codec?
no
lament crow
because it would change the AST
i.e. how the code behaves
there are other tools that can reorder imports already
ivory quest
ruff
lament crow
I would have said the same, yeah
covert ridge
Hi guys, girls, people
This is presumably a really common request so hoping someones written someway around it
As a developer I want to be able to postpone triggering the file watcher if there is another change in certain short period of time ("quiet period") - i.e. a developer touches one line in the scope of watched files, saves the file and a second later she makes another change, thus triggering another run.
What steps will reproduce the problem?
Enable file watchers.
Make a change in your files, triggering the file watchers to start the build
Realize you wanted to make another tiny change in some other file.
Waiting for the previous build to finish takes lot of time
What is the expected result?
When the developer makes the first change a sort of "quiet period" starts (e.g. 10 seconds) at the moment waiting out any other potential change in watched files the developer might still want to make. If she does not make another change during the quiet period, file watchers start the build. Any change during this quiet period prolongs the quiet period (resets the timer).
Ideally the length of the quiet period would be configurable.
What happens instead?
currently there is not an option to set the quiet period
Also not only does it avoid waiting, but sometimes im halfway through writing a line and it reformats and moves my cursor which is infuritating
really good tool though
dense jungle
Hi guys, girls, people
This is presumably a really common request so hoping som...
this sounds like a bug with some IDE, not black itself?
rare niche
Hi all, I have a question. I write some Python code in markdown document.
Just like this:
class Solution:
def pivotInteger(self, n: int) -> int:
for x in range(1,1000):
if (1 + x)* x= (x + n) * (n - x + 1):
return x
return -1
How can I automatic format the code with black?
fallow lance
I'd probably use https://black.vercel.app/ to copy-paste the snippets back and forth.
rare niche
I'd probably use https://black.vercel.app/ to copy-paste the snippets back and f...
Thanks for your sharing. I think it would be better to format the python code in markdown document automatically.
https://github.com/doocs/leetcode/blob/main/run_format.py#L62-L84 Here is my current solution. And I am looking for a better way 🙂
elder tusk
you could try something like https://github.com/hukkin/mdformat-black
you could also use black.format_str(...), see https://github.com/psf/black/issues/779
young sand
https://github.com/adamchainz/blacken-docs will also format Python code blocks in markdown files
fallow lance
Oh nice!
rare niche
That's great! Thank you all, guys.
naive wolf
Hello folks. I am curious that if there are some issues when I use Black with pre-commit to configure options via pyproject.toml file.
I use Python 3.11, pre-commit 3.3.1, and Black 23.3.0. I run the below codes and command on MacOS Ventura 13.4.
I install packages in .venv directory created by poetry.toml file.
The below one is my pyproject.toml file.
[tool.black]
line-length = 120
target-version = ["py311"]
The below one is my .pre-commit-config.yaml file.
- repo: https://github.com/psf/black
rev: 23.3.0
hooks:
- id: black
I run pre-commit command with poetry because I manage my packages with it. Thus, I create a command to run lint with Makefile file like the one below.
.PHONY: lint
lint:
poetry run pre-commit install && poetry run pre-commit run
The pyrpoject.toml , .pre-commit-config.yaml , and Makefile exist in the same depth of directory. It looks like the one below.
├── .venv
├── Makefile
├── poetry.lock
├── poetry.toml
├── pyproject.toml
├── .pre-commit-config.yaml
└── src
When I run make lint command, Black fixes line length even it is not over 120. Can I get any help?
flat krakenBOT
graceful trout
slick move right there with the zero. I too believe zero should always be put on a new line. It's too special not to be.. 🤦♀️
ivory quest
slick move right there with the zero. I too believe zero should always be put on...
you can always increase or decrease line length setting
graceful trout
you can always increase or decrease line length setting
that wouldn't really help much tho. The default length is ok, mostly. But it'd be better if there was some checking of the length of elements. As far as I remember, a list gets expanded if it doesn't fit within the max line length, but.. cmon, 1 element with a total of 3 extra characters? Couldn't it like, check the amount of elements in the lists, and not do that if you're barely over the limit and resulting in 2 extra lines with 1-2 characters only? That doesn't seem like a difficult "fix", and would look a lot better in.. pretty much any case I can think of. Basically, to have a range as the limit, with a strict and soft limit, not just a strict limit..
And even if for some reason doing that is impossible.. why couldn't it just do this instead.. looks a lot better 💀
little parrot
you can do that yourself if you add a trailing comma right?
graceful trout
you can do that yourself if you add a trailing comma right?
I'm pretty sure blackie chooses for u, whether u have a trailing comma or not, and that depends on the length, and no, I don't believe you can do what I'm asking for. At least I don't know how to
graceful trout
And even if for some reason doing that is impossible.. why couldn't it just do t...
this immediately gets formatted to the above version
little parrot
can you paste the code?
graceful trout
can you paste the code?
best to @ me, chances are I won't see the message otherwise:
test_data["Fare"].fillna(
train_data.groupby(["Pclass", "Embarked", "Parch", "SibSp"]).Fare.mean()[3]["S"][0][
0
],
inplace=True,
)
lol ok yeah doesnt matter
flat krakenBOT
elder tusk
hmm
so yesterday i switched pre-commit at work over to using mypyc black because i wanted things to be faster https://github.com/psf/black/issues/3405#issuecomment-1600079594
today i discover that now black is hanging very often
py-spy tells me:
__enter__ (multiprocessing/synchronize.py:95)
get (multiprocessing/queues.py:102)
_process_worker (concurrent/futures/process.py:237)
run (multiprocessing/process.py:108)
_bootstrap (multiprocessing/process.py:315)
_launch (multiprocessing/popen_fork.py:71)
__init__ (multiprocessing/popen_fork.py:19)
_Popen (multiprocessing/context.py:277)
start (multiprocessing/process.py:121)
_adjust_process_count (concurrent/futures/process.py:675)
submit (concurrent/futures/process.py:697)
run_in_executor (asyncio/base_events.py:814)
<dictcomp> (black/concurrency.py:163)
schedule_formatting (black/concurrency.py:161)
_run (asyncio/events.py:80)
_run_once (asyncio/base_events.py:1890)
run_forever (asyncio/base_events.py:596)
run_until_complete (asyncio/base_events.py:629)
reformat_many (black/concurrency.py:108)
new_func (click/decorators.py:26)
invoke (click/core.py:760)
invoke (click/core.py:1404)
main (click/core.py:1055)
__call__ (click/core.py:1130)
<module> (black:8)
hmm but there are way more processes than i expect, another classic multiprocessing.cpucount() thing?
oh this sounds quite a bit like https://github.com/psf/black/issues/2513
okay i'll patch that in and see if it helps. weird that it's only happening with mypyc...
oh but maybe the actual difference is some ci caching changes that work made
anyway, could be nice for --workers to be specifiable by env var as well. is that a patch you'd be willing to take?
dense jungle
anyway, could be nice for `--workers` to be specifiable by env var as well. is t...
kind of wary of that as we end up with three different ways of configuring Black (command line, config file, env vars), and it will be confusing to reason about how they interact. Seems like we already accept BLACK_CACHE_DIR though
why should --workers be special and deserve an env var?
elder tusk
it's not critical by any means, but feels weird to change our incantations to specify --workers=fixed_number when the fixed number only matters in CI. the assumption that local machines and CI have same number of cores doesn't really hold
dense jungle
it's not critical by any means, but feels weird to change our incantations to sp...
ok that makes sense, you'd want to use the same pre-commit invocation both locally and in CI, but you reasonably want a different value for --workers in both cases
an env var is a good fit in that case
flat krakenBOT
Describe the bug
We started to use black and used it to reformat the whole repository. One week later, on the same black version, many lines have been changed back to what they were previously.
Merge request 1, introducing black:
Merge request 2, one week later:
I don't mind which vers...
flat krakenBOT
flat krakenBOT
onyx current
Intresting
flat krakenBOT
flat krakenBOT
fallow lance
Does black have a special case for # type: ignore comments? After switching to pyright and # pyright: ignore comments people noticed that black is much more liberal with re-shuffling the comments which leads to frustrating cycle of fighting with it to keep the comments on the lines they intended to be on. Curious why we didn't notice it before.
dense jungle
Does black have a special case for `# type: ignore` comments? After switching to...
yes. there's an open issue about extending the special treatment to pyright comments
fallow lance
yes. there's an open issue about extending the special treatment to pyright comm...
where can I upvote?
I think I found it https://github.com/psf/black/issues/3399
Actually this seems different...
dense jungle
where can I upvote?
I guess there's just a PR https://github.com/psf/black/pull/3661
stiff spear
Hi.
A question about how black uses pyproject.toml. We have a big repo with some nested pyproject.toml's. In this case black will pick the closest pyproject.toml file based on current directory while we would preferably like to specifiy our black config only in the root pyproject.toml file.
In contrast, ruff does exactly this and it works pretty weel for us https://github.com/astral-sh/ruff/blob/main/crates/ruff/src/settings/pyproject.rs#L48 It will check that there is actually a tools.ruff section before deciding to use a pyproject.toml file.
If this is something that seems reasonable I am willing to create a PR for it
elder tusk
Hi.
A question about how black uses pyproject.toml. We have a big repo with som...
this is something i've wanted too, if no one else chimes in definitely make an issue / PR for it!
dense jungle
Hi.
A question about how black uses pyproject.toml. We have a big repo with som...
makes sense to me!
rancid abyss
anyone aware of a pseudo-repr function thats able to render strings or tuples in a black safe manner - i'd like the setuptools_scm python version file template to yield black safe code, without having to sort out details like line length or quoting myself - my impression is that the alternative would need to have black format it and i'd like ot avoid that
slow ibex
Wdym by "black safe"?
String that is already formatted in black-like style (so applying black to it does nothing)?
flat krakenBOT
flat krakenBOT
flat krakenBOT
late dewBOT
:incoming_envelope: :ok_hand: applied timeout to @pliant fossil until <t:1688324939:f> (10 minutes) (reason: duplicates spam - sent 4 duplicate messages).
The <@&831776746206265384> have been alerted for review.
flat krakenBOT
flat krakenBOT
bright glacier
@elder tusk you've beat two maintainers in issue closures in like a week 😄
next target: Cooper
flat krakenBOT