it would be great to get diff-shades set up in CI, similar to how mypy-primer works on typeshed
#black-formatter
dense jungle
bright glacier
I'm not even done the feature that powers those reports -- I just whipped up something quick and dirty in my site-packages install of diff-shades ๐
Yeah my development practices might not be the best when it comes to personal projects :p
plain atlas
i- i see
bright glacier
How helpful are the reports I'm attaching by the way @dense jungle ? Is there something you'd like as a fellow maintainer? If I can get the data backend performant enough I plan on adding line count, line additions, line deletions, and files changed statistics.
Anything from mypy-primer you'd like is also welcome ๐
dense jungle
How helpful are the reports I'm attaching by the way <@!783088578363523104> ? Is...
they're great! that's basically exactly what I'd want
it's unfortunate that the conditional expressions report included the changes from the blank lines PR, I guess that's addressable with a better baseline
bright glacier
Yeah that's just me being lazy as each run take 20 minutes to do. I first compile black with mypyc (this ends up being a time saver in the long run!) and then run diff-shades taking 15 minutes.
plain atlas
so obviously a workflow would be great
bright glacier
Yea but I'm a busy human
bright glacier
16 cores?
plain atlas
the runners have at least that many
dense jungle
except for the ones used for the rest of CI
plain atlas
wdym?
bright glacier
because also it would only take 2 minutes because 16 cores
diff-shades is configured with slightly more projects than black-primer which takes 8-12 minutes in CI - I don't understand what you mean each VM has 16 cores.
the extra projects are just my own :p
Linux and Windows VMs have 2 cores while the Mac ones have 3.
tired shard
uhh
bright glacier
is this black from main?
dense jungle
uhh
you likely have a different syntax error elsewhere. because of that, black tries the 2.7 grammar which explodes on the print keyword
tired shard
nope
bright glacier
jelle's guess seems likely then
plain atlas
flake8 will tell you where your syntax error actually is
bright glacier
plain atlas
<@!717983911824588862> https://docs.github.com/en/actions/using-github-hosted-ru...
well then
tired shard
you likely have a different syntax error elsewhere. because of that, black tries...
the code works perfectly fine and no errors from pyright
dense jungle
the code works perfectly fine and no errors from pyright
can you hack parsing.py so it prints the errors for all grammars instead of just the last one
or try --target-version py310 or something
bright glacier
Oh do you have pattern matching / parenthesized with?
tired shard
yep
bright glacier
Yeah that's locked under --target-version py310 at the moment.
bright glacier
Read the changelog :p
flat krakenBOT
raw comet
black can do 3.10 now?
flat krakenBOT
dense jungle
black can do 3.10 now?
yes, but you need to pass --target-version py310 explicitly for match syntax to work
flat krakenBOT
bright glacier
actually I do have an analysis from the blank line PR that I could use as the baseline for PR #2278's report - I just totally forgot about it ๐ฆ
dense jungle
let's make a new release soon
plain atlas
yay 21.12b1
dense jungle
I'd like to implement somebody's suggestion that we print a warning suggesting to use -t py310 if we detect match syntax
red sedge
"it looks like you're playing with matches, would you like some help" ๐ฅ๐
bright glacier
This feels much better already @plain atlas :p
bright glacier
honestly nothing, I'm just playing with nox instead of tox
plain atlas
ohhhhh
my pc is lagging rn so if i'm slow to respond its because my pc is
tldr a windows service has a memory leak
okay killed
plain atlas
honestly nothing, I'm just playing with nox instead of tox
yeah, I haven't looked at nox much
I'm a bit excited to try it, however
given that its fairly nice, and allows the user to test on multiple environments locally
drowsy sonnet
Okay, so... I can't figure out if I'm seeing something that is new -- because I can't find a corresponding issue for this.
Black formats this file as:
"""This is a subpackage because the directory is on sys.path for _in_process.py.
The subpackage should stay as empty as possible to avoid shadowing modules that
the backend might import.
"""
from contextlib import contextmanager
from os.path import abspath, dirname
from os.path import join as pjoin
try:
import importlib.resources as resources
def _in_proc_script_path():
return resources.path(__package__, "_in_process.py")
except ImportError:
@contextmanager
def _in_proc_script_path():
yield pjoin(dirname(abspath(__file__)), "_in_process.py")
All this additional whitespace seems excessive, to the point of hurting readability.
errant barn
For me there's only one space before except on the playground. Would you suggest that the space before contextmanager is removed?
drowsy sonnet
What version of black are you using? I'm seeing this on the latest stable release.
Ah, main has this fixed somehow.
toxic stormBOT
errant barn
But what do you think of the spacing otherwise?
drowsy sonnet
I think what main does is reasonable!
errant barn
Agreed, although I probably would get rid of the space before the function.
But that seems to be discussed in psf/black#450
toxic stormBOT
flat krakenBOT
jolly oriole
Maybe it's counting that as an operation ยฏ\_(ใ)_/ยฏ
flat krakenBOT
plain atlas
@muted haven i was just thinking that you would submit a fix, lol
muted haven
flat krakenBOT
plain atlas
I left a review ^^
neither comment of which actually applied to the code change itself ๐
muted haven
thanks though I don't necessarily know how to respond to this
https://github.com/psf/black/pull/2667#discussion_r761979694
I just add patma related changes end-to-end I guess, not sure if there is an order to follow ๐
plain atlas
yep Richard got you covered
tldr no one cares 
which makes sense, by putting it anywhere in that block, it can slightly reduce merge conflicts
dense jungle
I might reorganize the changelog a bit before the release
bright glacier
Fixed in https://github.com/psf/black/commit/b0c2bcc9537d238c8a580294ecbc41de465...
we'll be doing a release soon so just stay patient with us :)
@plain atlas what do you think of using this?
late dewBOT
.github/workflows/ci.yml lines 14 to 16
concurrency:
group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.sha }}
cancel-in-progress: true```.github/workflows/lint_test.yml lines 15 to 17
concurrency:
group: ${{ github.workflow }}-${{ github.repository }}-${{ github.ref }}
cancel-in-progress: true```basically the idea is that if we rapidly update a PR, the older CI runs would be autocancelled
plain atlas
!remind 10M double check this
late dewBOT
late dewBOT
!remind 10M double check this
It has arrived!
plain atlas
basically the idea is that if we rapidly update a PR, the older CI runs would be...
Thinking about it, it should work, although I would consider using github.ref instead of sha. The sha means it will always run on every commit, and continue to run, unless there is a pull request.
However, that may be desired behavior, as using a mix would mean that it would cancel checks if two consecutive pulls were merged to main.
I suppose my point above is be careful which workflows you put it on ๐
Additionally, I would consider github.sha vs github.ref, I have made a context dump here: https://github.com/onerandomusername/context/actions/runs/1536895896
bright glacier
Thank you I'll look at it later as I'm helping out in the help channels ATM
bright glacier
I honestly have no idea
plain atlas
!src remind
late dewBOT
late dewBOT
config-default.yml lines 261 to 264
reminder_whitelist:
- *BOT_CMD
- *DEV_CONTRIB
- *BLACK_FORMATTER```how do we want to detect when the line the parser fails on is probably a match statement? .startswith("match") but that's a bit imprecise :)
dense jungle
how do we want to detect when the line the parser fails on is probably a match s...
oh I was going to just try parsing the file with the match grammar
bright glacier
oh that works too
I'll leave the PR for you to do then as I rather not implement that change - not for technical reasons, I just don't feel like it.
dense jungle
sure, writing the code now
bright glacier
You're so fast compared to me it surprises me every time ๐
dense jungle
just procrastinating from doing real work
bright glacier
I finished most of my work today, I just have some notes to fill out and I should be good to go for today.
dense jungle
bright glacier
that period after match ... looks weird but I don't know whether that's pre-existing or not. I'm not sure whether the backticks are necessary - but they do look weird imo as we don't use backticks as part of our UI language so far.
These are just nitpicks, the overall idea seems fine to me.
dense jungle
thanks, I'll take out the backticks
I put the period in, feel like we need something to separate the code from the message. Maybe I should put in a newline instead
bright glacier
Oh yeah a newline would work better IMO. I thought you had used a newline but it turns out my viewport is just the right size to make it seem like that :p
dense jungle
also apparently we support parenthesized with before 3.10 too? ```(black) jelle@mbpt-root black % black -t py38 -c 'with (x as y, y as z): pass'
with (x as y, y as z):
pass
Yup, we only use a single grammar and the support for parenthesized with was implemented in the grammar itself. On the other hand pattern matching requires adding some names to the grammar object at runtime (on top of the grammar changes) so the parser knows whether to treat match / case a soft keyword
dense jungle
oh interesting, I thought parenthesized with also required backtracking/PEG
flat krakenBOT
bright glacier
PY310_HINT: Final[
str
] = "Consider using --target-version py310 to parse Python 3.10 code."
I guess this is what the 25+ issues about splitting on the RHS vs LHS is all about because damn this is not pretty.
dense jungle
agree
plain atlas
ditto agreeee
bright glacier
Pretty happy with this, now time to implement the rest of the show and compare command ๐
plain atlas
yeah from what you showed me it looks great
muted haven
I recall there was a benchmark suite for black @bright glacier, would it be possible to run it with -tpy310 to get some profiling data?
I plan to do some improvements over performance today, but it would be nice to learn how slow it is already ๐
bright glacier
What's the status on the TOML bare CR situation btw?
bright glacier
!remind 1D look into doing more outreach in regards to testing the mypyc wheels
late dewBOT
plain atlas
What's the status on the TOML bare CR situation btw?
not part of the scope, the issue on toml-lang was closed
the issue on pypa/pipis still open
bright glacier
plain atlas
you've probably already checked them, since it's been 2 hours
bright glacier
It's still open unless you mean a different one.
red sedge
I think you just pass bytes to tomli and let it decide
bright glacier
Yeah I suppose
bright glacier
@dense jungle how you feel about running the 3.10 grammar check in a thread that has a timeout? https://github.com/psf/black/pull/2668#issuecomment-985982856
I feel like it won't be worth it especially there are files passed to black IRL that are massive
dense jungle
I guess it's a nice additional safety but not sure it's really worth it
bright glacier
I vote nay then as I don't want to deal with multithreading with mypyc or 3.6. It's on the error path anyway so I doubt it'll be a problem IRL.
plain atlas
for the longest time i thought lgtm meant 'looks good to merge'
and then i learned it meant 'looks good to me'
lament crow
you got 3 words out of 4, not bad :P
plain atlas
true
lament crow
I heard 'let's get this merged' abbreviation in the past too
plain atlas
makes sense
dense jungle
@bright glacier we could do another release now, or maybe wait for Batuhan's possible perf improvements?
then again, that could just go into the next one
bright glacier
OH yeah
I guess it doesn't hurt to wait a little bit.
This release candidate is quite packed with changes and it's only been 17 days
sour jay
lgtm
plain atlas
lol, do you have a source?
lament crow
lol, do you have a source?
Had to do a bit of searching but I think this is the original source:
https://twitter.com/jschauma/status/1019410471999467525?t=Nan8j9m00u7jKQRJO2ME6g&s=19
bright glacier
I recall there was a benchmark suite for black <@429445826251587616>, would it b...
It turns out 1) blackbench sucks - I had to hack the source code to support this usecase, and 2) the included benchmarks don't really have any usage of match surprisingly enough.
โฏ pyperf compare_to main-9424e795b-normal.bench.json main-9424e795b-py310.bench.json -G
Slower (1):
- parse-black/strings: 74.8 ms +- 7.3 ms -> 231 ms +- 32 ms: 3.08x slower
Benchmark hidden because not significant (16): parse-black/__init__, parse-black/brackets, parse-black/comments, parse-black/linegen, parse-black/lines, parse-black/mode, parse-black/nodes, parse-black/output, parse-comments, parse-dict-literal, parse-flit/install, parse-flit/sdist, parse-flit_core/config, parse-list-literal, parse-nested, parse-strings-list
Geometric mean: 1.07x slower
I think it's fine if they don't use match statements, if I understand correctly the worry was that it would slow down parsing other code that happens to use match as an identifier
is there a benchmark that calls re.match a lot?
also thanks for running this!
bright glacier
No, I meant literally any reference to match the word not the statement.
I grepped through all of the test files and only black/strings has any ...
https://paste.pythondiscord.com/lomazuhifi.py this is the only file that uses match as an identifier
โฏ blackbench dump black/strings | grep "match" --line-number
17: overlapping matches.
40: match = re.match(r"\s*\t+\s*(\S)", line)
41: if match:
42: first_non_whitespace_idx = match.start(1)
136: match = re.match(r"^([" + STRING_PREFIX_CHARS + r"]*)(.*)$", s, re.DOTALL)
137: assert match is not None, f"failed to match string {s!r}"
138: orig_prefix = match.group(1)
142: return f"{new_prefix}{match.group(2)}"
191: matches = re.findall(
200: for m in matches:
But yeah it does seem to slow down parsing a fair bit if match is used as an identifier
dense jungle
oh yeah ```...$ rm -rf ~/.cache/black/ && time black graphql_/schema/search/_definitions.py
All done! โจ ๐ฐ โจ
1 file left unchanged.
real 0m0.474s
user 0m0.440s
sys 0m0.028s
...$ rm -rf ~/.cache/black/ && time black -t py310 graphql_/schema/search/_definitions.py
All done! โจ ๐ฐ โจ
1 file left unchanged.
real 0m2.744s
user 0m2.692s
sys 0m0.048s
this file uses the variable name "match" like 60 times
bright glacier
I just hacked up a super simple micro benchmark and yeah it's not great
Mean +- std dev:
[main-9424e795b-normal.bench-mini] 20.6 ms +- 1.5 ms
[main-9424e795b-py310.bench-mini] 194 ms +- 25 ms
9.41x slower
whyy is the unexpected EOF test crashing ....
muted haven
I have no idea ๐ค will check again
bright glacier
oh, CI is failing on main too :(
somehow https://github.com/psf/black/commit/dc8cdda8fdd6941103240ae3279034d2acdc69bc broke the previous commit
bright glacier
I have no idea ๐ค will check again
right, driver.parse_string can raise ParseError, TokenError, or even IndentationError
now why did CI not catch on jelle's PR? well to avoid (practically) duplicated workflows, one for the push and one for the pr event, we cancel workflows on psf/black for PR event
this means CI doesn't test the revision when the PR would be merged into main
bright glacier
@muted haven this should fix the failing test, sorry for the noise!
flat krakenBOT
bright glacier
should we add a changelog entry for https://github.com/psf/black/pull/2670?
muted haven
I did not include one since I though this wouldn't be the only speed-up we are trying to get but I am not sure at this point
if I can make more that targets this release
bright glacier
It depends how confident you are as we want to release soon :)
plain atlas
I did not include one since I though this wouldn't be the only speed-up we are t...
you could include one in this pull, and then append to it in the next pr, if it hasn't been released yet
muted haven
I've looked into more holistic solutions (e.g determining token type on the tokenizer), but it seem very very hard to do so (primarily because of stuff like match ({a: b}) can be either a call to a function named match() or a real match statement). So it would probably require someone (possibly me, but who knows maybe someone else wants to do it) to take a look at my initial implementation for parso, which avoids stack copies at all (It pseduo-emulates the parser stack and DFA pushes).
I don't think I will have time during the week, so I'll try to give it a shot at the next weekend. No promises / expectations though ๐
muted haven
I just hacked up a super simple micro benchmark and yeah it's not great
```
Mean...
@bright glacier btw how did you this benchmark?
I tried to run the given file with black on an isolated environment, and I've only observed a degredation of 1.6X;
root@bac268808ff0:/# hyperfine 'touch t.py;black --check t.py'
Benchmark 1: touch t.py;black --check t.py
Time (mean ยฑ ฯ): 158.5 ms ยฑ 2.5 ms [User: 146.0 ms, System: 12.4 ms]
Range (min โฆ max): 155.7 ms โฆ 165.6 ms 18 runs
root@bac268808ff0:/# hyperfine 'touch t.py;black -tpy310 --check t.py'
Benchmark 1: touch t.py;black -tpy310 --check t.py
Time (mean ยฑ ฯ): 243.3 ms ยฑ 4.8 ms [User: 222.7 ms, System: 20.2 ms]
Range (min โฆ max): 239.2 ms โฆ 254.9 ms 12 runs
root@bac268808ff0:/# black --version
black, 21.11b2.dev26+gd9eee31 (compiled: no)
Yeah I suspect it's probably due to black's startup time that the slowdown is less noticable
muted haven
Possibly. So I assume you only benchmarked parse()?
bright glacier
import tempfile
from pathlib import Path
import pyperf
import black
from blib2to3 import pygram
# First try the relevant function post-refactor in 21.5b1, and fallback
# to old function location.
try:
from black.parsing import lib2to3_parse
except ImportError:
from black import lib2to3_parse
runner = pyperf.Runner()
code = Path(r"{target}").read_text(encoding="utf8")
with tempfile.TemporaryDirectory(prefix="blackbench-parsing-") as path:
# Block the parser code from using any pre-existing cache.
pygram.initialize(path)
runner.bench_func("{name}", lib2to3_parse, code, set((black.TargetVersion.PY310,)))
it's effectively this script
muted haven
good to know, thanks! (will definitely use this for the future PRs)
bright glacier
note that this script is a bit more complex because it has to handle old versions of black where the APIs changed
parse-black/strings
===================
Mean +- std dev: [uhh-default] 67.0 ms +- 2.0 ms -> [uhh-initial] 227 ms +- 27 ms: 3.39x slower
Mean +- std dev: [uhh-default] 67.0 ms +- 2.0 ms -> [uhh-speedup] 154 ms +- 3 ms: 2.30x slower
parse-match-identifiers
=======================
Mean +- std dev: [uhh-default] 21.8 ms +- 2.7 ms -> [uhh-initial] 196 ms +- 23 ms: 9.01x slower
Mean +- std dev: [uhh-default] 21.8 ms +- 2.7 ms -> [uhh-speedup] 135 ms +- 10 ms: 6.19x slower
Geometric mean
==============
uhh-initial: 5.53x slower
uhh-speedup: 3.77x slower
nice win @muted haven ๐
โฏ pyperf compare_to uhh-initial.json uhh-speedup.json -G
Faster (2):
- parse-black/strings: 227 ms +- 27 ms -> 154 ms +- 3 ms: 1.47x faster
- parse-match-identifiers: 196 ms +- 23 ms -> 135 ms +- 10 ms: 1.46x faster
Geometric mean: 1.47x faster
from a different perspective
muted haven
Btw do you ship blib2to3 mypy-c compiled? @bright glacier
bright glacier
Not via PyPI yet, but I'm trying to community test the compiled wheels (all of black is compiled, not just blib2to3)
I should probably rebuild the current published set of wheels as main has a whole new set of features at this point ๐
I'll probably get ลukasz or something to tweet this issue out or something because right now I have one successful report for Linux 3.8 which is me ๐
silent apex
ooh ooh ooh I can do this rn, brb
bright glacier
Worst come to worse I'll just ask the whole core team to help out as we do have all of the hardware necessary to do so
silent apex
so wait, is the command just pip install black --extra-index-url https://ichard26.github.io/black-mypyc-index/simple/?
bright glacier
theoretically although you're the second person to try as far as I know
you might need the upgrade flag
It's hard hosting on your PyPI
silent apex
installed fine for me, I think I'm good
I have black, 21.11b2.dev14+gb336b39 (compiled: yes)
sour jay
so wait, is the command just `pip install black --extra-index-url https://ichard...
Download the wheel and do pip install name_of_wheel.whl no?
silent apex
๐คทโโ๏ธ I just tried to follow the instructions on https://ichard26.github.io/black-mypyc-index/
bright glacier
Download the wheel and do `pip install name_of_wheel.whl` no?
well you assume people know how to choose a wheel from a list of entries lol
that's why I hosted my own PyPI style index with github pages :p
sour jay
Lol
bright glacier
installed fine for me, I think I'm good
good to know although please don't forget to open a field report as stated in that issue because my memory sucks L)
bright glacier
you don't have to do it now, just sometime after you feel confident in describing your experience with it
although the name of the game here is more about quantities of positive field reports because with enough people y'know someone's gonna manage to break your code
silent apex
err how do you clear black's cache? (I assume it does)
plain atlas
rm -r ~/.cache/black
silent apex
ah fun
bright glacier
silent apex
tyty
bright glacier
I honestly made myself a simple utility to do this because I run into the cache way too much while developing
silent apex
wowowowow
black-compiled took half the time of regular black in formatting python-discord/bot
impressive, great job on this richard
plain atlas
neat
bright glacier
black-compiled took half the time of regular black in formatting `python-discord...
lovely! and it didn't crash once I presume?
silent apex
not at all
bright glacier
phew
silent apex
smooth sailing
I have filed a field report
and I shall test on WSL for another later
bright glacier
thank you :)
plain atlas
now test by formatting the stdlib
plain atlas
I've formatted it before
followed directly by git checkout .
sigh how do i installed compiled black
silent apex
now test by formatting the stdlib
I'm doing it
bright glacier
can you wait a little bit
bright glacier
I have a new set of wheels
just gotta push them
dawn, you don't have to wait, but arl shall wait :)
plain atlas
just gotta pull cpython
plain atlas
no its pulled
silent apex
formatting stdlib now, let's see how it goes
bright glacier
what's your OS, Python version, and arch?
silent apex
fuck it broke
bright glacier
yeah that's expected
there's an issue on the tracker with various crashes for formatting cpython
silent apex
ah okie
plain atlas
what's your OS, Python version, and arch?
wsl, ubutnu, x64, and any of these versions
bright glacier
do you have a preference lol?
plain atlas
not really, no
bright glacier
pypy is not supported btw
plain atlas
ยฏ_(ใ)_/ยฏ
3.9.9 ig
or 3.10 because there's probably some random bug with 3.10 :kekw:
plain atlas
lmao
plain atlas
you can
plain atlas
i'll test in a moment, i'm working on my nuke discord bot
bright glacier
excuse me what ๐
plain atlas
wait-- that sounds bad, that's not what I mean
I've been testing slash commands and now i have a bunch of those that don't exist, so I need to clear them
bright glacier
d'ah
plain atlas
So I'm nuking my slash commands
the file is literally named nuke.py
okay done
they gone
bright glacier
or 3.10 because there's probably some random bug with 3.10 :kekw:
?
there
@dense jungle I prefer to do a release now with Batuhan's performance improvement merged so I can update the mypyc wheel index with a compiled release (instead of a development version for the time being). Batuhan said they'll need some time for additional performance wins.
Thoughts?
plain atlas
huh i got this error when formatting the stdlib
maximum recursion depth exceeded while calling a Python object
bright glacier
that's already known lol
plain atlas
lmfao was about to say, better not be expected
bright glacier
basically there's an assignment chain of 1000+ assignments
https://github.com/psf/black/issues/2407#issuecomment-906060619 <- I should probably update this with our new 3.10 support
plain atlas
late dewBOT
Lib/test/test_exceptions.py line 1758
# Mutating locals() is unreliable, so we need to do it by hand```lmfao
bright glacier
Yeah talk about unreasonable input
plain atlas
@bright glacier I ran it, do you want the good or bad news first
bright glacier
bad news first
plain atlas
i didn't run it with time
silent apex
basically there's an assignment chain of 1000+ assignments
...why
bright glacier
because the "did you mean" feature of NameError / AttributeError shouldn't slow it down too much, so if there's too many locals the feature is disabled
bright glacier
i didn't run it with `time`
did it feel faster?
plain atlas
idk i walked away from my pc
bright glacier
:>
plain atlas
oh come on!
bright glacier
Also I think we should consider the fact that we crash on that function with 800 locals to be a feature.
- Jelle
not totally unreasonable ๐
bright glacier
do the crashes look similar to these ones https://github.com/psf/black/issues/2407#issuecomment-906060619
if not then it sounds like I have a mypyc bug to deal with
plain atlas
!remind 40M back from thing i have to do, now do other thing and other thing, but not confused with thing a or b
late dewBOT
plain atlas
I'll update you when I'm done with thing cffc41bd
silent apex
do the crashes look similar to these ones https://github.com/psf/black/issues/24...
uh no I don't think so
silent apex
I'll get you them in text in ~20 minutes
plain atlas
`black .` in `cpython/lib` crashed in ~2 seconds for me
fyi I'm using 3.10 as the runner, and cpython main as the stdlib to format
plain atlas
with this wheel
plain atlas
https://github.com/ichard26/black-mypyc-wheels/actions/runs/1541832497
from this build
afk
silent apex
nvm, I got it now https://paste.pythondiscord.com/ubolabuqac.coffeescript
err, maybe it's cause I cloned master but am using 3.9.9 rn?
plain atlas
I cloned master and am using 3.10
but black is supposed to be able to
OH
@silent apex specific -t py310 on the cl
silent apex
oh frick
plain atlas
I didn't even look at your error lol
silent apex
nope still dies
little parrot
you need to be on 3.10
silent apex
I'll install 3.10 and try then
bright glacier
I'mma try a newer revision of mypy actually
plain atlas
you need to be on 3.10
black shouldn't be showing that error to user
bright glacier
there's a performance win which inlines most of the refcnt operations
silent apex
I'll install 3.10 and try then
Stuff is still breaking, and now I get that the process pool is terminated as well
bright glacier
what happens if you try a pypi version of black?
You're not running black within cpython/lib right? because then imports would use stuff from your checkout which might not go so well
silent apex
You're not running black within cpython/lib right? because then imports would us...
That is exactly what I was doing, woops
late dewBOT
!remind 40M back from thing i have to do, now do other thing and other thing, bu...
It has arrived!
dense jungle
<@!783088578363523104> I prefer to do a release now with Batuhan's performance i...
yeah let's do a release
do you want to do it?
bright glacier
Sure, playing gartic phone with the staff here atm though =p
bright glacier
what
plain atlas
oh nvm a simple mistake nvm nvm
dense jungle
I think you most likely have an empty black directory on your PYTHONPATH somewhere
so it interprets it as a namespace package
plain atlas
that's close enough to what happened, yeah
dense jungle
I guess not necessarily empty
bright glacier
yeah let's do a release
plz review psf/black#2670, thanks
toxic stormBOT
plain atlas
yeah, its black's repo lmao
bright glacier
I can do the release this time
dense jungle
thanks, sounds good!
also we've been running mypyc black at Quora since last week or so and everything has been smooth, I should submit a report
bright glacier
oh! ๐
bright glacier
Sorry for the delay by the way, it turns out I underestimated how much time I had left before I was up to play with staff oops.
bright glacier
@dense junglehttps://github.com/psf/black/releases/edit/untagged-9e05f026ee64102be3a9 looks good?
I'm totally down to remove that contributors part if you'd like
I saw the "auto-generate changelog" button and was curious to what it does ... then played around with its output :p
wait, do github releases now include this by default? cause my extra effort is useless then
Doesn't seem like it, probably due to me tagging them
plain atlas
<@&831776746206265384>
bright glacier
I filed an incident already but alright :)
plain atlas
half expected a scam link after that
when they didn't send, I was like, "fuck i should have dmed modmail"
plain atlas
I filed an incident already but alright :)
oh, you deleted it...?
bright glacier
I didn't
but the staff team have filed an incident for the mods to take a look at
plain atlas
ah
dense jungle
I saw the "auto-generate changelog" button and was curious to what it does ... t...
I like it, it's a nice touch
looks like I missed all the moderation fun
bright glacier
!remind 1D update docs for stable revision
late dewBOT
plain atlas
hmmm
plain atlas
!remind 1D update docs for stable revision
I want to make a workflow that will be triggered by someone manually running it to trigger a new stable docs build
this means that all maintainers can update the stable docs, just as they can update the tag
worth making, perhaps?
bright glacier
that seems unnecessarily complicated
also this shouldn't be necessary as RTD should theoretically pick up the updated stable tag but doesn't
plain atlas
its not, its a few lines, to use curl and a token to trigger a build
oh right i meant to actually make it trigger on the tag update
which wouldn't be complicated, either
not when a maintainer triggers it as that's already being trigger--
wait a sec brb
i-
plain atlas
also this shouldn't be necessary as RTD should theoretically pick up the updated...
i think the problem is on black's end
on this page, https://github.com/psf/black/settings/hooks, there should be a rtd webhook
bright glacier
no the webhook is being fired, it's just picking up the wrong revision somehow
plain atlas
you sure?
plain atlas
nice
dense jungle
nice, new wheels showed up at https://ichard26.github.io/black-mypyc-index/simple/black/ ๐
bright glacier
how did you notice?
bright glacier
Yeah that's because I have to trigger the build and upload the wheels manually.
bright glacier
flat krakenBOT
grave fox
Do you guys a GitHub action out to automatically run black on code in a commit or PR?
plain atlas
there's both an action and a pre--dang it jelle :D
grave fox
perfect
late dewBOT
!remind 1D update docs for stable revision
It has arrived!
bright glacier
Damn, compiled black really is faster - I just ran it against the pip project and it's probably twice as fast
that is stupid fast startup with compiled black ๐
.. and there's literally a faster startup PR for black (not blackd) so this could get faster (although it does some things I'm uneasy with)
bright glacier
pre-commit.ci might not like our local black hook fwiw for autoupdating
bright glacier
I'm going to try to pick up psf/black#2095 and rewrite it (while addressing feedback) and hopefully drive it to completion
toxic stormBOT
bright glacier
hmm, I don't like the existing infrastructure for running transformers - I mayy try to clean it up although I'm nervous I might break something
... is it defined what makes an exponent / base "complex"? https://github.com/psf/black/pull/2095#discussion_r619853433
dense jungle
... is it defined what makes an exponent / base "complex"? https://github.com/ps...
It might need discussion but I'd say everything but a bare name or literal is complex
And maybe unary operators on non-complex expressions are still non-complex
2**-2 vs. 2 ** -2
though https://github.com/psf/black/issues/538#issuecomment-962448644 wants no spacing for more complex operators
plain atlas
@bright glacier could you edit the title to __b__ump?
bright glacier
I fucked up the commit title too
plain atlas
aw
bright glacier
Oh haha I did not realize I swore in a public channel (not that it matters but in general I try to not do that publicly)
gusty grail
you're grounded for 6 months 
plain atlas
Oh haha I did not realize I swore in a public channel (not that it matters but i...
lol yeah
flat krakenBOT
flat krakenBOT
flat krakenBOT
flat krakenBOT
grave fox
So a question about formatting items like function definitions/calls with arguments. Say I have something like this:
def some_really_long_name(some_big_arg, another_big_arg):
# do something
Now if black sees that the function definition line exceeds the line length I set, it'll format it like this:
def some_really_long_name(
some_big_arg,
another_big_arg,
):
# do something
My question is if it's possible to force black to format all functions one way or the other, not have a mix-and-match approach to it? I would like to keep the styling the same for readability, so going with the latter option for all my function definitions and calls would be preferable over the former
errant barn
I'm afraid not, we'll fit it onto one line if it can!
dense jungle
So a question about formatting items like function definitions/calls with argume...
Only in one direction. If you add a comma at the end the magic trailing comma will split it up
errant barn
Although, you could use
dense jungle
But I feel like that's pretty ugly for one-argument functions.
grave fox
Oh for one argument functions I'd be keeping it on a single line
My minor issue was when I would have some heavily indented piece of code where one function call would be formatted in a single line, but a following function call right below would just exceed the line length limit and split into multiple lines, which makes everything hard to read
But hey that tip to add a comma to the last argument works like a charm, thank you!
red sedge
woah did django land black?
red pendant
failed mypy when running pre-commit run -a from the basics of trying to setup the repo for black, here's a screenshot:
should this be an immediate concern?
dense jungle
failed mypy when running ```pre-commit run -a ``` from the basics of trying to s...
Probably related to your version of pytest. Would be good to get this fixed.
red pendant
yeah I found my issue https://github.com/python-gitlab/python-gitlab/issues/1539
dense jungle
that looks unrelated, totally different error
flat krakenBOT
steady plaza
I've made a .pre-commit-config.yaml with:
default_language_version:
python: python3.9
repos:
- repo: https://github.com/psf/black
rev: 21.12b0
hooks:
- id: black
I run pre-commit run --all-files.
Get error in file "black\cache.py", line 13": ModuleNotFoundError: No module named '_black_version'
I've tried reinstalling everything and cleaning pre-commit. Anyone have a clue?
stuck vapor
I can't reproduce that @steady plaza , it works fine for me
perhaps try updating pre-commit?
flat krakenBOT
steady plaza
I can't reproduce that <@!258915060426997760> , it works fine for me
perhaps tr...
Tried in another environment and it works. So weird. I think _black_version.py should be created when setup.py is run but _black_version.py does not exist in my .cache for pre-commit.
red pendant
How would I go about running a coverage test with my own test for black? I see something called .coveragerc is this something we are suppose to use?
errant barn
How would I go about running a coverage test with my own test for black? I see s...
Do you mean that you have written custom tests for Black and would like to measure coverage?
red pendant
actually, I was just about to write my own test case for black
it has to do with testing the # skip format
but i'm not sure where I would actually add that test case
errant barn
Oh I see
So adding tests in general
There are a bunch of test files here: https://github.com/psf/black/tree/main/tests/data They have a specific (although simple) structure of input and expected output separated by # output
errant barn
Yeah it can be a bit hard to parse what's going on ๐
red pendant
no problem, i greatly appreciate the guidance
it probably saved me quite a bit of time
so i just added a test in data called fmtskip7
the outside files evaluate these different source files so I have to add the string "fmtskip7" to SIMPLE_CASES in test_format.py
errant barn
If you created the file, yes. But if it was already there, then not needed!
red pendant
actually, i'll try testing this out right now instead.
sorry if i'm asking a lot of questions..how would I run the specific unit test I created with coverage?
errant barn
Yikes that might be a bit difficult ๐ since the tests are generated on the fly. There might be a pytest syntax for that but at least I'm not familiar with it.
red pendant
tox -e py
Isn't it enough to run overall coverage?
errant barn
```
tox -e py
```
Yep this performs the whole test run
and prints coverage info as well
red pendant
a user reported this bug here regarding the # fmtskip generating an error if it's put together
is this something we need to resolve? Becausei n documentation, we have # fmt skip which has a clear space between the hashtag and fmt
errant barn
Definitely, either we fix the comment and then ignore, or accept the ignore without a space. An internal failure is not an option ๐
red pendant
yeah i replicated the error on my machine
errant barn
Nicely done
red pendant
and added a test fmtskip7
errant barn
I'll respond to the issue as well
red pendant
a, b = 1, 2
c = 6 #fmt: skip
d = 5
yeah test_format.py loses coverage
if you're going to be responding to this issue, do you mind assisting me a bit? This is my first repository i'm trying to make contributions to, and I'd like to become more independent
errant barn
Sure!
errant barn
Yep, verified here too!
red pendant
?
red pendant
the format is corrected for the comment
errant barn
I have not used them
red pendant
fmt off and on i believe are for blocks
errant barn
that much I know :D
Oh, they need to be in the same level of indentation and start a block, so I guess inline comments are not handled at all
so we're good ๐
red pendant
so we just need to fix the issue with #fmt skip
im looking at the comments.py
FMT_SKIP: Final = {"# fmt: skip", "# fmt:skip"}
it can't be as easy as just...
errant barn
It's probably that easy ๐ but what about #fmt: skip etc.
red pendant
so basically we want to check the format and correct it
errant barn
I think let's hear from other maintainers on the issue for now, because they might disagree with my preference
I'd say 1) don't fix 2) accept arbitrary spacing
red pendant
yeah it seems if you add any amount of spacing the internal error dissapears
errant barn
Oh! So my example doesn't error?
red pendant
but it renders the skip ineffective
no it does
but something like:
a=1
a=1 # fmt: skip
a=1
I see, so that's corrected to # fmt: skip and not formatted
then it could be possible to reformat other types as well!
red pendant
errant barn
Oh.
red pendant
yeah that's what i meant
errant barn
Nevermind ๐
errant barn
And the comment isn't even fixed
red pendant
the error specifically only happens when you have ```
#fmt: skip
but yeah it doesn't reformat in other cases of arbitrary space
errant barn
Well this is a mess huh. Best to add this to the same issue, because this is definitely related.
red pendant
I can make a seperate bug report and link to the old one
errant barn
That's fine as well!
red pendant
there's no way to generate a visible report from the test coverage either right?
like to see where certain LOC ran
terse jay
Hi guys, how do I make when a person leaves their email on my site, then the bot will send him an email, information that he bought on my site? If you can help me, I'll pay
errant barn
Hi guys, how do I make when a person leaves their email on my site, then the bot...
wrong channel, also against the rules my mate!
gamble: !rule 9
!rule 9
late dewBOT
errant barn
YES it worked
red pendant
prior to the run with the test that has the buggy skip, test format was at 100% coverage
terse jay
I'm sorry, but where do I find help
errant barn
Try the lobby or general channels, or the help channels if you have specific questions ๐
red pendant
now its at 96%
errant barn
that function specifically?
terse jay
Thanks
red pendant
i think what might've happened is the internal error we suffer occurs prior to running these functions
errant barn
You're not talking about the file test_format.py are you?
Maybe I've confused myself
red pendant
yeah im referring to the test_format.py test coverage
errant barn
Odd that adding a data test affects that..
red pendant
anyway to debug?
errant barn
Oooh unless, is the test failing with the internal error?
red pendant
^ yes
errant barn
Oh I see, the format tests are naturally expected to pass. So coverage isn't really valid
red pendant
errant barn
It's the coverage of Black source that matters
If the test suite itself fails, then we can't gather coverage information
So we'd need to first fix the issue, then make the test and finally measure coverage
red pendant
errant barn
Or make a test runner that expects the failure. But we'd rather fix the bug, right? ๐
red pendant
yes
errant barn
Although it's good to have failure tests as well for user errors, but that's beside the point.
red pendant
this isn't doing its job I believe
def make_comment(content: str) -> str:
"""Return a consistently formatted comment from the given `content` string.
All comments (except for "##", "#!", "#:", '#'", "#%%") should have a single
space between the hash sign and the content.
If `content` didn't start with a hash sign, one is provided.
"""
Could very well be
mint barn
i know this is a python server but can anyone help me with this java error in cmd
plain atlas
!ot
late dewBOT
bright glacier
man, I don't know how possible it is to check whether we're in a new expression with the line transformer based system
AFAICT all of the AST information is lost after the AST visitation logic and the visited line is compacted into a list of leaves >.<
dense jungle
man, I don't know how possible it is to check whether we're in a new expression ...
what do you mean by "new expression"?
bright glacier
a = [x**y, x**y]
``` right now with the PR code, the lack of spaces between the power operator is allowed because the whole line has more than one hugging power operatorline.leaves=[Leaf(NAME, 'a'), Leaf(EQUAL, '='), Leaf(LPAR, ''), Leaf(LSQB, '['), Leaf(NAME, 'x'), Leaf(DOUBLESTAR, '**'), Leaf(NAME, 'y'), Leaf(COMMA, ','), Leaf(NAME, 'x'), Leaf(DOUBLESTAR, '**'), Leaf(NAME, 'y'), Leaf(RSQB, ']'), Leaf(RPAR, '')]
and this is what I'm working with ...
I feel like I'm missing something obvious though.
dense jungle
```python
a = [x**y, x**y]
``` right now with the PR code, the lack of spaces be...
Is that bad? That example looks fine to me with no spaces
bright glacier
It's just inconsistent with return x ** y / return x**y, a**b for example
dense jungle
Maybe we shouldn't space that either? The rule could simply be that we never space ** with simple LHS and RHS
bright glacier
Apparently ambv and a few people agreed that a single power op should not hug but honestly I don't understand whyyy exactly
dense jungle
might be worth just having a discussion with some of the interested people about exactly what the rule should be
bright glacier
Thinking about it some more (actually, a lot more) it seems to me like the easiest improvement here is to do what @paugier was suggesting all along, especially clearly in his latest comment, with the following additional twist: if ** is the only operator in the expression, then don't hug.
https://github.com/psf/black/issues/538#issuecomment-593957202
dense jungle
yeah, I don't really see a strong argument there for special-casing single expressions
flat krakenBOT
bright glacier
this feels fragile but then again this entire part of the codebase is fragile :p
skip: skip was honestly quite undertested IMO
tired shard
Could I get some more reviews on my PR?
flat krakenBOT
tired shard
I could possibly add root to ctx.obj, it would require passing but would be less compared to root as some of the functions already take in ctx
errant barn
Mjeh, could be worth a try, but let's see what the others think. I'm not sure what else the context is being used for
tired shard
one thing can be done, make root a class variable for Report, and move find_project_root to teh main black cli function so we can pass root when we are making Report
errant barn
Sounds possibly good as well ๐๐ฟ
tired shard
ok @errant barn i tried doing the above change and it looks much much better, pushing the code, would make it easier to decide also :)
errant barn
ok <@!273104906083106816> i tried doing the above change and it looks much much ...
Thanks a bunch! I'll have a look right away
errant barn
@tired shard psst, I think you didn't push it
flat krakenBOT
flat krakenBOT
bright glacier
I'm a little confused at why we're normalizing the paths to be from the root when emitting the log messages
Right now the code is careful to not modify the passed in SRCs as much as possible so when we do log 'em it's clear they're from what the user passed in
for example, the fact they're prefixed with black/ strikes me as odd given I explicitly passed in src
tired shard
Right now the code is careful to not modify the passed in SRCs as much as possib...
Actually yeah that's not needed since we have already explicitly shown the project root and sources path
tired shard
@errant barn pushed, i did push yest but the pre-commit failed, and then I never looked at it until now (wasn't at home)
bright glacier
this was surprisingly simple ๐ still got one major feature to add and a few minor ones
mint barn
what's the best way to make black still exit with 0 if it couldn't reformat a specific file because of a match statement, but still exit with non zero if something got reformated?
lament crow
Are you using latest version? Because there is basic support for 3.10 features added
mint barn
I have 21.12b0
lament crow
are you using target version 3.10?
as in, --target-version py310 in CLI arguments or this in pyproject.toml:
[tool.black]
target-version = ['py310']
no
lament crow
Yeah, you need this to have 3.10 support for now as it's still experimental
mint barn
hello guys , who knows how to connect MongoDB data base to a project using pythoin?
when ever I connect it it says err":{"name":"MongoError","message":"bad auth Authentication failed.","ok":0,"errmsg":"bad auth Authentication failed.","code":8000,"codeName":"AtlasError"}}
jagged fossil
hey guys, im trying to make a really basic plugin to copy paste docstrings from pdf. i havent been able to find any resources on how to make a plugin like black for pycharm. can anyone point me in the right direction?
gonna move this to help channel also
bright glacier
!remind 1D do a more proper review on psf/black#2656
toxic stormBOT
late dewBOT
bright glacier
I implemented your mypy_primer feature request for diff-shades Jelle :)
I guess if we're using a github URL I could rewrite the URL to point to the tree at the recorded revision ๐ค
dense jungle
thanks this is great ๐
short forge
<@&831776746206265384>
glacial gate
!ban 898297969684660224 Racism
late dewBOT
:incoming_envelope: :ok_hand: applied ban to @neon oasis permanently.
short forge
this is not a help channel #โ๏ฝhow-to-get-help
also the error message is pretty clear
swift fossil
Howdy y'all. I've been using Black forever, and I feel like it'd be a good time to give back. I'm gonna go through the contributing guide and try my best at some stuff. :'] Just wanted to say hello, etc, etc.
errant barn
Howdy y'all. I've been using Black forever, and I feel like it'd be a good time...
Yo, glad to hear and welcome! ๐
late dewBOT
!remind 1D do a more proper review on psf/black#2656
It has arrived!
viscid bane
How do I tell vscode to use the black setting that I defined in the pyproject.toml?
upbeat portal
idk if it's the right channel to ask, I started using Black a few weeks ago and one thing I noticed was the use of parenthesis in some cases (like backslashes)
I'm a bit confused about what those parenthesis are
# in:
if some_long_rule1 \
and some_long_rule2:
...
# out:
if (
some_long_rule1
and some_long_rule2
):
...```is (some_long_rule1 and some_long_rule2) a tuple with a boolean in it ?
(<- python beginner)
swift fossil
In PEP8 [https://www.python.org/dev/peps/pep-0008/#indentation] they specify this as the preferred method to multiple lines for bool masks, args, etc.
(PEP8: "The preferred way of wrapping long lines is by using Python's implied line continuation inside parentheses, brackets and braces. Long lines can be broken over multiple lines by wrapping expressions in parentheses. These should be used in preference to using a backslash for line continuation.")
So, it's not a tuple, it's just syntax which wraps around stuff for multilines and the like. A one element tuple looks like this: (1,) instead of (1).
plain atlas
well i forgot i asked to do this issue
!remind 2y actually finish that pr that makes black and isort conflict
late dewBOT
lament crow
2 years!?
bright glacier
๐ 
lament crow
you could comment on the issue to get unassigned to not discourage potential contributors :P
plain atlas
2 years!?
i don't check my github notifications enough, figured i'd add some leeway ยฏ_(ใ)_/ยฏ
plain atlas
you could comment on the issue to get unassigned to not discourage potential con...
nah, I'm gonna try it right now
bright glacier
I get emails for GitHub notifications <.<
lament crow
thought you were serious-ish with those 2 years and that you're not interested in working on it any time soon :P
bright glacier
Yes, it's why we have the --skip-trailing-comma flag
upbeat portal
In PEP8 [https://www.python.org/dev/peps/pep-0008/#indentation] they specify thi...
Thanks for explanation!
btw I'm kinda getting better at python (still beginner though), I'm still looking for a friendly open-source project that I can start contributing
is this a good idea to join this project? even if my contributions will be very limited?
bright glacier
Being an older project with a fair amount of cleanup needed there's quite a few rough edges to be honest but overall it's still a relatively easy project to contribute if you choose your issues the right way which I can help out with :)
Also, the core logic behind black is very tree (i.e. node and leaf) heavy so experience in those areas will be required if you want to make changes to the actual formatting -- FWIW when I first contributed I didn't have much experience with mangling tree structures but that didn't stop me from contributing in other ways
upbeat portal
Well I watch some AnthonyWritesCode streams and I've done AoC21/day10, so I have some veeery basic notions about linters/parsing
I discovered thanks to AoC, graphs and DFS this week. Is a tree a kind of graph (or inverse)?
bright glacier
yeah kinda -- lemme pull up an example tree black could operate on (after parsing)
upbeat portal
Being an older project with a fair amount of cleanup needed there's quite a few ...
I'd be very happy to help on some "easy" issues, please feel free to suggest ๐
bright glacier
!paste
late dewBOT
bright glacier
with open("daylily.txt", encoding="utf-8") as f:
blob = f.read().splitlines()
that's the blib2to3 AST for this code 
https://github.com/psf/black/issues/2560
https://github.com/psf/black/issues/2492
https://github.com/psf/black/issues/2097
Seems easy enough issues
dense jungle
^ wow my hack fixed bugs instead of introducing bugs
we should have a regression test for this though, another easy issue
flat krakenBOT
upbeat portal
Seems easy enough issues <:this:470903994118832130>
thanks! it's 1AM here, I'll have a look tomorrow ๐
you created a tag just for me (and other open-source beginners) โค๏ธ
very kind of you @bright glacier
bright glacier
I've made a "stable release" milestone fwiw: https://github.com/psf/black/milestone/1 -- I don't like using Projects as I find they take too much effort to manage and I end up forgetting about them (also the pre-existing stable project is in disorder), feel free to reorganize it Jelle
None of these are true blockers but they'd be nice to get in -- only formatting issues / PRs in this milestone would count as potential blockers due to the stability policy
Honestly should we aim to get ESP enabled by the stable release? Given there's some formatting changes I'd like to see made to ESP + that pesky performance penalty it feels optimistic that we'd be able to get it in.
Oh yeah, I suppose we should probably make a decision in regards to parenthesized withs by then too as those are gonna get more and more common
tired shard
Could I get another review on psf/black#2525 and I have mentioned the changes in psf/black#2526
toxic stormBOT
tired shard
https://github.com/psf/black/issues/2560
https://github.com/psf/black/issues/249...
For is_type_comment issue, the formatting of type comments would go into https://github.com/psf/black/blob/1c6b3a3a6fbc50b651d4ac34247903041d3f6329/src/black/comments.py#L105 right?
late dewBOT
src/black/comments.py line 105
def make_comment(content: str) -> str:```
flat krakenBOT
flat krakenBOT
mint barn
.
upbeat portal
Is black conflicting with flake8 ?
dense jungle
upbeat portal
black autoformats my_list[i:i+1] into my_list[i : i + 1]
lol exactly the issue I had
I just switched to mypy but now I have another problem/question ( #help-pretzel )
plain atlas
!remind 2D mark that isort issue up for grabs ๐
late dewBOT
bright glacier
https://black.readthedocs.io/en/stable/faq.html#why-are-flake8-s-e203-and-w503-v...
I'm so happy we added a FAQ, I'd've been happy even if this was the only entry
i'd've -- I saw this once on a YouTube video and I love it now.
stuck vapor
i'd've -- I saw this once on a YouTube video and I love it now.
was not expecting to find that in the Collins English dictionary, nice! https://www.collinsdictionary.com/dictionary/english/idve
bright glacier
hmm I wonder if y'all'd've would be in there too ๐
no I'm not kidding ๐ I'm genuinely curious
errant barn
I think we'd'ven't
bright glacier
'hank goodness
bright glacier
well this escalated quickly
Uh kinda? I was born in Canada but my mother's tongue isn't English but then again I can't really speak my mother's tongue ..
Like I know French better than I do my family language :?
errant barn
oh ๐ well close enough!
cause I've heard that there is no central authority on english language, so if it's in use, then it's correct ๐
especially if a (kinda) native says it
bright glacier
Wait, I might using "mother's tongue" incorrectly
I thought it was literally language your mother speaks but apparently means "native language"? Ugh I need to crack open a dictionary
bright glacier
Like I live in a non-English/French household but I can only speak languages of the place I live (which is English et le franรงais)
So technically if the bar for knowing a language is super duper low I'd count as a trilingual but practically I'd call myself 1.5 bilingual
bright glacier
well... that's awkward
Chinese in particular Mandarin
Can't write or read it all
errant barn
oh shoot, cool
bright glacier
Je peux seulement รฉcrire /lire en franรงais ou anglais
dense jungle
My wife is kind of like that, her family is Chinese but she can barely speak it now
errant barn
joo just nii, jep
bright glacier
I can't even read the latinized writing aka pinghi?
Romanization of Chinese (Chinese: ไธญๆๆไธๅ; pinyin: zhลngwรฉn lฤdฤซnghuร
well I was pretty wrong :p
errant barn
now to guess what's the meaning of all those accents
bright glacier
It's fun knowing some French since I can now attach meaning to accents and seem more cool than I really am
dense jungle
now to guess what's the meaning of all those accents
it's tone, though don't ask me which tone ๐
errant barn
Sure :D I only know basic umlauts, and the stress รฉ on my last name
bright glacier
I heard there's four ones, flat, rising, falling, and some wavy one ..?
errant barn
No wait, I can hear my father disapproving from 300 kilometers, not a stress, but an acute accent ๐
bright glacier
un accent aigu ๐ฅ
I should probably do my french gรฉo homework now lol.
Every damn time I get French ads inbetween my background music, it's really something.
errant barn
Beautiful language though, even though I'm quite clueless to its meaning
bright glacier
https://github.com/psf/black/issues/2688#issuecomment-994116954 this is my best guess at what they're aiming for roughly ๐คท
dense jungle
except the error message they posted says "cannot format file.py: INTERNAL ERROR"
unless that's from the template?
bright glacier
hmm it might be
flat krakenBOT
bright glacier
unless that's from the template?
So basically the whole main post is useless garbage while the comment they added is the actual issue, sigh.
dense jungle
right, I think you got the actual issue out though, thanks!
bright glacier
oh wow we actually assert that the right ANSI codes are expected in the output >.<
I was not expecting that at all.
plain atlas
https://github.com/psf/black/issues/2688#issuecomment-994116954 this is my best ...
that is the most comedic issue title i have seen
bright glacier
hmm I can't reproduce psf/black#2594
toxic stormBOT
red sedge
Thanks!
bright glacier
hmm I can't reproduce psf/black#2594
ah, typed-ast supports <> but not the built-in AST haha
bright glacier
man, honestly I might approve psf/black#2526 even though I don't understand the changes in regards to us printing the SRC
toxic stormBOT
plain atlas
man, honestly I might approve psf/black#2526 even though I don't understand the ...
monorepos.
( or similar )
bright glacier
yeah I know I just hate myself for going back and forth on this PR so many times
basically I'mma give up on fighting for my opinion here since clearly it's dumb
plain atlas
yeah I know I just hate myself for going back and forth on this PR so many times
flip is a command
heads approve it, tails deny it /s
bright glacier
@tired shard I'm so sorry for being so flip-floppy on this verbose PR, it looks good to me except now I realize why including the root in the log messages was a good idea (for the SRCs log mind you, I still disagree it's a good idea in the other logs).
Basically, my opinion here is use felix's output design because it's way better than what I can come up with but let's try to avoid touching the other printed paths in the logs since people are probably used to how it works
but overall, if it feels dumb, then ignore my opinion, I'm as done with this PR as you are most likely and I don't want to fight it anymore
PS. it crashes if you format a directory that doesn't share a common base within the project root
This is probably my fault and I'm sorry for this too.
also, it seems like felix's output has the "identified X as project root" line before the "using configuration from project root" line which seems better to me
bright glacier
all sources that are about to be formatted relative to the project root, meaning that if I'd run black . ../tests in one level into the project e.g. in project/src, I'd see something like about to format: project/src and about to format: project/tests
If my defeated ranty paragraphs were confusing, the TL;DR is to apply this comment to the sources line (and maybe the "reformatted X" lines too if just applying it to the SRCs line is dumb - i.e. revert my feedback).
bright glacier
https://github.com/psf/black/runs/4528535309, OK I'm done today
if someone wants to fix this I'd be happy but I'm not doing so right now.
flat krakenBOT
plain atlas
if someone wants to fix this I'd be happy but I'm not doing so right now.
gotcha covered, go take some time for yourself
flat krakenBOT
slow ibex
# this is invalid code:
for x in f(lambda: i := 0,):
pass
# black formats it to:
for x in f(
lambda: i := 0,
):
pass
remove the ,
dense jungle
https://github.com/psf/black/pull/2695 it works ๐
tired shard
yeah thanks richard, I will look into your feedback and address it soon :D
late dewBOT
tired shard
!remind 3d that
late dewBOT
bright glacier
you can reply to a message and the bot will use the linked message as the content field if it's not provided explicitly
tired shard
yeah, i was the one who created a issue for that lol
bright glacier
and I reviewed the contributing PR :)
flat krakenBOT
dense jungle
maybe their editor is doing something weird?
tired shard
<@!736234578745884682> I'm so sorry for being so flip-floppy on this verbose PR,...
I am still printing them relative to root for SRCs log and I think I am using felix's message design just using confgi is before identified, i would do that change.
tired shard
PS. it crashes if you format a directory that doesn't share a common base within...
hmm but I do call .absolute() on it
late dewBOT
src/black/__init__.py lines 533 to 536
paths = '", "'.join(
str(Path(source).absolute().relative_to(root)) for source in src
)
out(f'Sources to be formatted: "{paths}"', fg="blue")```Do you mean Path.resolve()?
tired shard
doesn't absolute also do that
errant barn
You know, I thought it did, but looking at pathlib's documentation I cant see a reference to it ๐
dense jungle
I think there's a longstanding bug about clarifying the behavior of those two
Isn't one of them undocumented?
tired shard
probably, well that works :D I will push the fix
errant barn
The link was visited as well ๐
tired shard
purple lol
errant barn
Oh, from the discussion: Path.absolute() means exactly Path.cwd() / p
tired shard
I am still printing them relative to root for SRCs log and I think I am using fe...
for this, we print out the srcs log in get_sources which called pretty later than the config log, so i am not sure if it is oky to move it down
felix had also probably suggested that
and i gave the same reasoning
here
dense jungle
now you have .absolute().resolve() though
errant barn
One option could be to only calculate the root once, maybe by caching or something.
Or pull the find_project_root out from find_pyproject_toml and just pass the found root there. Now it just directly calls it with the argument and it's only used there.
late dewBOT
src/black/files.py line 74
path_project_root = find_project_root(path_search_start)```for moving the root log above config?
tired shard
hmm, i could move find root and method to the main function and then cache them to ctx.obj[]
edit"
and richard review
which can then be used in get sources
errant barn
That could be nice, a small performance gain as well ๐
tired shard
not really, we only call find root once i think
tired shard
!remind delete 3716
late dewBOT
tired shard
ah yes
flat krakenBOT
frigid grotto
Hi, I've been using black for a while and want to contribute and was looking at https://github.com/psf/black/issues/2097
i see in the code comments that
All comments (except for "##", "#!", "#:", '#'", "#%%") should have a single
space between the hash sign and the content.
but the formatter only adds a space if there isn't one, and doesn't remove extra space
i was wondering if all comments should only have one space or should that only be for type comments
dense jungle
i see in the code comments that
```
All comments (except for "##", "#!", "#:", ...
Only type comments I think. Otherwise we may mess up commented out blocks of code.
frigid grotto
ah i thought it was required to be type:
tired shard
NOot sure
errant barn
as long as we're fixing the comments, I think #\s*type\s*:\s*whatever is fine in my opinion, but worth it to check especially from Jelle
frigid grotto
ok ngl when i wrote that i completely forgot regex existed
can change it to use it and make it a lot cleaner
tired shard
#\s*type\s*:\s*(.*) and then rewrite it using the group
frigid grotto
yup
errant barn
let's discuss details in the issue
dim haven
This turned out hard to read after some working sessions.. I just removed the tool that I wrote which formats first with black and afterwards it makes tabs of size 2, now I'm just using black hahaha
fringe coral
Is there a list of which Black version is good for some specific Python version? I'm getting an error with black 21.12b0 and Pyhton 3.8
Traceback (most recent call last):
File "/opt/miniconda3/envs/<env>/bin/black", line 6, in <module>
from black import patched_main
File "/opt/miniconda3/envs/<env>/lib/python3.8/site-packages/black/__init__.py", line 49, in <module>
from black.files import find_project_root, find_pyproject_toml, parse_pyproject_toml
File "/opt/miniconda3/envs/<env>/lib/python3.8/site-packages/black/files.py", line 27, in <module>
from black.handle_ipynb_magics import jupyter_dependencies_are_installed
File "/opt/miniconda3/envs/<env>/lib/python3.8/site-packages/black/handle_ipynb_magics.py", line 15, in <module>
from typing_extensions import TypeGuard
ImportError: cannot import name 'TypeGuard' from 'typing_extensions' (/opt/miniconda3/envs/<env>/lib/python3.8/site-packages/typing_extensions.py)
Simply by running black __init__.py on an empty init file
dense jungle
Is there a list of which Black version is good for some specific Python version?...
The Python version is fine, your typing-extensions is too old
fringe coral
I see. So this is basically an issue with conda messing up. Thanks
dense jungle
yeah, we require typing-extensions 3.10.0.0 in setup.py, looks like you have an older version
flat krakenBOT
bright glacier
https://github.com/psf/black/issues/2681 honestly, would it be better to just delete these log messages? I don't see much value having 'em and we keep getting new issues asking about them >.<
dense jungle
https://github.com/psf/black/issues/2681 honestly, would it be better to just de...
agree
bright glacier
technically the ideal option would be to integrate blib2to3's grammar caching with black's cache mechanism but that'd be complicated as this cache attempt occurs on import
bright glacier
technically the ideal option would be to integrate blib2to3's grammar caching wi...
actually no, it's not that complicated and I have a working implementation. Time to write a test (or two).
