What web stuffs?
#black-formatter
neon loom
bright glacier
basically websites. I can't share the more recent as it has private information but I was proud of the result.
the second latest would be https://ichard26.github.io which still has no real content yet :p
As much as people hate on CSS, I haven't endured much pain from it yet π although to be fair I'm working on small sites that don't even need multiple stylesheets
neon loom
Nice π
bright glacier
will check tomorrow
!remind 18H Don't forget to clear out some pending reviews once the diff-shades integration is useful again!!!
late dewBOT
bright glacier
hmm I might be in a meeting at that time, but it should be over soon after so it's probably fine.
bright glacier
For separating the test cases between preview and stable, there's two main ways we could do it:
- Keep them in separate files (affixed with some kind of marker / stored in a specific directory)
- Enhance the current test data format so an expected result can be defined for either (or both) modes
The latter would keep duplication down, but would be less flexible overall. Although to be fair it's not like we'd be forced to choose only one. We could implement both. I just want to document the general process.
dense jungle
I like the second idea, maybe add a separate # preview-output section
errant barn
Me too, all for reducing duplication
late dewBOT
!remind 18H Don't forget to clear out some pending reviews once the diff-shades ...
It has arrived!
bright glacier
I guess I'll review the currently open PRs now and then I can look into improving our testing infrastructure after
bright glacier
Removing multiple unnecessary parenthesizes for for assignment targes is unstable #2952
ah whatever, the typo can live on
bright glacier
It's this kind of instability I'm worried about with stuff like this
bright glacier
Finding lots of bugs in my reviews today 
dense jungle
good catch! I was worried about that sort of thing in my reviews but didn't look hard enough
bright glacier
if the PR above has a bug, it'd be on me with how thorough I'm trying to be π
!remind 3H finish off some more reviews
late dewBOT
bright glacier
I wonder how many non magic trailing comma instability bugs we're potentially hiding with the second pass hack
winged whale
I am setting default formatter to be black formatter
How to specify that? Thank you
bright glacier
you're missing a comma :p
bright glacier
lemme pull out my vscode config file and see what the right value is
winged whale
I dont know why "[python]" is still orange
bright glacier
"[python]": {
"editor.defaultFormatter": "ms-python.python"
},
this is what I have
dense jungle
are you missing a comma on line 7 above
cloud quarry
put a comma at the end of line 7
winged whale
Sorry got it
Do I need to click "save" or anything to take effect?
I can't see any changes when I push shift + alt + f
bright glacier
did you set
"python.formatting.provider": "black",
too?
bright glacier
it needs to be outside of the python block, i.e. in the main configuration scope
bright glacier
πΊ lovely!
late dewBOT
!remind 3H finish off some more reviews
It has arrived!
long frigate
(replying to @bright glacier here, since it seems more topical than #unit-testing )
Black no longer crashes on parenthesised context managers, but it doesn't add parens when needed - even if I specify -t py39 or py310. See eg
https://black.vercel.app/?version=main&state=_Td6WFoAAATm1rRGAgAhARYAAAB0L-Wj4AHRAK5dAD2IimZxl1N_Wg0-BAclsm6AXNYZqFDXdkKKtCkCz8fX-xEYu0zyMr0UnfF3cCKW4WPbUzY_SKtXxPCEVnsSqmPTCseINvICJiA4wa-ldoG94oFLw5QLkhuSPUbE5T9V5MKsv_YeT2anIazy9fx0vYdDMgP7wrueuEOoN7VlkkuODYIdaPA0rvUajAO_7G49LXyDn3uRuxsdN_lCE6LsITgGVfotiSdZKmXAtNonIAAAAHUbwYesYXDYAAHKAdIDAAD0D13AscRn-wIAAAAABFla
and https://github.com/psf/black/issues/664
winged whale
Is that a preferred way for you guys to breakdown list comprehension into multiple lines?
errant barn
Yeah
winged whale
I want to keep it remain single line, can I change the setting?
errant barn
You can change line length limits, but I'd maybe just move the comment
winged whale
I see. Thank you
errant barn
ππΏ
flat krakenBOT
bright glacier
(replying to <@!429445826251587616> here, since it seems more topical than <#463...
yeaahh I misread the room
It's a bit tricky with how it was an undocumented feature, but IIRC I had a idea to make it work. It just involves making it special as usual with hacks :p
nova tinsel
hey richard, i accidentally edited the duration of this reminder. you need it to remind you april 23?
can you do !remind list, i just want to make sure that i did indeed edit your reminder
bright glacier
Yea that was the intention, but honestly it's more of a general reminder for "sometime in the future"
!remind list
late dewBOT
nova tinsel
ok you can !remind edit duration 4276 1m
bright glacier
!remind edit duration 4276 1m
late dewBOT
nova tinsel
sry :) i'll be more careful haha
bright glacier
no worries!
long frigate
It's a bit tricky with how it was an undocumented feature, but IIRC I had a idea...
I do love a good hack! Sadly most of my best in Hypothesis are so cursed that I don't want to tell people about them for fear of inspiring others π
molten kernel
how do i comment out a certain area of text from black formatting?
bright glacier
surround it with # fmt: off and # fmt: on -- note they have to be on the same level of indention
you might be able to try # fmt: skip but it's rather buggy right now
dense jungle
looks like f-string time
bright glacier
is this a graphQL query? they are always a pain to write out :)
bright glacier
My solution when I last dealt with graphQL was to add placeholder values and use str.replace on 'em -- not exactly pretty
faint haven
possibly one of the few cases where string.Template could be an alright solution (https://docs.python.org/3.10/library/string.html#template-strings)
bright glacier
I forgot string.Template was a thing π
glacial swan
Ok cool
thank you
I can't wait to use it
so I don't need to use my hackyish
function anymore
bright glacier
!remind 5H do a systematic review of the current code style and make sure the docs are up to date
late dewBOT
bright glacier
My wifi connection is awful on my main machine and I don't know why... loading any GitHub page is unreliable :/
well -- at least I can still access the web lol
flat krakenBOT
bright glacier
fun fun fun, 2.4GHz works but tops out @ 2 Mbps while 5GHz doesn't work at all (pictured below)
Do we have an issue about taking advantage of parenthesized with @dense jungle ?
I thought we did, but I can't find one hmmm
dense jungle
Not sure, can take a look
bright glacier
hmmm I remember we added a section to the future style docs with a better style, I guess whatever issue had that shared around would be it
late dewBOT
!remind 5H do a systematic review of the current code style and make sure the do...
It has arrived!
dense jungle
turned on this option so we can update branches and get more diff-shades goodness π
bright glacier
dense jungle
https://black.readthedocs.io/en/latest/the_black_code_style/future_style.html#us...
Seems like it's new in #2174 (your big docs rewrite)
bright glacier
it goes way back to March 2020 https://github.com/psf/black/issues/664#issuecomment-594771153
I forgot we also agreed to use backslashes π
dense jungle
hm did we actually commit it? I checked out the commit before your PR but don't see it
dense jungle
I forgot we also agreed to use backslashes π
yeah let's not
Since we have a nicer option in 3.9+
oh the rest of the section was from somewhere else, but it looks like you added the sentence "Although when the target version is Python 3.9 or higher, Black will use parentheses instead since theyβre allowed in Python 3.9 and higher."
bright glacier
That sounds about right but I can barely remember :)
Can we get away marking parenthesized with as a 3.9 feature? It could break code using any implementation that isn't CPython or PyPy, but then again it's only those two that are used in practice
dense jungle
Can we get away marking parenthesized with as a 3.9 feature? It could break code...
Yes, I think it's fine
bright glacier
I'm fine with it too. Wish it was made official but oh well π
flat krakenBOT
dense jungle
This is pretty bad, would appreciate contributions addressing this
frozen badge
the fun moment when you write a message to report a bug and then 2 issues get opened before you can send
dense jungle
Either pinning click or making us not use _unicodefun
We should be able to do a release today to deploy the fix
fair spade
Hello everyone, I am trying to use black but I am fairly new with this tool. I wanted to ask help here instead of opening a ticket.
I have a project running under python 3.9 inside an venv (called .env). I have installed black under this venv with the pip command as instructed in the get started.
When I launch black i get the following output:
black
Traceback (most recent call last):
File "/Users/.../myProject/.env/bin/black", line 8, in <module>
sys.exit(patched_main())
File "src/black/__init__.py", line 1423, in patched_main
File "src/black/__init__.py", line 1409, in patch_click
ImportError: cannot import name '_unicodefun' from 'click' (/Users/.../myProject/.env/lib/python3.9/site-packages/click/__init__.py)
Any idea of what's missing? Also, I am running on macOS if it can help the troubleshooting
frozen badge
Hello everyone, I am trying to use black but I am fairly new with this tool. I w...
do pip install click==8.0.4
fair spade
do `pip install click==8.0.4`
Thanks @frozen badge , actually I have 8.1.0 installed, would it be the source of my issue?
frozen badge
yes
flat krakenBOT
frozen badge
they removed something that black uses in the 8.1.0 release which was like 30 minutes ago
fair spade
ah ah ah, i started to use black at quite the wrong moment π
dense jungle
ah ah ah, i started to use black at quite the wrong moment π
very unlucky lol π
fair spade
that's me π
frozen badge
ah ah ah, i started to use black at quite the wrong moment π
same happened to a friend of mine which is why im here 
fair spade
Is there any open ticket that i can follow about this issue?
I didnβt found anything earlier when i tried to search for one...or maybe i missed it π€
fair spade
Thanks a million @frozen badge
flat krakenBOT
dense jungle
thanks for the quick PR @bright glacier ! fyi you can edit the changelog from the web UI at https://github.com/psf/black/blob/ichard26-patch-1/CHANGES.md
frozen badge
(or press . in the repo which should open github.dev/vscode web )
dense jungle
hm, something still wrong with the mypyc build in https://github.com/psf/black/runs/5725257122?check_suite_focus=true
bright glacier
OK, so let's get a release out ASAP. We'll just ignore mypyc and we can do a follow up point release with compiled wheels later.
Unless we want to drop 3.6 like right now which seems untenable..?
bright glacier
hm, something still wrong with the mypyc build in https://github.com/psf/black/r...
I assume 8.1.0 dropped other stuff from the API that we were using? hence mypyc freaking out
dense jungle
hm, seems bad to have a release without wheels
dense jungle
I assume 8.1.0 dropped other stuff from the API that we were using? hence mypyc ...
why isn't that build using click<8.1.0?
bright glacier
it's just a speed up, we'll still have the pure python one obviously
bright glacier
why isn't that build using click<8.1.0?
oh hmm, yeah
did I mess up the syntax?
oh right, the workflow installs click as it's a dependency of the helper script and that's still unbounded
dense jungle
I'm going to try removing the patch entirely as some people on GH were suggesting
bright glacier
I think when I get back to my main machine I'll set up some CI on another repo that tests out the development versions of our dependencies
this blew up way more than I expected
bright glacier
I'm going to try removing the patch entirely as some people on GH were suggestin...
We could also just fix the patch code as it was catching the wrong exception apparently
Also I now realize click 8.1.0 dropping 3.6 is not a problem as we still support the older 8.x.x versions.. ah the joys of stress
dense jungle
think we do still need the patch ```% LANG=C python -m black src/black/init.py
,,,
RuntimeError: Click will abort further execution because Python was configured to use ASCII as encoding for the environment. Consult https://click.palletsprojects.com/unicode-support/ for mitigation steps.
This system lists some UTF-8 supporting locales that you can pick from. The following suitable locales were discovered: af_ZA.UTF-8, am_ET.UTF-8, be_BY.UTF-8, bg_BG.UTF-8, ca_ES.UTF-8, cs_CZ.UTF-8, da_DK.UTF-8, de_AT.UTF-8, de_CH.UTF-8, de_DE.UTF-8, el_GR.UTF-8, en_AU.UTF-8, en_CA.UTF-8, en_GB.UTF-8, en_IE.UTF-8, en_NZ.UTF-8, en_US.UTF-8, es_ES.UTF-8, et_EE.UTF-8, eu_ES.UTF-8, fi_FI.UTF-8, fr_BE.UTF-8, fr_CA.UTF-8, fr_CH.UTF-8, fr_FR.UTF-8, he_IL.UTF-8, hr_HR.UTF-8, hu_HU.UTF-8, hy_AM.UTF-8, is_IS.UTF-8, it_CH.UTF-8, it_IT.UTF-8, ja_JP.UTF-8, kk_KZ.UTF-8, ko_KR.UTF-8, lt_LT.UTF-8, nl_BE.UTF-8, nl_NL.UTF-8, no_NO.UTF-8, pl_PL.UTF-8, pt_BR.UTF-8, pt_PT.UTF-8, ro_RO.UTF-8, ru_RU.UTF-8, sk_SK.UTF-8, sl_SI.UTF-8, sr_YU.UTF-8, sv_SE.UTF-8, tr_TR.UTF-8, uk_UA.UTF-8, zh_CN.UTF-8, zh_HK.UTF-8, zh_TW.UTF-8
Yeah having black refuse to work sounds untenable even if it's only on 3.6
flat krakenBOT
dense jungle
^ this works locally
bright glacier
click, not black but thanks!
dense jungle
lol oops
dense jungle
this is still broken https://github.com/psf/black/runs/5725593238?check_suite_focus=true
Looks like it could be a bug in Click's typings
@bright glacier could we pin to <8.1.0 just for the mypyc build?
flat krakenBOT
bright glacier
<@!429445826251587616> could we pin to <8.1.0 just for the mypyc build?
it's because it installs click early without the bound: https://github.com/psf/black/blob/ac7402cbf6a0deb5c74e9abcffc5bd7b1148fda5/.github/workflows/diff_shades.yml#L59-L67
late dewBOT
.github/workflows/diff_shades.yml lines 59 to 67
- name: Install diff-shades and support dependencies
run: |
python -m pip install https://github.com/ichard26/diff-shades/archive/stable.zip
python -m pip install click packaging urllib3
python -m pip install -r .github/mypyc-requirements.txt
# After checking out old revisions, this might not exist so we'll use a copy.
cat scripts/diff_shades_gha_helper.py > helper.py
git config user.name "diff-shades-gha"
git config user.email "diff-shades-gha@example.com"```yes, just pushed a commit to pin there π
dense jungle
diff-shades is building fine on my PR now too. Shall we wait for CI to pass on it, merge it, and release 21.3.0?
dense jungle
um 22.3.0
bright glacier
diff-shades is building fine on my PR now too. Shall we wait for CI to pass on i...
Everything feels super rushed, but I don't see what could go wrong if we leave off publishing the mypyc wheels
dense jungle
Everything feels super rushed, but I don't see what could go wrong if we leave o...
Any reason not to publish the wheels too?
bright glacier
those wheels are still the most unreliable part of the release process
dense jungle
We were overdue for a release anyway π
bright glacier
Any reason not to publish the wheels too?
No? They're just slow to build, like an hour or so
dense jungle
and you still have to manually run some script for it, right?
bright glacier
It's still manual, yes but I trust mypyc enough where I can probably skip the manual testing so it's basically just "fire the workflow, wait 30 minutes, download the wheels and use twine to push 'em"
after all we've been using it in the diff-shades workflow for a while now
dense jungle
OK, let's just do that for this release too then
bright glacier
I've been meaning to upstream the mypyc wheel build workflow, but clearly I haven't gotten to it yet
dense jungle
the list of linked issues on https://github.com/psf/black/issues/2964 is scary
bright glacier
OK so after calming down and figuring out what is even happening the plan is to essentially:
- merge psf/black#2966
- go through the release process for black 22.3.0
- release it
- and at the same time also build the mypyc wheels
toxic stormBOT
bright glacier
do we want to wait to release everything at once or upload the wheels later once they're done?
dense jungle
Right. I can do steps 1-3. I think the wheels can be added later
Given the amount of breakage the faster we release the better
bright glacier
I agree.
dense jungle
will prepare the release PR noow
bright glacier
52 thumbs up π
flat krakenBOT
bright glacier
OK I'll fire the mypyc build workflow now
dense jungle
bright glacier
gosh the overthinking part of me is now worrying that the fix doesn't work, ahhhhh
dense jungle
that would be embarrassing
dense jungle
time to eat lunch
bright glacier
time to actually drink my french vanilla, I didn't drink it because I was too focused on this madness :p
bright glacier
why :/
I'll push the other ones I have right now, but this is annoying and odd. Seems like actions/create-artifact made a broken zip file?
bright glacier
seems rather similar to https://github.com/psf/black/pull/2682/files
long frigate
Psst, fully automatic release processes are great https://hypothesis.works/articles/continuous-releases/
dense jungle
Psst, fully automatic release processes are *great* https://hypothesis.works/art...
our releases are pretty automated too (apart from the mypyc stuff). I feel like releasing every change immediately would create a lot of noise for users, but it's an interesting model
bright glacier
I'm setting up a repo that runs black's test suite against the latest development version everyday and I'm hitting quite a few issues
at this point 4 of our dependencies have dropped 3.6 support, and the latest aiohttp also breaks our test suite which is just lovely
bright glacier
jeez that's a concerning amount of overrides https://github.com/ichard26/black-deps-ci/blob/4e25ac3dea340a956e35dab37b2ac5c4177c105e/generate.py#L9-L24
So in total 7 of our user/test dependencies don't support 3.6 anymore (also yeah once aiohttp 4.0.0 goes out of alpha our CI is blowing up)
and I didn't even include all of our test dependencies as pytest is the only one necessary to run the test suite
dense jungle
I think https://github.com/pallets/click/issues/2227 is affecting us too, saw that same error in the mypyc output
bright glacier
well that's great 
long frigate
our releases are pretty automated too (apart from the mypyc stuff). I feel like ...
Sure, but you still get disproportionate benefits from automating the last part of the process - anyone authorised can cut a release just by merging.
For smaller projects I just python setup.py sdist bdist_wheel && twine upload --skip-existing dist/* at the end of passing builds on main - it only releases if I've bumped the version number!
neon loom
We need use GitHub releases tho as we upload binaries there that people use now ...
Something make me uneasy about a commit being able to cause all that uploading and potential mess with a few bugs
long frigate
Still doable with github actions!
And you can use the CODEOWNERS file to restrict permissions appropriately.
round ore
whats going on everybody
flat krakenBOT
Upgrade on library click https://pypi.org/project/click/8.1.0/ broken, I think we can solve restricting version of library to 8.0.4
red sedge
Sure, but you still get disproportionate benefits from automating the last part ...
ew setup.py sdist
pipx run build && pipx run twine upload --skip-existing dist/* plz
lament crow
re: #2966 he could have had 21.10b0 pinned, just not click
flat krakenBOT
long frigate
`pipx run build && pipx run twine upload --skip-existing dist/*` plz
ew, build-time dependencies π
flat krakenBOT
red sedge
?
dense jungle
@lament crow thanks for the support in telling people they have the wrong version of Black installed π
lament crow
hah, sure
bright glacier
try:
from aiohttp.test_utils import unittest_run_loop
except ImportError:
# unittest_run_loop is unnecessary and a no-op since aiohttp 3.8, and aiohttp 4
# removed it. To maintain compatiblity, we can make our own no-op decorator.
def unittest_run_loop(function: Callable[..., None]) -> Callable[..., None]:
return function
tests/test_blackd.py:22:5: error: All conditional function variants must have identical signatures
def unittest_run_loop(function: Callable[..., None]) -> Callable[..., None]:
not sure why mypy is erroring here hmm, especially as I can't even tell what other function variants exist
dense jungle
it's very particular about this
little parrot
Is the first param named function in the unit test version?
dense jungle
like maybe the argument name is different
I was looking at an issue the other day where someone suggested mypy should tell you what the difference in the signature is
bright glacier
late dewBOT
aiohttp/test_utils.py lines 487 to 498
def unittest_run_loop(func: Any, *args: Any, **kwargs: Any) -> Any:
"""
A decorator dedicated to use with asynchronous AioHTTPTestCase test methods.
In 3.8+, this does nothing.
"""
warnings.warn(
"Decorator `β@unittest_run_loop`β is no longer needed in aiohttp 3.8+",
DeprecationWarning,
stacklevel=2,
)
return func```and take *args and **kwargs
bright glacier
It'd be nice if mypy emitted the other signature variants but I guess that's too difficult
(although I'm still surprised how it's finding this signature as I don't think typeshed has aiohttp, and I have aiohttp 4 installed)
dense jungle
It'd be nice if mypy emitted the other signature variants but I guess that's too...
yes that's what the issue I referred to asked for π it should be possible
dense jungle
(although I'm still surprised how it's finding this signature as I don't think t...
aiohttp is py.typed I think
bright glacier
β― pip list | grep "aiohttp"
aiohttp 4.0.0a1
unittest_run_loop doesn't exist since aiohttp 4
dense jungle
maybe mypy is looking at a different venv?
late dewBOT
aiohttp/test_utils.py lines 453 to 466
def unittest_run_loop(func: Any, *args: Any, **kwargs: Any) -> Any:
"""A decorator dedicated to use with asynchronous methods of an
AioHTTPTestCase.
Handles executing an asynchronous function, using
the self.loop of the AioHTTPTestCase.
"""
@functools.wraps(func, *args, **kwargs)
def new_func(self: Any, *inner_args: Any, **inner_kwargs: Any) -> Any:
return self.loop.run_until_complete(
func(self, *inner_args, **inner_kwargs))
return new_func```turns out it was not even deprecated in 4.0.0a1 Β―_(γ)_/Β―
I guess they keep multiple active branches hence this situation..?
flat krakenBOT
flat krakenBOT
lament crow
Ugh
You would think that a project aimed at developers wouldn't get as many issues about something that has already been reported and fixed...
flat krakenBOT
ripe imp
Hello. I have an issue using black formatter for my python code. I installed it and ran it: black main.py. But instead of formatting indentations it just crashes.
errant barn
What's the message? And have you managed to narrow down the cause at all?
ripe imp
error: cannot format main.py: Cannot parse: 17:0: print("This country code was not found") All done! 1 file failed to reformat.
The problem was that print line had an indentation problem. I thought it should fix it to see if it works, but it crashed. I then fixed indentation myself and black returned: All done! β¨ π° β¨ 1 file left unchanged.
errant barn
Ah, we can't really do much if the file is invalid Python
ripe imp
so it can only work for correct syntax?
errant barn
Yep, since we rely on Python parsers to do the formatting in the first place
ripe imp
okay, thank you. And another question. I have installed black and use it in my local machine as a script in my package.json. But if I upload this project to github and someone will clone repo, will he be able to run script?
errant barn
Hmm, I'm not that familiar with the machinery, but as long as they have your configuration (if any) and have installed Black, it should be fine!
Ordinarily you'd specify Black as some type of dev requirement for people to easily install everything they need
ripe imp
I did not configure my black , I used default configuration if it matters
errant barn
Should be that much easier then
ripe imp
okay! Thank you so much!
errant barn
You're most welcome!
flat krakenBOT
flat krakenBOT
bright glacier
https://github.com/psf/black/runs/5760994213?check_suite_focus=true hmm I guess mypyc just ignored that type ignore? that's fun
bah, I'm dumb, it's main that's still problematic
dense jungle
https://github.com/psf/black/runs/5760994213?check_suite_focus=true hmm I guess ...
I think it's because it's building main
flat krakenBOT
dense jungle
huh there is no #2980
bright glacier
Yay I can remove that override now :)
flat krakenBOT
worn badge
hi all! I just discovered that black defaults to an 88 char line length, while PEP8 specifies 79. Why does black allow a longer line length?
flat krakenBOT
short forge
79 is insanely short
Even 88 is short imho, i use 100
worn badge
I thought that was the point of the standard - to standardise and remove opinions
faint haven
hi all! I just discovered that black defaults to an 88 char line length, while P...
The reasoning is covered in the docs https://black.readthedocs.io/en/stable/the_black_code_style/current_style.html#line-length
worn badge
I missed that, thanks!
Just read it - does not really explain why black decided to break with PEP8 rather than trying to change PEP8
olive shell
Some teams strongly prefer a longer line length. For code maintained exclusively or primarily by a team that can reach agreement on this issue, it is okay to increase the nominal line length from 80 to 100 characters
<pep8.org>
(Although it does then say "comments and docstrings should still be 72 chars")
worn badge
My point is that 88 should not be the default, PEP8 should be the default. If teams want to go bigger that's up to them
I expect I am not alone in assuming that a PSF project would follow PSF standards out of the box
red sedge
PEP8 says not to follow it right at the top
But ultimately it doesn't matter in the grand scheme of things
teal goblet
imo 88 allows for a bit of padding in case your statement goes just over the limit.
flake8 actually uses 79 char limit
hearty canopy
ambv was kind of rude in that discussion, but he explained the rationale. Not much of why it's 88, but why it is now hard to change that default.
flat krakenBOT
silent apex
I'm guessing they're running a version of 3.11 that doesn't have tomllib
bright glacier
I honestly have no idea though, since 3.11a6 is the most recent prerelease and the docs already have tomllib
flat krakenBOT
dense jungle
I'm guessing they're running a version of 3.11 that doesn't have tomllib
yeah probably. GH Actions might not have the latest alpha yet
speaking of 3.11, do we have an issue already for except* support?
late dewBOT
silent apex
I honestly have no idea though, since 3.11a6 is the most recent prerelease and t...
I just installed 3.11.0a6 and import tomllib errors
dense jungle
I think we merged it after the alpha was released
silent apex
I just built with pyenv
Python 3.11.0a6 (main, Mar 31 2022, 16:53:31) [GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tomllib
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'tomllib'
flat krakenBOT
flat krakenBOT
neon loom
I think we should accept a tab argument.
errant barn
About damn time
neon loom
Tabs are the best.
flat krakenBOT
flat krakenBOT
flat krakenBOT
dense jungle
lol, great timing here
bright glacier
Yes, but the sheer amount of syntax tree code involved scares me :)
I'll get to it soon, I'm just cleaning up some of my various spreadsheets
flat krakenBOT
flat krakenBOT
bright glacier
that's a very fast review cycle for a formatting PR π
I'm still reviewing / editing psf/black#2926
toxic stormBOT
dense jungle
that's a very fast review cycle for a formatting PR π
I won't merge it yet
bright glacier
we should probably just rewrite this test as it's doing nothing ever since click 8 renamed its die on LANG=C function
it's complaining about test_black.test_shhh_click
bright glacier
Building wheels for collected packages: aiohttp, frozenlist, multidict, yarl
Building wheel for aiohttp (pyproject.toml): finished with status 'error'
error: subprocess-exited-with-error
Γ Building wheel for aiohttp (pyproject.toml) did not run successfully.
β exit code: 1
β°β> [22 lines of output]
running bdist_wheel
running build
running build_py
creating build
creating build/lib
creating build/lib/aiohttp
copying aiohttp/web_ws.py -> build/lib/aiohttp
copying aiohttp/worker.py -> build/lib/aiohttp
copying aiohttp/multipart.py -> build/lib/aiohttp
copying aiohttp/web_response.py -> build/lib/aiohttp
copying aiohttp/client_ws.py -> build/lib/aiohttp
copying aiohttp/test_utils.py -> build/lib/aiohttp
copying aiohttp/tracing.py -> build/lib/aiohttp
copying aiohttp/web_exceptions.py -> build/lib/aiohttp
copying aiohttp/web_middlewares.py -> build/lib/aiohttp
copying aiohttp/web.py -> build/lib/aiohttp
copying aiohttp/http_exceptions.py -> build/lib/aiohttp
copying aiohttp/web_app.py -> build/lib/aiohttp
*********************
* Pure build *
*********************
error: can't copy 'aiohttp/streams.py': doesn't exist or not a regular file
[end of output]
note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for aiohttp
Building wheel for multidict (pyproject.toml): started
Building wheel for multidict (pyproject.toml): finished with status 'done'
Created wheel for multidict: filename=UNKNOWN-0.0.0-py3-none-any.whl size=965 sha256=f28320f7795a1425546c789fd3e7e8610cd6b2e9a4e06951d985ed4efd5a816e
Stored in directory: /Users/runner/Library/Caches/pip/wheels/e4/80/84/c322ab45d6218ff609d81de8343014271fec74423a3b9bfbe7
WARNING: Built wheel for multidict is invalid: Wheel has unexpected file name: expected 'multidict', got 'UNKNOWN'
Not sure why it's now crashing as aiohttp=3.8.1 is not recent at all
dense jungle
could just be random
red sedge
cosmic ray hit the streams.py inode?
bright glacier
that would've been a real fun april fools, CI crashes due to a cosmic ray βοΈ
flint cedar
anyone know why black sometimes doesn't conform with pep8
dense jungle
anyone know why black sometimes doesn't conform with pep8
Can you give examples?
flint cedar
like, instead of
[item1,
item2,
item3]
it does
[
item1,
item2,
item3,
]
i think i've read somewhere in pep8 that the first one is preferred? i may be wrong rhough
dense jungle
Seems like you should read PEP 8 before claiming Black doesn't conform with it π
plain atlas
Traceback (most recent call last):
File "/home/runner/work/monty-python/monty-python/.cache/py-user-base/bin/black", line 8, in <module>
sys.exit(patched_main())
File "/home/runner/work/monty-python/monty-python/.cache/py-user-base/lib/python3.9/site-packages/black/__init__.py", line 1372, in patched_main
patch_click()
File "/home/runner/work/monty-python/monty-python/.cache/py-user-base/lib/python3.9/site-packages/black/__init__.py", line 1358, in patch_click
from click import _unicodefun
ImportError: cannot import name '_unicodefun' from 'click' (/home/runner/work/monty-python/monty-python/.cache/py-user-base/lib/python3.9/site-packages/click/__init__.py)
errant barn
psf/black#2964
toxic stormBOT
plain atlas
ye i figured updating dependencies would help
waiting for them to finish updating lol
now i get to relock it again since i had a 21.x build
hm
what is the policy on backports with the new stability policy
errant barn
meaning we'd release 21.x.x now? :D
lament crow
lol
flat krakenBOT
red sedge
!paste
late dewBOT
red sedge
I'd expect redundant parenthesis in yield to go away too
# Cr@zY Br@ck3Tz
async def main():
yield (
(((((((((((((
((( (((
((( (((
((( (((
((( (((
((black(1)))
))) )))
))) )))
))) )))
))) )))
)))))))))))))
)
tbh a whole bunch of stuff....
# Cr@zY Br@ck3Tz
async def main():
print(
(((((((((((((
((( (((
((( (((
((( (((
((( (((
((black(1)))
))) )))
))) )))
))) )))
))) )))
)))))))))))))
)
yes, there are a few issues about that. @mint barn-git has been sending a series of PRs cleaning up redundant parens in a few places
red sedge
oh neato
dense jungle
but unfortunately it needs specific code for each case
red sedge
reminds me of
bright glacier
what is the policy on backports with the new stability policy
We don't do backports.
I'd rather not do backports given how much complexity that'd introduce into our development flow, but I'm fine doing post release patches on a purely adhoc basis.
Still not convinced we need to patch the older versions yet as I haven't seen many people ask for it.
dense jungle
I think it doesn't make much sense to do backports within the 22.* releases, since everyone not using preview mode should be able to upgrade safely without formatting changes
Maybe once we release 23.1.0 next year there will be calls to continue 22.* backports. I don't really want to do that though.
bright glacier
What's the best way forward for dealing with the click 8.1+ fallout @dense jungle ?
dense jungle
What's the best way forward for dealing with the click 8.1+ fallout <@!783088578...
does anything else need to be done?
Only problem I know of is a CI failure with mypy on our end but that should be fixable
bright glacier
I'm thinking to just update our type ignores and add a click>=8.1.0 pin (so everyone's cached hooks get evicted)
bright glacier
does anything else need to be done?
oops sorry I meant our lint job specifically
dense jungle
I'm thinking to just update our type ignores and add a click>=8.1.0 pin (so ever...
sure
bright glacier
man, there's a typo in the commit message >.<
naive wren
bright glacier
Please refrain from spamming on topic channels @naive wren .
!remind 20H try to respond to https://github.com/psf/black/issues/2883 ...?
late dewBOT
tired flax
hey, is there a way to make black use single quotes instead of double quotes?
tired flax
hm
teal goblet
black is opinionated like that.
You can probably force it on an individual basis by using a " character in your string.
late dewBOT
!remind 20H try to respond to https://github.com/psf/black/issues/2883 ...?
It has arrived!
flat krakenBOT
flat krakenBOT
bright glacier
if anyone feels like debugging why black's test suite is failing with the latest development versions (i.e. from the main/master branch) here ya go: https://github.com/ichard26/black-deps-ci/actions/runs/2105909618
I'll note that it's only failing on Python 3.8 and higher since only those versions are supported by the latest versions of the jupyter deps (or at least that's my guess)
bright glacier
if anyone feels like debugging why black's test suite is failing with the latest...
@stuck vapor could you take a look into this? installing ipython from github main fails test_ipynb.test_set_input
def test_set_input() -> None:
src = "a = b??"
with pytest.raises(NothingChanged):
> format_cell(src, fast=True, mode=JUPYTER_MODE)
E Failed: DID NOT RAISE <class 'black.report.NothingChanged'>
from black import format_cell, Mode
src = "a = b??"
dst = format_cell(src, fast=True, mode=Mode(is_ipynb=True))
print(repr(dst))
you can notice the difference using this
here's the diff between ipython 8.2.0 and main: https://github.com/ipython/ipython/compare/8.2.0...master
late dewBOT
:incoming_envelope: :ok_hand: applied mute to @sick patrol until <t:1649408812:f> (9 minutes and 59 seconds) (reason: newlines rule: sent 119 newlines in 10s).
spark bone
@sick patrol Please stop spamming your question everywhere. Open a help channel #βο½how-to-get-help
!unmute 764471561343336490
late dewBOT
:incoming_envelope: :ok_hand: pardoned infraction mute for @sick patrol.
bright glacier
can i help
You want to help with black's development? Could you elaborate? :)
bright glacier
well all of my projects' CI is failing thanks to the click issue, this is quite annoying :)
this time it's diff-shades
fast heart
I've just tried out black for my project, it looks great. But I'm a bit confused about this change
Does it only add commas when function calls and the like are multiline? And, what's the motivation behind it?
errant barn
The idea is that for collections the trailing comma allows for easier adding of new elements or editing, since the rows are uniform.
i.e. calls, lists, dicts etc.
high drift
im sorry if this place isn't for asking help, but would you guys happen to know why im getting this error whenever i try to use black?
$ black test.py
/usr/bin/env: './python3': No such file or directory
i am on android, but using replit which runs on linux afaik
i was able to fix it with someone's help (#help-rice message) by changing the first line of this file from #!/usr/bin/env -S ./python3
but would you happen to know if this is a permanent fix or a hacky fix?
lament crow
using python -m black instead of black when running black is probably a good workaround
I'm not familiar with REPL to know what it does that causes the built entrypoints to not work
high drift
oh i see i see, that worked. thanks!
bright glacier
https://github.com/psf/black/blob/22.3.0/src/black/__init__.py#L1438 of course this is breaking diff-shades CI on main
late dewBOT
src/black/__init__.py line 1438
from click import _unicodefun```it's missing a type annotation -- the baseline revision (being 22.3.0) fails to compile now
not sure how I built the mypyc wheels without error now I think about it though
dense jungle
not sure how I built the mypyc wheels without error now I think about it though
maybe you had click 8.0.1 installed?
bright glacier
late dewBOT
mypyc-requirements.txt lines 5 to 13
# A bunch of packages for type annotation information
mypy-extensions >= 0.4.3
tomli >= 0.10.2
types-typed-ast >= 1.4.2
types-dataclasses >= 0.1.3
typing-extensions >= 3.10.0.0
typing_extensions != 3.10.0.1; python_version >= '3.10'
click == 8.0.4
platformdirs >= 2.1.0```I used a spell checker and read over my writing a few times over, hopefully that shows :)
lament crow
Huh, I thought there's more projects on diff-shades's list
bright glacier
I just stole black-primer's list of projects and added a few of my own projects :p
lament crow
I'm sure it's millions of lines of code anyway, just didn't think it would be just 20 or so projects π
bright glacier
I'm sure it's millions of lines of code anyway, just didn't think it would be ju...
... out of 2 178 151 lines, 10 484 files & 23 projects
roughly two million :)
django and typeshed are responsible for a sizable portion IIRC, along with pillow or pandas?
bright glacier
I can't remember Β―_(γ)_/Β―
https://github.com/psf/black/pull/2996#issuecomment-1089417251 I like this comment, it makes me happy to see it.
Could be further improved, like including a breakdown of the changes per project or whatever (hidden in a dropdown), but it's good as it is.
hmm, actually that feature would've probably saved jpy-git some work
gosh it's been a while since I've last ran this script. haven't used it much ever since I made https://ichard26.github.io/ghstats/
bright glacier
@dense jungle btw could you update the stable tag? I don't have access to my main Linux machine right now (I'm stuck waiting for windows updates to finish)
bright glacier
whee thank you!
dense jungle
pip install black
There are extensions for various IDEs available
It's part of the standard Python extension, so there's no need for a Black-specific one
you have to enable it
the instructions linked in the documentation you quoted
bright glacier
well it formats your code, so the most the integration will do is make it easier to run black (by adding a shortcut or even automatically running it)
stuck vapor
<@633587720270381077> could you take a look into this? installing ipython from g...
sorry just saw this - will take a look
stuck vapor
sorry just saw this - will take a look
In the latest release of IPython:
>>> TransformerManager().transform_cell('a = b??')
"get_ipython().set_next_input('a = b');get_ipython().run_line_magic('pinfo2', 'b')\n"
In the development version:
>>> TransformerManager().transform_cell('a = b??')
"get_ipython().run_line_magic('pinfo2', 'b')\n"
The latest version looks more correct to me, because the a = part will be ignored anyway if you type that line into an IPython shell:
In [2]: object??
Init signature: object()
Docstring:
The base class of the class hierarchy.
When called, it accepts no arguments and returns a new featureless
instance that has no instance attributes and cannot be given any.
Type: type
Subclasses: type, weakref, weakcallableproxy, weakproxy, int, bytearray, bytes, list, NoneType, NotImplementedType, ...
In [3]: a = object??
Init signature: object()
Docstring:
The base class of the class hierarchy.
When called, it accepts no arguments and returns a new featureless
instance that has no instance attributes and cannot be given any.
Type: type
Subclasses: type, weakref, weakcallableproxy, weakproxy, int, bytearray, bytes, list, NoneType, NotImplementedType, ...
So I think I'd suggest to update the test.
What's the process for dealing with dev versions of dependencies? Should one wait until the next IPython release to update the test?
bright glacier
currently working on upstreaming the mypyc workflow, hopefully this doesn't crash
bright glacier
and I managed to mess twice, once by forgetting a few changes and twice by using the wrong commit message initially
flat krakenBOT
plush zinc
Hi, I am trying to install black and use in Vscode Jupter notebook using pip install black[jupyter]. But, I get an error : zsh: no matches found: black[jupyter]
mint barn
How can I make a command to be only used in a server
Not in other servers
Like id: 123456
And the command should be used in the server with this id only
flat krakenBOT
dense jungle
Hi, I am trying to install black and use in Vscode Jupter notebook using `pip in...
Sounds like your shell is trying to do command expansion. Try quoting the part with the square brackets: pip install "black[jupyter]"
flat krakenBOT
modest citrus
You have to select black in the "Provider" list. After that, you can do Ctrl+Shift+I on Linux or Alt+Shift+F on Windows to format the current file.
flat krakenBOT
bright glacier
those are touching different files @errant barn
errant barn
...... I'm
flat krakenBOT
bright glacier
haha at least you closed Jelle's PR :p
errant barn
fk, at least someone is awake
bright glacier
fk, at least someone is awake
it's only 5:17 PM over here on the better other side of the pond
errant barn
I think I'll sleep now
bright glacier
good call :)
lament crow
do single quotes work in Windows
bright glacier
I have no idea Β―_(γ)_/Β―
silent apex
do single quotes work in Windows
just tested, it works in powershell but not cmd
bright glacier
well that's fun.
dense jungle
just tested, it works in powershell but not cmd
oh good catch, do double quotes work in both?
lament crow
ye
silent apex
yes
lament crow
and they don't expand in zsh
which I actually find a bit weird considering how other things expand in double quotes but I'll take it
bright glacier
yay for os tridiversity
dense jungle
ok thanks for checking! someone wants to make another PR? π
lament crow
I actually ran into that problem with extras previously and had to update docs in a different project so I was guessing there must have been something about single quotes that isn't going to work considering the docs in that project are using double quotes lol
unique mulch
In regards to method chaining, should black try to automatically reformat this into multiple lines without the user having to explicitly add parentheses.
From:
# Exceeds 88 character limit
apple = Fruit()
apple.eat().remove_seed().plant_seed().eat().remove_seed().plant_seed().eat().remove_seed().plant_seed()
To:
apple = Fruit()
(
apple.eat()
.remove_seed()
.plant_seed()
.eat()
.remove_seed()
.plant_seed()
.eat()
.remove_seed()
.plant_seed()
)
I know this is a bad example
silent apex
I believe black does not do things like that for AST safety but I am not 100% sure
lament crow
black does the ORM thing (fluent interface) inside parens but you have to opt-in I guess
I don't think it's typical to split like this in non-fluent interfaces
dense jungle
I *believe* black does not do things like that for AST safety but I am not 100% ...
this doesn't change the AST
lament crow
so I feel like doing reformatting in such cases would look weird
gilded sable
There is a --preview flag or something like that to enable changing the AST I believe
not exactly sure
errant barn
There is a `--preview` flag or something like that to enable changing the AST I ...
It's for switching on experimental style changes, not an AST toggle
errant barn
hmm, I'm not 100% but we try to keep AST changes very minimal so I'd be surprised
although if the proposed docstring changes land it would be another case
gilded sable
okay thanks for the information
gilded sable
it does have something that changes ast right?
I remember I saw something like that mentioned somewhere
errant barn
some conversation about an AST toggle in psf/black#2150 though if you're interested!
toxic stormBOT
gilded sable
some conversation about an AST toggle in psf/black#2150 though if you're interes...
thanks
mint barn
black formatter
flat krakenBOT
bright glacier
oh I got it from <https://stackoverflow.com/questions/63664514/python-black-does...
hmmm I don't know what I was thinking saying that
IIRC I was the person that SO answer quoted
dreamy plume
what is this about
fast heart
How's the progress with this?
I plan on having an ast, and wanting to output it using black
dense jungle
not much has happened for a while
bright glacier
do you mean .*pycache.*? Black uses regexes not globs
tawdry sorrel
Hello
flat krakenBOT
bright glacier
!remind list
late dewBOT
bright glacier
we don't even run tests on python 3.11-dev
flat krakenBOT
granite stirrup
dense jungle
we don't even run tests on python 3.11-dev
the blib2to3 change should still work though
bright glacier
Cross-platform. This is off topic for this channel though :)
Or OS-independent.
Ah cool. Well anyway welcome to Python Discord! This is the #black-formatter channel where the core team and other community members discuss the Black python autoformatter.
#python-discussion is our all purpose general discussion channel. Under it we have some off topic channels and also some discussion channels ( #internals-and-peps #community-meta etc.)
https://github.com/psf/black/runs/6032901934?check_suite_focus=true ugh, Python 3.11-dev still won't work as Cython haven't done a new release yet.
queen apex
nice
flint cedar
um
this channel is for the black formatter
but nice language though, looks like C and Python's baby
wraith jacinth
this is annoying, how do i make it not annoying
elder tusk
this is annoying, how do i make it not annoying
--line-length 100
wraith jacinth
where do i input that?
silent apex
where do i input that?
You pass it in via the command line or you can put it in your pyproject.toml (or however you configure black)
bright glacier
If folks disagree that consolidation is better, please let me know! I'll add the mypyc workflow/jobs regardless where this PR goes.
it may be easier to review using git locally @dense jungle , unchanged lines (that only moved files) are visible locally
OK cool git diff main... should show the right context (after gh pr checkout 3017), I forgot how to merge diff a branch to main
Oh neat, GitHub for Mobile finally shows the workflow name prefix in the checks list UI. That kinda kills half the value of this PR though
flat krakenBOT
neon loom
@bright glacier Wow @ your large ass .yml actions
TIL a lot
How does the concurrency work?
flat krakenBOT
bright glacier
The answer is that I tested it out as much as I reasonably could on my repo
I disabled the docker upload and pypi upload bits but left everything intact and did a test release
I linked to the workflow run in the PR description @neon loom
neon loom
I must be blind. Cheers. Thatβs good enough then
neon loom
How does the concurrency work?
Any idea with this or should I be less lazy and go find docs?
bright glacier
I'll be able to explain once I'm back on my laptop, currently on my phone :p
silent apex
Any idea with this or should I be less lazy and go find docs?
afaict, the concurrency.cancel_in_progress means that something like this will happen:
- currently, PR #12345 is running the Lint workflow
- the author quickly pushes a commit to PR #12345 while the previous Lint is still running
- the old Lint immediately stops and a new Lint workflow is run
without it, the old Lint and the new Lint will run, which is probably a wasted workflow run
(I may be wrong on that, that's just my understanding of it)
bright glacier
It's exactly that 
... the run_id OR fallback is for workflows that aren't linked to a pull request, they will use an ID unique to each run
tough saffron
probably a silly question, but why does black remove the single empty line here?
if True:
foo
bar
no?
tough saffron
hum, i guess the playground is broken, then? i tried here https://black.vercel.app/
tough saffron
it works fine for me in firefox, but is broken in chrome and edge, hum
bright glacier
I'm on chrome Β―_(γ)_/Β―
tough saffron
oh god, what on earth is happening
bright glacier
I have no idea. We don't control the playground btw, JosΓ© Padilla does
tough saffron
this is funny, if i open your link, it works fine. if i open your link, and copy-paste your code in place, it works fine. but if i open your link, paste the code into notepad, copy it from there, and paste it back, it breaks
bright glacier
yup
tough saffron
i guess i'll see if they've got an issue for this
dense jungle
in inspector it shows that your payload has \r\n newlines and @bright glacier's has \n
tough saffron
yeah, I was thinking something like that was the problem. i'll make an issue i guess. i'll mention this
dense jungle
let me see if I can repro this with Black directly
doesn't repro on the command line
tough saffron
aha, thanks! i'll mention this too, then.
flat krakenBOT
tough saffron
can i also ask something else. i ran black on my tests and it did this, although the documentation seems to be saying that black likes call chains? https://black.vercel.app/?version=stable&state=_Td6WFoAAATm1rRGAgAhARYAAAB0L-Wj4AGRAOldAD2IimZxl1N_WlXnON2nzOXxnCQNGadArwI9yclOzPPmLbUxjbxC6mtuVV_e4twQ-I1pjv2w0mNjcCzcXh6X4gcvLDct8vjHtypdFJYoO3igalTFV4Cuyqu3sjr16j__A5Gb0cit769ikJS5GW1RgMvqycH9Aajuk2uQvEN4kiCbe0gcOSkzaXUb8KTpAU3mkgSC3LbuxIF3T06UhBHDMuGn9hyWYozhMwFxzcgGULUY7XtDaR4eowPYQvFJrg-f3k0ABCq5oOdehG_qBV8I8e525rsKELHORgSI9ZREYaJfQxtz50w1pocAAAAAACsLGfWQ8s--AAGFApIDAACglefiscRn-wIAAAAABFla
dense jungle
can i also ask something else. i ran black on my tests and it did this, although...
only in some circumstances, not sure we ever do it within lambdas
mint barn
black used to be amv/black for the plugin in vim what is it now?
mint barn
wait
how do i get the latest version of black (i think it came with PEP 9 or smthn?) for neovim?
dense jungle
how do i get the latest version of black (i think it came with PEP 9 or smthn?) ...
I don't use it myself, but check out https://black.readthedocs.io/en/stable/integrations/editors.html#vim
glad pilot
If I suspect mypyc is generating invalid C code for Black, how do I prove that? How can I see the intermediate .c files that get built by mypyc in the process of creating a wheel for Black?
bright glacier
they're not stored anywhere in the built wheel AFAIK
glad pilot
(I'm trying to track down https://github.com/psf/black/issues/2845 - the segfault is uninteresting, but the assertion error is interesting)
bright glacier
the C output is stable though so I could probably generate it locally looking at the build logs
glad pilot
looks like the problematic type is StringMerger, and it looks like mypyc is generating code that sets tp_vectorcall_offset and Py_TPFLAGS_HAVE_VECTORCALL without setting tp_call, which would be wrong.
bright glacier
OK, according to the logs the 22.3.0 wheels used mypy==0.910
glad pilot
I'm familiar with the C API, but I know approximately nothing about mypyc - but that assertion error certainly seems to indicate that the code it's generating is wrong...
bright glacier
glad pilot
ah! Looks like the class is inheriting __call__ from its superclass. That's probably the test case here...
bright glacier
If I suspect mypyc is generating invalid C code for Black, how do I prove that? ...
assuming you don't care what version of mypy / black you're using, simply running python setup.py --with-mypyc bdist_wheel and stopping it once it hits running bdist_wheel will allow you dig into build/ for the __native*.c file
you might need some extra packages for give mypy more type information, feel free to use this list: https://github.com/ichard26/black-mypyc-wheels/blob/9bd31a4592455d91cb6a505baa8dce6b29e3d21e/mypyc-requirements.txt#L5-L13
late dewBOT
mypyc-requirements.txt lines 5 to 13
# A bunch of packages for type annotation information
mypy-extensions >= 0.4.3
tomli >= 0.10.2
types-typed-ast >= 1.4.2
types-dataclasses >= 0.1.3
typing-extensions >= 3.10.0.0
typing_extensions != 3.10.0.1; python_version >= '3.10'
click == 8.0.4
platformdirs >= 2.1.0```(I'll note that mypy beyond 0.910 generates very broken code currently for black so you'll want to use that if you're compiling black)
glad pilot
ack - I'm gonna try messing around without Black and seeing if I can reproduce this for a minute, now that I think I've spotted what's interesting about what Black is doing...
yep! It's pretty trivial to reproduce now that I guessed what's going wrong.
$ cat test.py
class Base:
def __call__(self) -> None:
return None
class Derived(Base):
pass
$ mypyc test.py >/dev/null
$ python3-dbg -c 'import test'
../Objects/typeobject.c:5392: PyType_Ready: Assertion "type->tp_call != ((void *)0)" failed
Enable tracemalloc to get the memory block allocation traceback
object address : 0xc518a0
object refcount : 1
object type : 0x97cbe0
object type name: type
object repr : zsh: segmentation fault (core dumped) python3-dbg -c 'import test'
the segfault is uninteresting (it's trying to print an object that hasn't yet been fully initialized, and it's accessing a field that hasn't yet been set), but the assertion error is either a bug in CPython, or a bug in mypyc
bright glacier
thank you! @glad pilot
glad pilot
It's selfish - I keep bumping into this because I keep installing my own modules into a venv made by a debug interpreter to test it out, and then forgetting I'm using a debug interpreter, installing my linters, and segfaults π
bright glacier
I mean, it's probably just black segfaulting
... well hmm mypy is compiled itself but I haven't seen any reports of that
glad pilot
the assertion error message didn't quite give enough context to see what was happening, but it was easy enough to figure out once I applied some gdb
FWIW, I highly recommend testing C extension modules with a debug build of the interpreter
it catches lots of subtle bugs - if you're going to be including mypyc-generated stuff, running python3-dbg in your test suite would be a good idea
bright glacier
Hmm, how would that work with cibuildwheel?
glad pilot
absolutely no idea. π
bright glacier
ah, well then I'll figure it out eventually
probably worth adding a mypyc build CI job for pushes to main
diff-shades (runs on pushes to main and PRs) tests mypyc indirectly but it doesn't use a debug build of cpython
glad pilot
I'm completely new to github actions - looks like Pablo didn't set up anything using -dbg interpreters for memray, so... π€·
Maybe it matters more when you're writing your extension code by hand, and less when mypyc is generating it for you, though.
bright glacier
I'm pretty sure mypyc is still technically alpha even though black and mypy are both using it π
glad pilot
I believe so, yeah.
stuck vapor
Has it already been tried to get black to use mypyc when it's run as a pre-commit hook?
dense jungle
Has it already been tried to get `black` to use `mypyc` when it's run as a pre-c...
why wouldn't it use mypyc then?
bright glacier
it's annoying since pre-commit clones black from github directly avoiding the compiled wheels. This is also why --required-version breaks with pre-commit :/
stuck vapor
Timing from running on the pandas codebase:
With standard pre-commit hook, i.e.
- repo: https://github.com/psf/black
rev: 22.3.0
hooks:
- id: black
$ rm -rf ~/.cache/black/; time pre-commit run black --all-files
black....................................................................Passed
real 0m31.356s
user 5m33.922s
sys 0m1.585s
With local hook, i.e.:
- repo: local
hooks:
- id: black
name: black
entry: black
types_or: [python, pyi]
language: python
additional_dependencies: [black==22.3.0]
$ rm -rf ~/.cache/black/; time pre-commit run black --all-files
black....................................................................Passed
real 0m14.093s
user 2m29.339s
sys 0m3.660s
Problem is, trying to get it to use mypyc when installing as part of the pre-commit hook results in
File "setup.py", line 42, in <module>
from mypyc.build import mypycify
ModuleNotFoundError: No module named 'mypyc'
Do you know if there's a package on PyPI which contains everything necessary to run mypyc?
bright glacier
you'd need to install mypy and a bunch of dependencies (or their types- counterparts) so mypy[c] has enough type information
I should probably document this somewhere in the development docs.
stuck vapor
Thanks - just tried with
additional_dependencies: [
mypy,
mypy-extensions,
tomli,
types-typed-ast,
types-dataclasses,
typing-extensions,
click,
platformdirs,
'setuptools-scm[toml]',
wheel
]
but still get
from mypyc.build import mypycify
ModuleNotFoundError: No module named 'mypyc'
that's because it's a build time dependency (along with the others)
when pre-commit shells out to pip to install the local clone, the [build-system] table in pyproject.toml will only have setuptools, setuptools-scm, wheel available
stuck vapor
π€ reckon there's some other way around this then?
bright glacier
the cleanest way would probably be to use a shim repository that installs black from pypi or builds black manually, but that seems untenable
stuck vapor
the cleanest way would probably be to use a shim repository that installs black ...
might go with a local hook then, with as many files as pandas has it makes quite a noticeable difference
bright glacier
the cleanest way would probably be to use a shim repository that installs black ...
Perhaps we could make another repo that does this, but it wouldn't be great being able to use different repos as the source
python packaging overall doesn't support conditional build time dependencies very well, aiohttp has to deal with this too
dense jungle
does mypy itself deal with this when it is used as a pre-commit hook?
bright glacier
I have no idea but everyone uses it as a hook via another repository so in theory they could do some fancy setup.py logic to workaround this
well that's certainly one way to use the compiled wheels in pre-commit
it seems like these mirrors are automated which is neat. I'd imagine they don't support specifying commit SHAs, but that's fair.
stuck vapor
https://github.com/psf/black/blob/main/.github/mypyc-requirements.txt
adding these to pyproject.toml , I no longer get that ModuleNotFoundError, now it's just
Parsed and typechecked in 2.211s
Compiled to C in 0.000s
src/black/parsing.py:30:1: error: Library stubs not installed for "typed_ast" (or incompatible with Python 3.8)
src/black/parsing.py:30:1: note: Hint: "python3 -m pip install types-typed-ast"
src/black/parsing.py:30:1: note: (or run "mypy --install-types" to install all missing stub packages)
src/black/parsing.py:30:1: note: See https://mypy.readthedocs.io/en/stable/running_mypy.html#missing-imports
src/black/parsing.py:30:5: error: Name "ast3" already defined on line 25
which is odd because types-typed-ast is definitely in there π€
bright glacier
that's strange :/
stuck vapor
that's strange :/
it's a different error to the one before though, so it's progress π π
bright glacier
i'll quickly note that IIRC gcc doesn't work with mypycified black on linux btw, you'll want to use clang
I don't know where you are in the process but I'd like to highlight that speed bump ahead of time
flat krakenBOT
fast heart
Black has negatively affected the neatness of the code I write π
I just write dirty code out of laziness and run black
errant barn
But with less effort I presume!
mint barn
flat krakenBOT
bright glacier
we might have to file a bug report for myst-parser honestly, I couldn't figure it out either when I last tried π¦
the issue was windows specific IIRC
!remind 21H look into the options PR
late dewBOT
dense jungle
maybe we should just remove the links for now
flat krakenBOT
vapid citrus
Hey!!
I recently installed black on my arch system. But for some reason it's not working. It's a environment variable problem ig.
I have tried setting path like this:
set PATH /home/name/.local/bin
It works but then other unix commands won't work.
Important : The above solution (sort of) only works in fish. (shell)
Any Solutions for this?
I am not much familiar with environment variables so please don't judge me
dense jungle
Hey!!
I recently installed black on my arch system. But for some reason it's not...
generally you want to only add things to $PATH, not replace it completely
something like set PATH $PATH:/some/new/path
vapid citrus
I was also thinking the same as setting the path was disabling the other commands
set PATH $PATH:/home/xxx/.local/bin
This should work right?
dense jungle
I was also thinking the same as setting the path was disabling the other command...
yes. note that putting it at the end makes it lowest priority, so if there a black in an earlier PATH entry it will be preferred
flat krakenBOT
late dewBOT
!remind 21H look into the options PR
It has arrived!
ancient shoal
Can black-formatter format my casing style?
I used it, and it was great, but I have some very bad, inconsistent naming
errant barn
meaning variable naming?
ancient shoal
yes
errant barn
unfortunately not, that would change the AST, which we try to avoid
ancient shoal
I have functions defined in PascalCase, camelCase and snake_case, and lowercase in the project
errant barn
but a tool like that would be cool!
ancient shoal
thanks, I guess I'll have to do it manually
but I have to say, it's really fast and convenient
errant barn
or spend 10x the time and automate ;D
ancient shoal
showed me syntax I didn't even know existed
errant barn
much appreciated ππΏ
lament crow
pylint can at least let you know about such issues in your code
But I don't know of anything that would automatically change code to adhere to such convention, seems dangerous :P
pure wolf
I just encountered a bug in on of my project, that was introduced by black: I have a dict variable where one element on a line was deemed too long by black so it split the value, a string, in two. But the string is an f-string and when split, black does not put an f-string on the second line and it happens a {} is on that line. The value is then broken at evaluation
# before black
row_attrs = {
"onclick": lambda record: f"window.location='{reverse('organization_follow_up_detail_view', args=(record.pk,))}';",
}
# after black
row_attrs = {
"onclick": lambda record: f"window.location="
"'{reverse('organization_follow_up_detail_view', args=(record.pk,))}';",
}
It thus broke the lambda. Is that addressed in most recent versions of black or is it a current known bug? Otherwise, I would be more than happy to open an issue.
dense jungle
I just encountered a bug in on of my project, that was introduced by black: I ha...
It's unlikely we ever had that bug. Can you reproduce it running Black directly on a file?
pure wolf
I'm trying right now but I can't reproduce it somehow. Black is run many times a day against the code locally and in a pipeline. I wish I could pinpoint the moment and what triggered that kind of choice. I'll keep trying some scenarios, otherwise that will hard for you to diagnose that so I'll just go with it and I'll open an issue if it ever happens again.
dense jungle
Sure. The reason I say it's unlikely is that we have AST safety checks that should make Black crash if it ever made that change. Those have been there since the beginning. It's slightly more likely that this happened if you run with the --fast and --experimental-string-processing flags.
pure wolf
Gotcha. However I've run Black without any flag, inside a docker container docker-compose run --rm django black .
Thank you for your help and taking time to answer me.
little parrot
This isn't the right channel to get general help with python, see #βο½how-to-get-help
flat krakenBOT
silent apex
^ additionally, you often order code to keep related functions together, and black can't know that
dense jungle
^ additionally, you often order code to keep related functions together, and bla...
agree. and do we order classes and functions separately? and do we do AST analysis to make sure we don't put child classes before base classes?
flat krakenBOT
flat krakenBOT
bright glacier
for context, code-review-doctor would actually break gallery if it PRed this suggested change
I know automating deep code linting is hard, but I am a bit suspicious of any service that claims it can detect logical issues.
Though, looking through its PR history, this service doesn't seem half-bad, huh
flat krakenBOT
upper dirge
?
flat krakenBOT
flat krakenBOT
flat krakenBOT
static iron
Black Lives Matter guys
mint barn
π§’
late dewBOT
It has arrived!
bright glacier
!remind 5H merge https://github.com/psf/black/pull/2525
late dewBOT
bright glacier
There's some code I need to check manually and that coverage drop on PyPy is still super odd, but overall the PR looks OK to me.
silent apex
blib2to3 iirc, a fork a lib2to3
bright glacier
!remind 5H merge https://github.com/psf/black/pull/2525
actually, checking this PR by hand nets me even more questions
gah, why is everything more complicated than it initially appears :)
dense jungle
blib2to3 iirc, a fork a lib2to3
that's right
I just realized that we have a README stating why we forked blib2to3 (cc @bright glacier since we were talking about this around your blog post). The initial reasons were ```+Reasons for forking:
+- consistent handling of f-strings for users of Python < 3.6.2
+- better ability to debug
+- ability to Cythonize
oh lol that's fantastic
dense jungle
now there's a few more in https://github.com/psf/black/blob/main/src/blib2to3/README
plain atlas
neat
i'm currently using libcst
although it depends on how long they take to fix this bug
dense jungle
don't use blib2to3
bright glacier
LibCST will be 1000x better to work with than blib2to3
dense jungle
it's the most painful parse tree I've ever used
silent apex
libcst is cool but I kinda wish it was untyped (not that that's relevant to most people, or even means anything)
plain atlas
why??? @silent apex
silent apex
I'm personally a huge fan of untyped trees with typed layers over them, I like the ease of manipulation and the uniform layers
dense jungle
I guess the f-string reference in the original blib2to3 docs is https://github.com/python/cpython/commit/e8412e684ed741df246e8430f4911b31b0d8be1f
bright glacier
thanks for reminding me of that README Jelle!
dense jungle
speaking of random stuff in the repo, maybe we should get rid of https://github.com/psf/black/tree/main/profiling. It's sometimes annoying when I git grep
late dewBOT
!remind 5H merge https://github.com/psf/black/pull/2525
It has arrived!
echo pebble
Is it possible to run black on a file and after that the file to throw errors?
errant barn
Is it possible to run `black` on a file and after that the file to throw errors?
What kind of errors? Black won't parse invalid Python files
Or for Black to destroy the previously valid code?
echo pebble
What kind of errors? Black won't parse invalid Python files
I have not encountered any error. I was just curious. Sorry if i wasn't clear.
errant barn
No worries π just wanted to dig deeper
And the other answer (to Black destroying a valid file) is that it's a bug
And yet another answer is that Black naturally can't validate the runtime behavior, so while 1 / 0 is formatted correctly, it will still produce a div by 0 error
echo pebble
Thanks for the explanation
errant barn
what. <@&831776746206265384> ^
molten hare
!cban 943359125197324339 we aren't interested in your trolling
late dewBOT
:incoming_envelope: :ok_hand: applied ban to @sand mason permanently.
errant barn
thaaaank you! <3
flat krakenBOT
flat krakenBOT
flat krakenBOT
flat krakenBOT
silent apex
@mint barn Hi, please use #bot-commands for testing the bots. I've removed your messages as to not clutter this channel.
swift mauve
Hi, I was wondering if Black respects when a file uses 2 spaces for indentation instead of 4 spaces, and if it can be configured for using 2 spaces? I have some MicroPython projects that I'd like to use Black on, and I read that 2 spaces should be used instead of the usual 4 spaces to fit more code into memory. I intend to continue to use 4 spaces for my non-MicroPython projects.
Thanks!
Source: https://randomnerdtutorials.com/micropython-programming-basics-esp32-esp8266/
dense jungle
Hi, I was wondering if Black respects when a file uses 2 spaces for indentation ...
No, we don't support that. Black is meant to create readable code, not save bytes.
If you need to minimize the number of bytes in your program, I'd recommend using a different tool that minifies the code.
swift mauve
If you need to minimize the number of bytes in your program, I'd recommend using...
Okay thanks for the advice! I'll try to add a minifier as part of the provisioning process so that the code can be written in 4-space-indentation and only minified right before flashing the microcontroller. Thanks again!
dense jungle
Okay thanks for the advice! I'll try to add a minifier as part of the provisioni...
Great! Of course there's a lot of other things you can also do in a minifier, like stripping comments and docstrings. Not sure if there's an existing tool for that.
Also, I'd be wary here that you're doing premature optimization. If you don't really need those few extra bytes, you're just complicating your deployment pipeline
dense jungle
this might be interesting for us: https://github.com/ofek/hatch-mypyc
neon loom
Hatch is going through the PyPA adoption vote at the moment fwiw
neon loom
I wonder if we can also use cibuildwheel
(If weβre not - didnβt look at Richardβs final automation)
dense jungle
cibuildwheel is awesome
neon loom
@bright glacier Did you look at using it at all for you wheel building?
bright glacier
The workflow is using it!
neon loom
Is it not merged yet? cooper-mbp:.github cooper$ grep -R cibuildwheel *
Shows nothing - Was looking before i asked to see how you did it
O right, I have to look at your repo don't I
O, you went envars rather than config - Would you be opposed if I move to config so easier to try build the wheels locally - Just trying to learn
late dewBOT
It has arrived!
bright glacier
@neon loom go ahead if you means you get to play with modern packaging and distribution conventions and techniques ^^
neon loom
Thank you.
neon loom
O I can maybe get us 3.11 CI if we need it. Or you talking about building wheels?
This is cool to see - same fuzzing (same data samples + size) is 2 seconds quicker in 3.11 vs. 3.10: https://pastebin.com/JAQTyx7Z
bright glacier
I have a branch already, I just need to retrigger CI on it- errm nevermind
neon loom
Ahhh ok. Thought so
bright glacier
cython still hasn't made a new release and that blocks some of our deps
bright glacier
yeah cython main supports 3.11-dev, but their release schedule is not fixed or predictable
not sure how I'd hack tox to install a build-time dependency from GitHub
neon loom
yeah cython main supports 3.11-dev, but their release schedule is not fixed or p...
Yeah I donβt think they have many maintainers right ?
neon loom
It looks like scoder is still maintaining it
Yeah - The poor 1 person
neon loom
mypyc is also broken on 3.11
bummer - they have an issue or PR tracking it I take it?
flat krakenBOT
neon loom
@bright glacier or any cibuildwheel gurus - What am I doing wrong here to make cibuildwheel try build a pure python wheel - https://github.com/cooperlees/black-mypyc-wheels (latest commit) when testing locally ...
Do you have (I am prob blind) how you tested locally documented somewhere?
Also, low pri, I'm going to move onto playing with something else now and come back to this
