internals-and-peps

[tidal marten]

Writing something so massive from scratch would be futile

[foggy juniper]

Hi all, i would like to share a numpy array between multiprocessing. My intention is to take the frame after it's processsing to django view. Can anyone help on this?. i can share the numpyarray in threads, but in multiprocessing, i cant do it. Kindly help

[undone hare]

Hey, see #❓｜how-to-get-help

[quiet wasp]

hi guys, how are you? Lately I've been getting more and more curious about Domain Driven Design. Does anyone have experience with that in Python? I'm posting here since I think this is a rather advanced (or just uncommon) topic throughout Python users, but if more are interested in that maybe a #software-architecture/design channel would be worth it in the Topical Chat/Help section. However, I'd love to discuss the matter with anyone who knows about it, and maybe try and implement something together too!

[outer prism]

which cloud should I learn if i want to get into machine learning and ds?

[undone hare]

Hey guys, can you take a look at #❓｜how-to-get-help

[quiet wasp]

I'm not asking for help, though, @undone hare 🤔

[undone hare]

Well, see the channel topic, this is about the implementation and the use cases of Python in particular, I’m not 100% sure if it fits the topic, maybe you could try in #python-discussion ?

[quiet wasp]

alright! How about the creation of a topical channel by the way? I mean something like #software-architecture/design where to discuss in general how to structure projects, who may decide if this could be a nice idea @undone hare ?

[undone hare]

We did have a channel like that, but it wasn’t very much used, kf you have any suggestion about the community though, feel free to send a message in #community-meta

[quiet wasp]

thank you!

[boreal rune]

i've recently figured out how to use asyncio

is there a reason why there's no asynchronous range() function availible in the language?

[peak spoke]

What would be asynchronous in it?

[boreal rune]

it just seems strange to me that i'd have to implement something like that myself

oh well i can't do stuff like: -

WAIT NEVERMIND sorry

[spice pecan]

range just runs a simple calculation to get you an item at a certain index

[boreal umbra]

@spice pecan I didn't know range supported __getitem__--thanks!

[olive marsh]

can I get users liked pages/follower/(people followed) if I provide login using Twitter API?

[spice pecan]

It implements the Sequence ABC

[boreal umbra]

@olive marsh that question isn't on-topic for this channel.

[desert peak]

@olive marsh look at their API documentation and figure it out

@quiet wasp kind of sounds like appropriate discussion for this channel or #esoteric-python

#python-discussion seems to be focused purely on newbies and doesn't have very much fruitful conversation

anyone know if there has been work on compiling a list of 3.9-supporting frameworks?

[mortal isle]

l

[unkempt rock]

what is python if __name__ == "__main__" ?

I think I'm good enough with python but to this date I still have no idea what exactly it means, if anyone does please @

[wide shuttle]

@desert peak Thanks

[desert peak]

@unkempt rock It means if someone is invoking your script directly, like python -m app or python script.py

[unkempt rock]

or double clicking?

[desert peak]

correct

if you execute a script directly, its __name__ is __main__ instead of the module name such as import script or whatever

[unkempt rock]

ok so i understand it someone already answered it in #python-discussion but why do i see it in alot of code bases? from my understanding using it makes no sense ?

[desert peak]

because you might want your module to be executable, but you don't want the executable stuff to run on import

[unkempt rock]

what?

[desert peak]

there might be usable functions you want to import from your script, but you also want the script to run from the cli

[unkempt rock]

ohhh

i see.

[desert peak]

and without that block

[unkempt rock]

thank you very much

[desert peak]

it would run that code whenever you import it

[unkempt rock]

ye i get it now thanks, will prolly be using this lots now

[desert peak]

it's a good practice to add it so you don't accidentally execute usercode on imports

[inland acorn]

!if-name-main there is also a tag for it 😄

[fallen slate]

if __name__ == '__main__'

This is a statement that is only true if the module (your source code) it appears in is being run directly, as opposed to being imported into another module. When you run your module, the __name__ special variable is automatically set to the string '__main__'. Conversely, when you import that same module into a different one, and run that, __name__ is instead set to the filename of your module minus the .py extension.

Example

# foo.py

print('spam')

if __name__ == '__main__':
    print('eggs')

If you run the above module foo.py directly, both 'spam'and 'eggs' will be printed. Now consider this next example:

# bar.py

import foo

If you run this module named bar.py, it will execute the code in foo.py. First it will print 'spam', and then the if statement will fail, because __name__ will now be the string 'foo'.

Why would I do this?

• Your module is a library, but also has a special case where it can be run directly
• Your module is a library and you want to safeguard it against people running it directly (like what pip does)
• Your module is the main program, but has unit tests and the testing framework works by importing your module, and you want to avoid having your main code run during the test

[unkempt rock]

wow

[desert peak]

not sure I'd word it "has a special case" - it's pretty common @inland acorn

[unkempt rock]

is theres a documentation for this bot?

[desert peak]

#bot-commands

[inland acorn]

it is open-source, and there is the !help and !tags commands

not sure I'd word it "has a special case" - it's pretty common @inland acorn
@desert peak feel free to open an issue on github if you think it should be reworded

[unkempt rock]

btw

you can eval stuff using the bot

[mint forge]

!source in #bot-commands @unkempt rock

[unkempt rock]

how is this secure?

i know theres probably checks and stuff but im also sure theres a bypass since you're running untrusted code

idk

just thought i'd point it out

[inland acorn]

I am not sure of we do it exactly, but here is the project that does it: https://github.com/python-discord/snekbox

[radiant fulcrum]

It's a Jailed python instance that is limited to a set of modules

[north root]

only one person has been able to break it

[feral cedar]

storytime?

[north root]

and we've patched that already

[unkempt rock]

oh wow

[radiant fulcrum]

Its basically you Vs Google

[north root]

yep

[unkempt rock]

It's a Jailed python instance that is limited to a set of modules
@radiant fulcrum limiting the modules is useless and shouldn't be count as security

[radiant fulcrum]

I mean it is because it limits what you can do lol

[north root]

https://github.com/google/nsjail

[desert peak]

how is the bot hosted? some VPS?

[unkempt rock]

you can still rewrite the modules inside the eval'd code using python without any modules

[north root]

we're sponsored by linode

[radiant fulcrum]

linode vps yeah iirc

[desert peak]

ah, very nice

[flat gazelle]

we don't really limit the modules, do we?

[radiant fulcrum]

you get access to a certain set

not all

but most

[desert peak]

!e

import os

print(os.system('ls'))

[fallen slate]

You are not allowed to use that command here. Please use the #bot-commands channel instead.

[north root]

you can use the whole stdlib and the extra modules we've installed

[peak spoke]

There's no module restrictions, some are just useless in the restricted environment; but this seems more fitting to #community-meta

[unkempt rock]

off-topic but what does meta mean lol

[radiant fulcrum]

the actual security side of it is Down to NSJail

the discord itself and the things related to it

[unkempt rock]

hi

[radiant fulcrum]

see the topic in meta

[desert peak]

@unkempt rock meta means referring to self, such as us talking about discord and the channel itself and not Python

[unkempt rock]

ah ok

i would assume if someone is aiming to mess with the eval they wouldnt actually break the sandbox rather they'd mess with results that the bot will take and display in channel

i cant say @ everyone? i mean its disabled why i cant say it -_-

[flat gazelle]

there was an incident where the ping got quoted by someone who could

[radiant fulcrum]

its returned in a embed block and also now is escaped

[desert peak]

"an incident" LOL

[inland acorn]

we have a check for code block escapes, and the result is run thru our normal chat filters

[radiant fulcrum]

it wasnt from the eval though

Error handling on a different command went ping pong

[unkempt rock]

we have a check for code block escapes, and the result is run thru our normal chat filters
@inland acorn ye i know but there must be a bypass, im 100% sure a string wizard can bypass it

[desert peak]

Can you @ user IDs?

oof, getting off-topic here

[unkempt rock]

saying a thing or two off topic shouldnt be a crime lol otherwise you would need to end each message with "this is an advanced discussion :)"

[odd ether]

ye i know but there must be a bypass, im 100% sure a string wizard can bypass it
@unkempt rock You're encouraged to try and break out (and then tell us how you did it)

[north root]

alright, this whole conversation has been off-topic, to be fair

[unkempt rock]

how? its python related lool

[odd ether]

Yeah, if you want to discuss our bot it should be in either #community-meta or #dev-contrib

[north root]

Discussion on the use cases, implementation and future of the Python programming language including PEPs, advanced language concepts, new releases, the standard library, and the overall design of the language.

[unkempt rock]

ah ok

[desert peak]

anyone think the macro or pattern matching PEPs get adopted for py310? they seem controversial. I also want null syntax

those three things would make python much more ergonomic to me

[silk pawn]

Are there any current optimization projects for the stdlib that have an idea in mind of what to change, but the person in charge doesn't have the time to do it?

Like how a ton of built in functions were migrated to PEP 590(?) vectorcall calling conventions, but that's still going on on 3.8, 3.9, and 3.10a1 iirc, so I'm wondering if there are optimization projects that need extra manpower/womanpower to finish, but have a clear idea of what they're trying to optimize

[icy tartan]

Greetings, is there a version of multiprocessing.Pool.starmap that returns an iterator like multiprocessing.Pool.imap and multiprocessing.Pool.imap_unordered do?
It's not crucial, but my tqdm-progress bar depends on it!

[boreal umbra]

@icy tartan that would be a good question for #async-and-concurrency

[icy tartan]

@boreal umbra ah, thanks. Based on the channel name I expected it to be only about async, not all other forms of concurrency. I stand corrected now that I actually looked at the channel description 🙂

Meanwhile I worked around my issue (in not the most beautiful way) by just rewriting the function to take one argument and unpacking it on the first line.

[glass robin]

What’s the controversy about pattern matching? I understand the one over macro, and I even doubt they would be userull in python

[cloud crypt]

some think it brings more problems than it solves

anyways, I got a small design question, basically I have a classic Point class with x and y fields, with overloaded addition to support adding coordinates of points

what should I use for isinstance check there?

isinstance(other, self.__class__) or isinstance(other, __class__)?

[deft pagoda]

self.__class__ or type(self)

[unkempt rock]

whats wrog with this

[deft pagoda]

Let's pretend we're building some library, a graph library, say, but you want the library to be agnostic to whether one uses a networkx.Graph, graph_tool.Graph, or igraph.Graph underneath it all. To accomplish this you build some "universal" Graph that provides a consistent interface to all three Graphs. We don't want to litter the start of each method with a conditional that checks which module we're using -- what we'd like to do instead is set our method the first time it's called to interface with the correct module. So how best to accomplish this? This is a rough draft of what I'm thinking: https://gist.github.com/salt-die/ac6b7e75df258bdd4cff6e259ee50909 . This hack is a decorator/descriptor for each method we intend on implementing. My feeling is that there's something simpler one could be doing also involving descriptors.

[stone mauve]

import random
from random import choice
list = ['Luis', 'B', 'C']
winner = choice(list)
print(winner)

@unkempt rock you don't need to type random.choice just choice because of how you imported it

[desert peak]

@unkempt rock #❓｜how-to-get-help

[stone mauve]

Is it frowned upon to use os.system() to run shell commands? sometimes there's just an easier way to do something with the shell and it's handy to have it in the python file.

[shy saddle]

Shot in the dark... Has anyone used python to create images in GIMP?

[desert peak]

!docs subprocess.run @stone mauve

[fallen slate]

`subprocess.run`

subprocess.run(args, *, stdin=None, input=None, stdout=None, stderr=None, capture_output=False, shell=False, cwd=None, timeout=None, check=False, encoding=None, errors=None, text=None, env=None, universal_newlines=None, **other_popen_kwargs)```
Run the command described by *args*. Wait for command to complete, then return a [`CompletedProcess`](#subprocess.CompletedProcess "subprocess.CompletedProcess") instance.

The arguments shown above are merely the most common ones, described below in [Frequently Used Arguments](#frequently-used-arguments) (hence the use of keyword-only notation in the abbreviated signature). The full function signature is largely the same as that of the [`Popen`](#subprocess.Popen "subprocess.Popen") constructor - most of the arguments to this function are passed through to that interface. (*timeout*, *input*, *check*, and *capture\_output* are not.)... [read more](https://docs.python.org/3/library/subprocess.html#subprocess.run)

[stone mauve]

@desert peak that's nice to know. Most of the time though I'm just running something simple or in my local bin and usually have os imported so this seems a little heavy for what I do. Thanks for telling me about this though.

[boreal umbra]

Is this a true statement that I wrote?

when people talk about "imperative programming", "object-oriented programming", etc, they're talking about general approaches to language design that can inform the design of a given language to varying extents.

[desert peak]

🤷🏽 I'm not sure I would contrast "imperative" alongside "object-oriented" @boreal umbra

Imperative vs. Declarative are approaches to programming. One states you use the language to instruct each stage of the way, the other states you say what you want and the language figures out the implementation details

[boreal umbra]

@desert peak I'm not contrasting specific "x programming" terms, I'm referring to the idea of "x programming" in general.

[crystal kindle]

hey all

https://www.youtube.com/watch?v=oZ4ktcZDfvY

high level play thing ready for tinkering

https://drive.google.com/file/d/1j97GWWaX5QMwHqNfWXYh_zPs5xZM_mQ_/view?usp=sharing

this is win64 - prototype

hook into actions in game engine is polling

import bpy
wm = bpy.context.window_manager
value = wm.xr_session_state.get_action_state(bpy.context, "action_set_name", "action_name", "/user/hand/left#path")

[raven ridge]

Is this a true statement that I wrote?
@boreal umbra I'd agree with that.

[boreal umbra]

I remember Guido said in an interview that python isn't intended to be a functional programming language, but I don't think he or anyone else contests that functions are a core component of the language

So I interpret that statement to mean that functional programming just wasn't the paradigm that he prioritized.

[raven ridge]

"functions" and "functional programming language" are relatively unrelated

C has no functional features whatsoever, but has functions, after all

the feature that Python has that's most strongly correlated with functional programming languages is probably functions as first order objects. That's a feature that Python has, and that's rare outside of functional languages.

[sacred yew]

@crystal kindle wrong channels

don't a lot of mainstream langs have functions as first-order objects?

js has it

java and c++ have lambdas now?

[raven ridge]

Java and C++ have lambdas, but neither of those have functions as first order objects.

C++ lambdas are functors, not functions - they're class instances, not functions.

Likewise for Java, I believe.

Yeah, Java's the same, from quick googling.

C++ also lets you pass pointers to functions around - but that's not quite the same thing as passing functions around.

[sacred yew]

lua perl php swift rust go julia according to wikipedia

[coral tree]

Is PyCharm Edu just PyCharm Community with the EduTools plugin?

[gleaming rover]

Is PyCharm Edu just PyCharm Community with the EduTools plugin?
@coral tree this is a channel for language discussion; you might want to try #tools-and-devops. thanks!

[mint forge]

not the channel

[weary minnow]

how can i access the "vscode" key itself without knowing the name

in the sense like i want to access the second object inside the json

i want to get the "vscode" as the key for the json

[unkempt rock]

Hi guys i am shrivardhana,I am in 7th grade,I am very interested in python.i finished my python course last week.i am looking forward to do my first project

[mint forge]

!projects @unkempt rock and wrong channel

[fallen slate]

Kindling Projects

The Kindling projects page on Ned Batchelder's website contains a list of projects and ideas programmers can tackle to build their skills and knowledge.

[unkempt rock]

!projects @unkempt rock and wrong channel
@♦ Inheritanc-e#1090 👍

[muted crane]

I've always had this question

Why do we have PEP 8?

[amber nexus]

To keep things consistent

[undone hare]

Because standards = good

Having the same style accross every code bases is very useful

[muted crane]

You can choose to make your code more readable than to follow a standard

Less effort reading the code in my opinion

[undone hare]

Also copying code from one code base to another is really easy

[visual shadow]

You can choose to make your code more readable than to follow a standard
@muted crane the issue is, what's readable for you might not be for someone else. And that leads to fragmentation. It's a lot nicer for everyone to agree on one standard as best possible.

As always, there can be cases where you might have to deviate. One rule i commonly flout is the max characters in one line rule.

But by and large, trying to conform to a standard helps everyone

[muted crane]

So I don't have to follow it strictly

[visual shadow]

But you should as best possible.

Because just imagine this very moment. You're unwilling to learn the standard because "it's slightly different from what you prefer.". In avoiding this step, you're creating this very same problem for everyone else who now have to figure out your way of doing things.

This is why standards help. Consistency helps everyone. And if a standard itself is bad, it should be challenged accordingly.

[muted crane]

I see

[visual shadow]

I do have a suggestion to ease the transition. Most modern editors have some kind of code linter or style analysis

Just turn it on, and eventually your warning messages will annoy you enough that you'll do something about it 😅

[muted crane]

I heard of those

[undone hare]

Most IDEs can also automagically format your code

[visual shadow]

I personally didn't learn about standards when I started coding. So I had to go through this transition phase too. Just finding that setting on my editor made it super easy.

[warm tulip]

warning messages anoying

[muted crane]

It sounds harder to code with these settings in addition to the problems you face in the code itself

[undone hare]

I don’t find it harder once you’re used to those

Besides, most of them have a reason, and will help you further down the road

For example the “attribute assigment out of __init__” in PyCharm can help you catch some nasty bugs

[visual shadow]

Honestly, pep8 standards are really well thought and designed

I just didn't know about them enough to appreciate it till I had the warning messages staring me in the face.

[flat gazelle]

what is just silly is demanding sorted imports, but most linters do not do that

[warm tulip]

i think i will configure my linter to use pep8 standards

It will help me in the future, I think

[muted crane]

Yeah, might do that too

[visual shadow]

Hands down my most favourite one is white spaces around operators.

[warm tulip]

yes

[flat gazelle]

you generally have the formatter do that

[warm tulip]

that is good

that is very good

[muted crane]

It's neat, yeah

[visual shadow]

It's one I used to get wrong. It really helped code readibility. (and had the side effect of reducing my tolerance for code that doesn't have these whitespace)

[warm tulip]

what is sorted imports

that sounds harsh

is that like import sys, os

[muted crane]

I only use whitespaces, but still do imports on the same line

Same for variable assignments

[undone hare]

The line limit is worst than he sorted imports IMO

Besides, I am a weirdo and I like sorted imports

[flat gazelle]

yeah, line limit is also not all that great

[muted crane]

It's 24 columns right?

[cloud crypt]

line_limit = 100 or I won’t participate in your project :p

[flat gazelle]

80 cols, 24 would be impossible

[cloud crypt]

79

[flat gazelle]

but generally people go for 100-120

[muted crane]

Sounds better

[flat gazelle]

the only reason for 79 are old 80 wide terminals, but that is not relevant in 2020

[warm tulip]

pep8 sounds harsh

[muted crane]

Why would they put a line limit?

[flat gazelle]

overly long lines are bad indeed

[warm tulip]

they are

[muted crane]

At least the code is readable, following all other standards

[warm tulip]

very bd

[cloud crypt]

if I were reading someone's code I'd rather have to scroll in one direction only

[warm tulip]

yes

ok

[radiant garden]

I tend to keep my code thin, but sometimes pop in a one-line double-nested f-string lambda lambda double listcomp for fun

[muted crane]

Wrong channel btw

[flat gazelle]

but also, self.adults = [member for member in all_members if member.is_of(self) and member.age > 18] is really no worse than

self.adults = [
    member
    for member in all_members
    if member.is_of(self) and member.age > 18
]

[cloud crypt]

is_of? heh

[radiant garden]

yeah I never was able to get multiline x-comps to "feel" right

[brave badger]

Vertically-spaced code is both meh and alright I guess

[cloud crypt]

I’d rather put ... for ... in ... on one line

[muted crane]

It feels weird to code a list comp like this tbh

[warm tulip]

I prefer my 300 character line python programs

tbh

[cloud crypt]

since it doesn’t make much sense to give an additional line

but honestly my code standards are the weirdest

[flat gazelle]

I generally just write things and run black and isort and autoflake on it

[cloud crypt]

https://github.com/nekitdev/iters.py/blob/master/iters/async_utils.py this is kinda style I’m writing in

[deft pagoda]

i'll give you a gross pep8 violation: https://gist.github.com/salt-die/28ff2564f770b0168af6bbfab8747d5a#file-kivy_custom_console-py-L111

[grave jolt]

I would say that generally, shorter lines are better, but breaking a line into multiple lines just to satisfy the line length may reduce readability

[warm tulip]

what does it do

[flat gazelle]

that is pretty

but yeah, pep8 does not like aligned :

[warm tulip]

Terminal was blocked on my school's iMac, so I used AppleScript to execute shell scripts instead

lmao

[radiant garden]

pep8 also doesn't like this, and I have no idea why https://gist.github.com/RocketRace/1672fde83395bfcfea8a50da3aa20676

[warm tulip]

no reason

i can see none

flawless code

[grave jolt]

but yeah, pep8 does not like aligned :
Even though CPython's source code does use alignment like that
and often they're screwed up in some place

[cloud crypt]

CPython's STD is non-PEP8 as hell

[undone hare]

CPython is pretty much not PEP8 compliant haha

[warm tulip]

lmfao

[flat gazelle]

well, it does predate pep8 and changing it is not exactly easy

[radiant garden]

hey, it gets the whitespace right

[flat gazelle]

supposedly the PRs often introduce bugs

[warm tulip]

ah

[cloud crypt]

hold up

imagine

[warm tulip]

or they are too lazy

[cloud crypt]

black on Lib/

lul

[undone hare]

It would be... fun haha

[cloud crypt]

I’ll fucking do it

[warm tulip]

i once made a fake "Downloading Windows" program that made a progress bar appeared and my computer lab teacher got spooked

[muted crane]

Wait what does black do?

[flat gazelle]

ast comparison taking hours

[muted crane]

Enforce pep8?

[undone hare]

Yes arven

And yeah, it will probably run for hours haha

[warm tulip]

pep8 is great (don't send hate!)

[cloud crypt]

I doubt compiling rust will be faster than running flake8 on library

[warm tulip]

ryme

rime

[cloud crypt]

also akarys I’m working on braces.py again lol

[flat gazelle]

being faster than compiling rust is a really low bar

[warm tulip]

what is ast comparison

is that like a render on steroids

[cloud crypt]

no?

[flat gazelle]

black, in order to ensure it did not screw anything up, checks if the ast of what it outputs is the same as the input ast

[cloud crypt]

it basically checks if formatted code produces same AST

[warm tulip]

checksum?

[cloud crypt]

as unformatted code

no

[warm tulip]

oh

[cloud crypt]

AST comparison is recursive soooooo

[warm tulip]

oh

oh no

thats gonna take a while

[flat gazelle]

you don't really need recursion to compare trees, but IDK if black uses recursion

[cloud crypt]

what is STD at, 250k lines?

[muted crane]

250k lines???!!

[warm tulip]

thats a lot of CODE

how long did that tak

[grave jolt]

Long-standing question from me: doesn't automatically PEP8'ing code break PEP8? Because of the "A Foolish Consistency is the Hobgoblin of Little Minds" section

[muted crane]

Enforcing pep8 into code can be breaking as far as I concern

[undone hare]

also akarys I’m working on braces.py again lol
@cloud crypt oh no

[flat gazelle]

yes, it can, which is why you write and test with pep8 already enforced

[warm tulip]

uhhh

uhhh

oh no

[cloud crypt]

C:\Users\nekit\OneDrive\Desktop\coding\cpython>black Lib --line-length=79```

oh no

lol

[warm tulip]

what have you done

[cloud crypt]

FUCK it'll do site-packages as well

[muted crane]

Lol

[cloud crypt]

I am at /http/ already

[flat gazelle]

well, see you in a few hours

[muted crane]

That might take a while

[cloud crypt]

yes

[warm tulip]

ok i go goodbye

see y

a

[muted crane]

^ pep8 in action right here

[cloud crypt]

/test/ already

[muted crane]

post traumatic pep8 disorder

Why did you do it though?

[undone hare]

test will take a while

[cloud crypt]

error: cannot format C:\Users\nekit\OneDrive\Desktop\coding\cpython\Lib\test\bad_coding.py: unknown encoding: uft-8
error: cannot format C:\Users\nekit\OneDrive\Desktop\coding\cpython\Lib\test\bad_coding2.py: encoding problem: utf-8``` haha

funny

[undone hare]

Haha

[muted crane]

Did someone challenge you to use black on CPython libs?

[cloud crypt]

no

I did it anyway

[muted crane]

Oh good god

[cloud crypt]

almost the end of the /test/

graphlib is apparently PEP8-compliant

1590 files reformatted, 117 files left unchanged, 7 files failed to reformat.```

[undone hare]

Is it done?

[cloud crypt]

yes.

I got a nice laptop so

[undone hare]

git diff | wc -l now

[cloud crypt]

windows

[undone hare]

F

install GNU utils

[cloud crypt]

OH MY FUCKING GOD

1590 files changed, 277192 insertions(+), 192645 deletions(-)```

[undone hare]

Hahaha

[cloud crypt]

[undone hare]

+8k lines

[brave badger]

I do not like that ratio

[cloud crypt]

now

I'm gonna run flake8

aifc, argparse and ast are only ones broken

and abc

also someone used bare except

[undone hare]

Now python -m test

[cloud crypt]

oh boy

how do I use threads there

[undone hare]

make test -j4 maybe?

[cloud crypt]

test_dict -- test_opcodes failed``` huh

AH

I know why

imagine basing your tests off of non-PEP8-compliant code

there are more but I see

[unkempt rock]

nekit is exposing the spaghetti toothpicks that python relies on

never expose the masters like this nekit!!!

[tough narwhal]

Someone here who knows, how to use pyautogui?

[desert peak]

@tough narwhal #❓｜how-to-get-help

[tough narwhal]

thnx

[south kelp]

Hello

[desert peak]

RE: formatting CPython - I do not like Black.

I think its rules are wrong - prefer yapf

I wish these highly adopted frameworks would reach a non-beta state, though

[undone hare]

I also dislike Black's style

But it is just a matter of personal preferences

[desert peak]

then again, I prefer using tabs in my personal projects with spaces for alignment which linters hate

[undone hare]

Ah welp

[desert peak]

thankfully flake8-tabs exists

it lets me have my own preference on indentation levels without impacting others @undone hare

[undone hare]

Interesting, how does it work?

[desert peak]

you use tabs for X level of indentation, but if you have a hanging indent for parameters or something, those will be spaces.

some_fn_call(par1,
             par2,
             par3)
# ⬆ spaces

def fn():
    call_thing()
    return 5
# ⬆ tabs

I'm still figuring out how to auto-format in this way using yapf, but it's my goal. my vscode config looks like this:

        "editor.detectIndentation": false,
        "editor.insertSpaces": false,
        "editor.renderControlCharacters": true,
        "editor.renderWhitespace": "boundary",
        "editor.rulers": [80, 100],

[flat gazelle]

not sure how much I like mixing space and tab like that

oh, discord converts the tabs in the msg

https://paste.pythondiscord.com/gaviqizeqo.bash

[desert peak]

oh derp

I even explicitly inserted tabs

[heady mauve]

With multiprocessing, does one process slowing down (more data to work with) affect other processes?

[desert peak]

it can, yes

[mellow widget]

!resources

[fallen slate]

Resources

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

https://pythondiscord.com/pages/resources

[hoary shore]

How random is the random module?

[peak spoke]

It uses the Mersenne Twister random generator

[desert peak]

the answer is usually: "random enough"

[cloud crypt]

but not secure

and Mersenne Twister is meh on its own

[desert peak]

@cloud crypt I don't think security matters. People who implement their own crypto have enough knowledge to use the secure forms of entropy.

[wide shuttle]

and Mersenne Twister is meh on its own
@cloud crypt

What do you mean by that? It's used a lot in simulations and statistical modeling.

[cloud crypt]

let me find a nice table of comparison

to begin with, it has a relatively large state buffer (2.5 KiB), is not cryptographically secure, and can take a while to start generating numbers that pass randomness tests

so yeah, we can just say random enough

[wide shuttle]

Well, sure, so don't use it for crypto

What I meant was, the Mersenne Twister is still used a lot in mathematical and statistical simulations and computations

[fresh cargo]

Well, it said on its doc "The pseudo-random generators of this module should not be used for security purposes"

[wide shuttle]

There's some discussion as to whether or not we should switch to other random number generators, but your review of the Mersenne Twister seems a bit one-dimensional.

[cloud crypt]

alright, I should mention advantages, then

[sacred yew]

iirc mt is slower and requires more memory than other prngs

[cloud crypt]

It has a huge period of 2^19937, which is nice but not something to really aim for

[sacred yew]

advantages are that its widely used

[cloud crypt]

and k-dimensional equidistribution

[wide shuttle]

My simulation studies have never been limited by the large memory state or speed of the Mersenne Twister

[cloud crypt]

indeed it’s not like

[wide shuttle]

It's widely used and accepted in academic circles, and has decent properties for most simulations

[cloud crypt]

a bad PRNG

[wide shuttle]

However, there may be newer, better alternatives now

[cloud crypt]

but there are better alternatives

yeah, exactly

[wide shuttle]

Well, it's by no means said that those alternatives will become new standards like the MT is

In academia, that is

[sacred yew]

what do other mainstream languages use?

i think v8 uses xorshift128+

java uses some shitty lcg

[fresh cargo]

C++11 is also Mersenne Twister I believe

Or maybe it's one of the available ones

[desert peak]

looks like Rust's rand crate can use whatever the OS supports:
https://rust-random.github.io/book/crates.html

[sacred yew]

i think glibc uses a lcg also

[desert peak]

ah, here we go:
https://rust-random.github.io/book/guide-rngs.html

this is way nicer than I remember

[magic python]

just bothered looking up alternative approaches to setup.py bc everyone says setup.py is the old way and i can't do basic things like pip install -e if I use toml 😩

why is packaging such a shit show

[desert peak]

That is not true

But yes, packaging is a terrible wart on Python

[magic python]

oh

pyproject.toml" file was found, but editable mode currently requires a setup.py based build.
maybe i've missed something

[desert peak]

it sounds like your version of pip is outdated or you're using the wrong flags

or setuptools

``sh

[magic python]

pip is > 20, python 3.8, flags - just pip install -e <pkg> , setuptools I'm not familiar with, if it sounds like there's something obvious i should read let me know

[desert peak]

$ python -m pip install -U pip setuptools

[magic python]

already had 47.1 installed

[desert peak]

it's on v50+ now

[magic python]

😊

upgraded, doesn't work

[desert peak]

what does your pyproject look like?

[magic python]

the toml file?

[desert peak]

yes.

[magic python]

https://pastebin.com/raw/5Tg91EUG

[desert peak]

oh, you're using poetry

why are you trying to pip -e?

just do poetry install

[magic python]

why are you trying to pip -e?
no idea, just thought that's how i installed something in editable mode

[desert peak]

nope, poetry installs your package into the venv automatically

in editable mode

[magic python]

only messing around with poetry for the first time

[sacred yew]

doesn't poetry have its own method

to install into dev

[desert peak]

indeed

[magic python]

what if i want to give poetry a path?

[sacred yew]

wait doesn't poetry put it in the venv already

[magic python]

well i have a package, which has a venv, and i want to install that package in another project with a different venv

[sacred yew]

but this is the wrong hcannel

#tools-and-devops

[desert peak]

@magic python if that's the case, then you build and distribute it

[magic python]

god i hate this too

[desert peak]

poetry build -f sdist

[magic python]

i'm just going to use setup

[desert peak]

I can't help that your dev process is weird 🤷

why would you want multiple packages installed in editable mode?

[magic python]

no worries, at least setup.py does what i expect without magic i don't care if i type a few more lines

[desert peak]

EVERYTHING SETUP.PY DOES IS MAGIC LOL

[magic python]

: )

[desert peak]

You just don't want new magic

[magic python]

no i don't

[boreal umbra]

if you do pip install -U some_module>=3, and you have version 4 installed, but version 5 exists, will pip not upgrade you to 5?

[desert peak]

@boreal umbra https://www.python.org/dev/peps/pep-0440/#version-specifiers

Looks like no

[boreal umbra]

@desert peak what's the point of asking people on the internet if I have to read stuff?

jk, I didn't know about this pep. Thanks!

[static hull]

Can someone tell me if there is something like automated message on chat after joining the meeting on Teams?

[mint forge]

not the channel

[static hull]

where then?

[undone hare]

!offtopic this would be the proper way to ask

[fallen slate]

Off-topic channels

There are three off-topic channels:
• #ot0-psvm’s-eternal-disapproval
• #ot1-perplexing-regexing
• #ot2-never-nester’s-nightmare

Their names change randomly every 24 hours, but you can always find them under the OFF-TOPIC/GENERAL category in the channel list.

[thorny nimbus]

Which is better performance
zip(itertools.count,iter) or enumerate functions interms of performance

[flat gazelle]

probably enumerate

[thorny nimbus]

probably enumerate
@flat gazelle Thanks for the fast reply

[boreal umbra]

the real question is when you want to zip multiple iterables and also enumerate them

[deft pagoda]

you can zip with a count object if you don't want to wrap enumerate around it all

oh, i guess i should've read a few lines up

i would probably still enumerate even if i was zipping a other iterables

[boreal umbra]

you have to do

for i, (a, b, c) in enumerate(zip(d, e, f)): ...

kinda kills the elegance, but is still better than rangelen

[spice pecan]

I think it conveys the purpose a lot better tbh

[boreal umbra]

@spice pecan conveys it better than rangelen? yes, by a million years, because then all the items you're iterating over are stated and not just one. The parens are weird for me though.

[spice pecan]

No, as in enumerate(zip(...)) conveys the idea better than zip(itertools.count(), ...)

rangelen being doodoo is a separate topic, though I do agree wholeheartedly

[desert peak]

I mean, the parens are just unpacking

you could do for i in enumerate()

[unkempt rock]

double quotes or single quotes?

[silk coyote]

double quotes or single quotes?
@unkempt rock it only depends on the string content, if you want to include in the string a single quote, you need to use double quotes, else if you need to include a double quote you need to use single... there are really no other differences

[unkempt rock]

@silk coyote righto

[cloud crypt]

I prefer double quotes for most stuff

[spice pecan]

I do single most of the time, that feels like the "default" version to me

[cloud crypt]

the other way around here

well I preferred single previously but then

idk

[spice pecan]

I do see the merit in double quotes for consistency with other languages

[cloud crypt]

yeah that kind of thing as well

I’ve had troubles accidentally writing 'string' before

[spice pecan]

I mostly stick to single as a habit, it's consistent with repr and adds less visual noise

[deft pagoda]

i often switch between ' and " as I switch projects

[unkempt rock]

I wonder about dictionary lookups and increments, while checking for keys' existence:

substrings[substring] = substrings.get(substring, 0) + 1

I generally use the pattern above to autoincrement a value associated to a dictionary key, whether it's missing or not.

The one thing that bothers me is that I suspect it performs two lookups.

Is that actually the case?

When I am performing an operation on a mutable object I can rely on setdefault() which returns the object. I can't do that with immutables though.

In C# one uses an out variable to refer to the value one changes:

[boreal umbra]

@unkempt rock couldn't you use defaultdict?

[unkempt rock]

dict.TryGetValue(id, out var currentCount); 
dict[id] = currentCount + 1;

@unkempt rock couldn't you use defaultdict?
@boreal umbra Yes. In this case I am limited to built-ins though.

[boreal umbra]

why

[unkempt rock]

Uni assignment.

[boreal umbra]

I hate that.

[unkempt rock]

Yes, I understand. 🙂

[undone hare]

It does perform two lookups, yeah

Although using a defaultdict and a += would have to perform one lookup and one assignment (because int is unmutable), which is pretty close to a lookup in term of complexity

So it is almost the same thing

Just way too much verbose, like any uni assignment

__import__('collections').defaultdict only use builtins though

Hehe

[unkempt rock]

Thanks for the answer, @undone hare!

[undone hare]

Anytime!

[unkempt rock]

(it's likely they will check the builtins globals for anything imported, in unit tests :-))

[undone hare]

Oh god, if it is unit tested, there sure are ways to bypass it

[cloud crypt]

ah yes

[flat gazelle]

you could also just make your own defaultdict, since I don't think they will care about performance. Consider dict.fromkeys though

[cloud crypt]

__import__

[flat gazelle]

if you already know what keys you will need, dict.fromkeys(l, 0) is better than defaultdict

[unkempt rock]

Oh god, if it is unit tested, there sure are ways to bypass it
@undone hare Yes, I am almost certain. I am being an honourable schoolboy though, so far.

[undone hare]

Haha, fair enough

[flat gazelle]

you can bypass most unit tests with type('',(),dict(__eq__=lambda*a:True))()

[still hamlet]

i have a doubt in pycharm

can anyone help me

[flat gazelle]

@still hamlet ask in #tools-and-devops

[undone hare]

That's so true lak haha

[desert peak]

isn't defaultdict a builtin?

well, built into the stdlib 😉

[flat gazelle]

defaultdict is part of the standard of the standard lib, but it is not part of the builtins module

[unkempt rock]

if you already know what keys you will need, dict.fromkeys(l, 0) is better than defaultdict
@flat gazelle Thanks. I am generating keys as I go though, so can't default-add all of them. Nice thinking though.

[cloud crypt]

@flat gazelle what about raising errors

[flat gazelle]

raise type('',(Types,To,Raise),{})()
``` unfortunately, `Exception.__subclasses__()` is unviable

[fallen slate]

:incoming_envelope: :ok_hand: applied mute to @ripe cobalt until 2020-11-05 18:58 (9 minutes and 59 seconds) (reason: duplicates rule: sent 5 duplicated messages in 10s).

[spark parcel]

<@&267629731250176001>

[smoky vessel]

omg that was inapropriate

[wide shuttle]

Let me have a look

[spark parcel]

NSFW image spam

[smoky vessel]

they deleted it immeditely tho

[north root]

!ban 491716476764553216 NSFW images spam

[fallen slate]

:ok_hand: applied ban to @ripe cobalt permanently.

[spark parcel]

Anti-spam filter @smoky vessel

[smoky vessel]

ye

ok cool problem dealt with thx everyone

[delicate ermine]

any best practices for developing packages locally? I.e I'm in an enviroment where I every everything but somepackage installed. I want to test out functionality of someModule inside codeToTestPackage.py. I don't want to add sys.path

├── codeToTestPackage.py
└── somePackage
  ├── init.py
  └── moduleX
    ├── init.py
    └── someModule.py

[boreal umbra]

@delicate ermine you can have a setup.py file and use a virtual environment where the package is installed

as a side note, remember to use snake_case_instead_of_camel_case

[delicate ermine]

so I want to test out some new code and this will not be distributed to users

so if I have a setup.py

it will install locally inside that enviroment?

and then I can test?

[strange sky]

as a side note, remember to use snake_case_instead_of_camel_case
@boreal umbra why do python programmers prefer snake case, whilst pretty much every other lang uses camel case lol

[flat gazelle]

we took it from C

it is referenced in pep8 IIRC

most camel case langs take after java, since java is so popular

[boreal umbra]

@strange sky because other languages are ugly and camel case is a primary reason for that. Though if you need an objective reason, a study showed that snake-cased words are slightly faster to read.

[strange sky]

Ohh I see, thanks guys

[boreal umbra]

@delicate ermine sorry, I didn't know that you had replied. if you have a setup.py file and you pip install it with the -e flag, the project will always be available for that virtual environment

[fast oxide]

Question guys. I’m running my python project from the root directory and all the directory paths I’ve coded into my project are relative to the root directory. The problem I’m having is when I run tests in a sub directory (root-dir/tests) it errors out because the relative paths don’t work. Is their a way to change the relative path on my testing directory so it works with all the paths I’ve integrated into my project

[feral cedar]

because other languages are ugly and camel case is a primary reason for that. Though if you need an objective reason, a study showed that snake-cased words are slightly faster to read.
@boreal umbra do you have this study? sounds interesting lol

[stable grail]

I wonder if that study also takes into considderation that some languages concatinate words

[muted crane]

I never knew why camelcase syntax is so hated?

[stable grail]

like in english you write Bus driver while in norwegian you write busdriver

[flat gazelle]

I still feel like kebab-case-is-best

[stable grail]

so maybe the "how you read words" is dependant on your personal language preference

[grave jolt]

why not just

`allow spaces in identifiers`

[stable grail]

i think that would break keywords

[grave jolt]

If the identifier is enclosed in ` `, it's fine

[stable grail]

right, i thought that was a discord f-up

[grave jolt]

I think some language do that

[flat gazelle]

fortran lets you do that, kotlin too, cl too

[stable grail]

why dont you implement that and see how it works out?

[grave jolt]

sure

(not sure why that would be an issue -- other languages already do that)

[stable grail]

maybe because how englishthe syntax is

I think i mean grammar

[grave jolt]

If the identifier is enclosed within ` and `, there is no ambiguity

[stable grail]

but i would love to test a branch that has this implemented

[grave jolt]

oh, you mean, implement it in Python

[stable grail]

yes, share it, let me compile it and test it

i have changed the grammar of python before, its not very hard, it just a bit time consuming

i got help from #esoteric-python

[grave jolt]

Well, it will probably take me a whole week to find where the definition of NAME is located 😅

[stable grail]

no rush, but if you manage to do so, let me know 😄

[desert peak]

CTRL+F NAME

tf did these lemons come from anyways? I love using them, but they're weird.

[stable grail]

its because one of our owners is a lemon

and has a beard

so we make lemojis

[muted crane]

Wasn't lemon the original owner?

[stable grail]

nope

[muted crane]

Was it a team?

[stable grail]

there was joe and one more, the other one left and joe has been running it ever since with an increasing amount of staff.

also, off-topic.. so lets not continue here

[unkempt rock]

Can we put multiple image with tkinter using URL?

I tried but it seems to be working only for 1 image

[grave jolt]

Hm, changing the identifier rules might be a little bit more involved than a single line of code

just a little bit

lol

aha, it's this part https://github.com/python/cpython/blob/master/Parser/tokenizer.c#L1381-L1469

async_hacks
o...okay

[radiant fulcrum]

Nice

[grand crag]

Good to know it !

[grave jolt]

attempt 1 at adding `-ed identifiers, compiling... :pipenv:

[grave jolt]

@stable grail 😎

seems to work

I will push to my fork now

done https://github.com/decorator-factory/cpython

[stable grail]

cool.. share the fork with me and ill compile and test it tomorrow 😄

aahh. thanks

will that overwrite my already existing cpython fork?

ill just grab the zip 😄

[grave jolt]

if you fork it, then yes, probably

you can just clone it to some other place

I should probably modify it so that the backticks aren't included in the name, i.e. `hello` is the same as hello

or maybe not

oh, I might just disallow backticked names without spaces

this all probably belongs to #esoteric-python

[solar delta]

Another question: whats the point of a controller class?

outside of mvc framework i dont see the abstraction purpose

[cloud crypt]

what’s the point of a controller class where

[solar delta]

Sorry outside of the mvc use case some people use it in abstract cases for example authorizationController or something

[cloud crypt]

this question sounds pretty Java

[solar delta]

yeah its a java styled language; Dart, im trying to implement it for data authorization meaning some users have authorization of group data sets

in python though ( i am building it a flask api for sharing customer data depending if they are in the group or not )

[desert peak]

@solar delta #web-development

this channel is to talk about Python the language

[unkempt rock]

what's going on at

line 6 and 7?

@boreal umbra help me out please 😢

[raven ridge]

They're assigning an attribute named store_name to the objects.

[strong lantern]

What is a pass? anybody

[raven ridge]

If you mean the Python pass statement, it's a statement that does nothing at all.

[cloud crypt]

^

[desert peak]

docstrings are preferable to a pass

as far as statements that do nothing

[swift imp]

Why dont dataclasses allow frozen fields

like for fricking sake

[brave badger]

@stuck imp Let's not dump GIFs on an on-topic channel

@swift imp To be fair though, dataclasses already imply immutability in some form, that being the entire dataclass being frozen

[swift imp]

Yeah but maybe I don't care about hashing and I want only want certain fields to be frozen

[brave badger]

As such, having one immutable field with other mutable ones is a bit off for me

[swift imp]

I dont see how its off, its like making a read-only property except you still get all the boiler plate reduction that dataclasses provide

I can get around it with a descriptor, its just lame to have to do it

[brave badger]

Fair enough

[severe citrus]

Is it possible to de-interpret python code ?

[boreal umbra]

@severe citrus as in, figure out what the source of pyc files was?

[severe citrus]

Something link how interpreter interpreted it ? what goes behind the see ?

like equivalent C/C++ code.

[boreal umbra]

It compiles to byte code and then interprets the byte code. Were you wanting to read the byte code?

[severe citrus]

Like I want to under stand. how below code is interpreted ?

my_list =[1,8,15]
g = (x for x in my_list if my_list.count(x)>0)

[fresh cargo]

Maybe dis.dis can help

[raven ridge]

!e ```
def f():
my_list =[1,8,15]
g = (x for x in my_list if my_list.count(x)>0)
import dis
dis.dis(f)

[fallen slate]

@raven ridge :white_check_mark: Your eval job has completed with return code 0.

001 |   2           0 LOAD_CONST               1 (1)
002 |               2 LOAD_CONST               2 (8)
003 |               4 LOAD_CONST               3 (15)
004 |               6 BUILD_LIST               3
005 |               8 STORE_DEREF              0 (my_list)
006 | 
007 |   3          10 LOAD_CLOSURE             0 (my_list)
008 |              12 BUILD_TUPLE              1
009 |              14 LOAD_CONST               4 (<code object <genexpr> at 0x7f81fb83d2f0, file "<string>", line 3>)
010 |              16 LOAD_CONST               5 ('f.<locals>.<genexpr>')
011 |              18 MAKE_FUNCTION            8 (closure)
... (truncated - too many lines)

Full output: https://paste.pythondiscord.com/ejurexijed.txt

[peak spoke]

You can find the code that's ran for the instructions in https://github.com/python/cpython/blob/master/Python/ceval.c

[brave badger]

Anyone seen this yet? Interesting stuff
https://twitter.com/anthonypjshaw/status/1325023688677208065

Anthony Shaw 🐍 (@anthonypjshaw)

Full Python test suite is running and passing with .NET JIT enabled, which means I can try some third party frameworks and libraries like Django, Flask, Pytest and see what happens.

Twitter · Nov 7, 2020, 10:33 AM

[grave jolt]

will it run on Linux?

[brave badger]

will it run on Linux?
@grave jolt Hopefully™

[true ridge]

will it run on Linux?
@grave jolt AFAIK, yes. On all 3 major platforms

[grave jolt]

Windows 7, Windows 8, Windows 10

[true ridge]

Windows 7, Windows 8, Windows 10
@grave jolt oh, sorry for the confusion. I meant FreeBSD, OpenBSD and NetBSD

[undone hare]

Is the tweet about a dotNET Python implementation

[peak spoke]

Sounds like pyjion to me

[sacred yew]

https://github.com/microsoft/Pyjion

@undone hare

[swift imp]

is there a way to tell copy.deepcopy to not copy certain objects? I'm having a little issue where something about pptx.enum.shapes.PP_PLACEHOLDER isn't deep copyable and its erking me

actually its pptx.enum.base.EnumValue

[safe hedge]

Hmm not entirely sure why that would cause an issue, the source code suggests it's just an int subclass

Maybe open a help thread and add the code/output you are getting

[little plover]

Guys. Anyone of You up to check an error in a 20-line code

actually 15

[visual shadow]

Wrong room. Take up a help channel and I'm sure someone should be able to help you. See #❓｜how-to-get-help

[plain spindle]

why does this typecheck?

from __future__ import annotations

from dataclasses import dataclass

from abc import ABCMeta
from typing import *

class User: pass
class Member: pass
class Guild: pass

class BaseEvent(metaclass=ABCMeta):
    ...

T = TypeVar('T', bound=BaseEvent)
TT = TypeVar('TT', bound=Type[BaseEvent])

Check = Callable[[T], bool]

class Message(BaseEvent):
    author: Union[User, Member]
    content: Optional[str]
    ...

class GuildLeave(BaseEvent):
    guild: Guild

HandlerCallback = Callable[[T], Awaitable[None]]

@dataclass
class EventHandler(Generic[TT]):
    checks: List[Check]
    callback: HandlerCallback[TT]

class Bot:
    def listen(self, t: TT) -> Callable[[HandlerCallback], EventHandler[TT]]:
        def deco(cb: HandlerCallback) -> EventHandler[TT]:
            self.add_listener(t, cb)
            return EventHandler(checks=[], callback=cb)
        return deco

    def add_listener(self, t: TT, cb: HandlerCallback) -> None: ...

bot = Bot()
@bot.listen(Message)
async def on_message(event: GuildLeave):  # expected this to error because the type of event must be Message
    pass

[wide shuttle]

What's doing the type checking? Are you using a static type checker?

Python does not enforce your type hints and annotations on any level by itself

[brave badger]

@plain spindle Checked on mypy, doing reveal_type(bot.listen(Message)) outputs this:

main.py:49: note: Revealed type is 'def (def (Any) -> typing.Awaitable[None]) -> main.EventHandler[def () -> main.Message]'

Bit of a shot in the dark but mypy's probably treating the call to add_listener as being passed a type variable instead of a concrete type

[grave jolt]

@plain spindle HandlerCallback is a generic type alias, why are you using it as a type hint directly?

[unkempt rock]

Can you code AI with python

does AI have another language and how do you write it

[grave jolt]

"AI" is a very broad term, what do you mean?

[unkempt rock]

like making tetris

its machine learning

[grave jolt]

You can ask about that in #data-science-and-ml

[unkempt rock]

but what kind of lang does it use ?

[grave jolt]

there's no "language for machine learning"

If you want to do ML in Python, you use Python

[broken palm]

I think he means modules, but there is a huge difference between languages and modules 😂

[grave jolt]

@plain spindle

class Bot:
    def listen(self, t: TT) -> Callable[[HandlerCallback], EventHandler[TT]]:
        ...

^ this is not right if you want to pass the type of the event as the argument. Consider this:

def f(x: int) -> int:
    return x

f(int)

this is cleary wrong, right?

You have to do this:

class Bot:
    def listen(self, t: Type[TT]) -> Callable[[HandlerCallback], EventHandler[TT]]:
...

Another issue is that you're using HandlerCallaback, which is a generic type alias.
Why is the deco function even accepting a HandlerCallback? If it should accept the event type specified in the type variable, you should just do

class Bot:
    def listen(self, t: Type[TT]) -> Callable[[TT], EventHandler[TT]]:
        def deco(cb: TT) -> EventHandler[TT]:
            self.add_listener(t, cb)
            return EventHandler(checks=[], callback=cb)
        return deco

...also, you should make HandlerCallback generic on the T bound to BaseEvent, not TT bound to to Type[BaseEvent]

[plain spindle]

it's a decorator so it accepts a func

that's why it's accepting a HandlerCallback

the HandlerCallback, being an event handler, is what accepts the event type (T) when the event is dispatched, not the decorator

[broken palm]

What do you think about the future of Python and Swift, I am kinda excited to see where it goes

[gaunt osprey]

Is there something changing in Swift? I thought it's still an ios only language

I think python will still thrive in long term, but it won't be like an exponential growth and rather a steady growth. I've heard of people want to do datascience in swift but that was just speculation when i hear about it

[deft pagoda]

what do you think the best way for telling if a function is defined inside a class or not? this is one hack-y solution, but it's not entirely fool-proof: (given inside a decorator for context)

from collections import Counter
from functools import wraps
from inspect import signature

def run_n(n):
    """Decorator that allows a function or method to only run n times."""
    def deco(f):
        instance_run_counts = Counter()

        @wraps(f)
        def wrapper(*args, **kwargs):
            # determine if a function was defined in a class
            if (params := signature(f).parameters) and next(iter(params)) == 'self':
                if instance_run_counts[id(args[0])] < n:
                    instance_run_counts[id(args[0])] += 1
                    return f(*args, **kwargs)
            elif wrapper.nruns < n:
                wrapper.nruns += 1
                return f(*args, **kwargs)

        wrapper.nruns = 0
        return wrapper

    return deco

[peak spoke]

You don't really have much more information at the definition time

[deft pagoda]

probably true, i thought something might exist in inspect or similar that could tell

[peak spoke]

You could go through the frame but that's even hackier

[unkempt rock]

Anyone know the answer?

[feral cedar]

you

!rule 5

[fallen slate]

Rules

5. Do not provide or request help on projects that may break laws, breach terms of services, be considered malicious or inappropriate. Do not help with ongoing exams. Do not provide or request solutions for graded assignments, although general guidance is okay.

[unkempt rock]

I have this function and I was wondering whether there was some way to return instead of print

def fib(n):
    for i in range(len(funclist := [])+n-1):
        (i==(n-2) and print(funclist[i-1]() + funclist[i-2]())) or funclist.append(lambda i=i:funclist[i-1]() + funclist[i-2]() if i>1 else 1)

[feral cedar]

does that actually work?

[unkempt rock]

feel free to try, but you need python3.8+ due to walrus

[feral cedar]

also, i think #esoteric-python is a better place for this

[unkempt rock]

thanks, will try there

[weary minnow]

this might be a silly question but could multi-threading / multi-processing be compared to quantum superposition

because if you think about it, if i wanted to display a progress bar while my program executed a function which it downloaded 5 different files

if i wanted to display only 1 progress bar

that would mean that the progressbar exists in 5 different states which constantly change every millisecond

so if you consider this

progress_bar = Bar(random_value)
time.sleep(0.01)
progress_bar.update() # makes a new random value

this would mean that a random bar for displaying the downloads is a feasible option

because essentially, it is switching between the progress of the 5 different downloads

which are all in different states

well not exactly superposition but it seems very similar

its just on a smaller scale

ps: for those who don't know what superposition is, its a certain object existing in different states at the same time

[feral cedar]

ehh...not exactly

[weary minnow]

@feral cedar well what do you disagree with?

[feral cedar]

sure, i agree, you could make the comparison and it does make sense

[unkempt rock]

Can somebody pm me and help me with something very easy?

[weary minnow]

lmaooo @feral cedar i know the question isn't directly related to python but i need a justification to place a random progress bar for my multiple downloads

[feral cedar]

why not just have 5 progress bars though

[weary minnow]

@feral cedar well thats the issue lol i can have that but its kinda broken because it keeps on glitching due to the values continuously changing on other states of the progress bar

its kinda hard to explain but its the the terminal doesn't know what to do lol

[broken palm]

@gaunt osprey you can now use python in Swift

[cloud hamlet]

What

[boreal umbra]

I noticed that numpy arrays doesn't support chained comparisons. I assume chained comparisons for native types have a special grammar?

[undone hare]

I remember reading the EBNF grammar, chained comparisons are weirdly parsed, but I do not know more about it

[grave jolt]

a == b == c is the same as a == b and b == c

Python doesn't know the types at parse time, so grammar doesn't affect that

[undone hare]

I'm not actually sure if it is interpreted like that

[grave jolt]

!e

import dis
def f():
    return a == b == c
dis.dis(f)

  3           0 LOAD_GLOBAL              0 (a)
              2 LOAD_GLOBAL              1 (b)
              4 DUP_TOP
              6 ROT_THREE
              8 COMPARE_OP               2 (==)
             10 JUMP_IF_FALSE_OR_POP    18
             12 LOAD_GLOBAL              2 (c)
             14 COMPARE_OP               2 (==)
             16 RETURN_VALUE
        >>   18 ROT_TWO
             20 POP_TOP
             22 RETURN_VALUE

[undone hare]

Was about to do that

[grave jolt]

well, the expressions are equivalent

[undone hare]

Hmm

[grave jolt]

the bytecode might differ

!e

import dis
def f():
    return a == b and b == c
dis.dis(f)

[fallen slate]

@grave jolt :white_check_mark: Your eval job has completed with return code 0.

001 |   3           0 LOAD_GLOBAL              0 (a)
002 |               2 LOAD_GLOBAL              1 (b)
003 |               4 COMPARE_OP               2 (==)
004 |               6 JUMP_IF_FALSE_OR_POP    14
005 |               8 LOAD_GLOBAL              1 (b)
006 |              10 LOAD_GLOBAL              2 (c)
007 |              12 COMPARE_OP               2 (==)
008 |         >>   14 RETURN_VALUE

[undone hare]

Yeah, it seems to just be an optimized version

[true ridge]

I remember reading the EBNF grammar, chained comparisons are weirdly parsed, but I do not know more about it
@undone hare It is actually not much different than any separation (like x (',' x)*), the only thing that is different is the separator can be any of the comparison operators. Later on it is translated to AST as is (left is single value, and all comparators are in a list, and also all of the comparison operators).

comparison: expr (comp_op expr)*

a == a in b    
Module(
   body=[
      Expr(
         value=Compare(
            left=Name(id='a', ctx=Load()),
            ops=[
               Eq(),
               In()],
            comparators=[
               Name(id='a', ctx=Load()),
               Name(id='b', ctx=Load())]))],
   type_ignores=[])

[undone hare]

Very interesting, thanks!

[clever shoal]

Thank you for this

[unkempt rock]

What's the fastest threading?

[reef egret]

Anyone have some solid examples of a non-ORM based Repository Pattern in Python? I've got a MONSTER of an older Database.py lib that I want to move to a proper pattern based organization while I also switch over to asyncpg. Any guidance would be wondeful 🙂

[boreal umbra]

@reef egret so you're wanting to construct a wrapper around an old database?

[reef egret]

Not exactly - Just to rewrite my existing db lib setup I made (its a giant monolith of a python file)

[boreal umbra]

so what do you mean by "pattern" in this context?

[reef egret]

Repository Pattern is a design pattern

[boreal umbra]

in my experience, there's very little discussion of design patterns in the Python community.

[reef egret]

👍 gotcha

wasn't sure how much of that translated into Python

If you have say... an Animal and Person class that contained SQL statesments and operations specific to them - but you wanted them to share a DB instance - would you just send the DB instance in as a constructor param?

[boreal umbra]

There are a couple of "patterns" that I guess exist in Python. For example, I like to have class method constructors

[reef egret]

so you're not making a new connection for each of them essentially

[boreal umbra]

but I haven't heard people talk about "design patterns" for the high-level architecture of a project.

[reef egret]

interesting, Google produced a ton of results they were just all ORM based

[boreal umbra]

is that bad?

[reef egret]

nah I just cant use an ORM for speed reasons

[boreal umbra]

ah

[reef egret]

sure makes dev easier though heh

asyncpg is pretty sweet though

I'd been using psycopg2 until now

[boreal umbra]

I don't think I have anything else to contribute, but we do have a #databases channel. activity across the server will probably pick up in a few hours.

[reef egret]

Also discovered this today - https://pypi.org/project/databases/

sweet thanks @boreal umbra !

[boreal umbra]

[reef egret]

[cloud crypt]

@unborn mortar all you need to do is to update your discord.py to 1.5.x (latest) version, along with implementing intents (https://discordpy.readthedocs.io/en/latest/intents.html)

[tacit wing]

@reef egret Wouldn't it just be easier to use SQL with python?

that's what I'm using atm

took like 5-10 mins to set up

shit nvm, didn't look into that link

lol

[reef egret]

Hehe

[bronze acorn]

Any idea on how I can save the configuration I made to my gui application that I made in python
So for example, this gui has theme setting which changes themes, the default theme is dark, if i changed it to white, i want when i close the program to save this configuration, so when i get back, it should be white, and so on for the rest of the settings

[mint forge]

not the channel?

[bronze acorn]

Sorry?

[gleaming rover]

Sorry?
@bronze acorn this channel is for discussion of the language

try #❓｜how-to-get-help

[mint forge]

@grizzled lichen not the correct channel for that

[grizzled lichen]

sorry im desperate no one replied to my help thing lol

do you however?

[fossil pumice]

!warn @grizzled lichen Please do not use #internals-and-peps as a help channel.

[fallen slate]

:incoming_envelope: :ok_hand: applied warning to @grizzled lichen.

[muted crane]

Oh, so now we warn people immediately? If they don't read the description or use the channel properly?

My main discussion here was about something really cool in python

It was about decorator syntax

It looks like there's more than one way to call decorator

For example there's the pie decorator syntax, pie decorator and space

List before def syntax

Why do we really have all this syntax when we could have only one?

[brave badger]

Could you list concrete examples of that?

Not really sure what you mean by these pie decorator syntax, pie decorator and space

[muted crane]

@classmethod
 def foo(arg1,arg2): 
...

This the pie syntax

[classmethod] 
def foo(arg1,arg2):
...

This is list before def syntax

There is also another variation where the list is after the function definition

[brave badger]

huh?

[muted crane]

It forms decorators like this

def foo(arg1, arg2) [ complicated(manyArgs=1, notTooUgly='yes'), even_more_complicated(42)]:
 ...

[brave badger]

I'm fairly certain that the latter one is not decorator syntax

[muted crane]

It is

[brave badger]

How exactly?

[muted crane]

Guido made it describing the extent of decorators

[brave badger]

Mind linking where?

[muted crane]

Here

https://wiki.python.org/moin/PythonDecorators

[brave badger]

ah, I mean those are just proposals

[muted crane]

So they aren't really in the language?

[brave badger]

No

[muted crane]

And what do you think about them?

[muted crane]

I see

Would you have them in the language?

Or how would you see it?

[brave badger]

This might just be because of my bias towards the proposal that survived but I do think that the @ syntax makes it clearer what you're actually doing with the function

[muted crane]

To be honest, the one that has a list after the function definition looks really cool

I guess it would give more freedom in declaring the decorator?

With the amount of arguments it gives of course

[brave badger]

Specifically, decorator application really is just

func = deco1(deco2(func))
```it perfectly makes sense for the decorators to be outside of the function, and to have them be signified by an `@`, which can be read as `apply`

[muted crane]

So that's why we chose the one which is commonly used today

[brave badger]

You could also consider the flexibility of it, considering that any valid Python expression can be a decorator

[muted crane]

For simplicity in reading the code

I guess using the @ syntax also makes it prettier to read , in comparison to the others

I should test on decorators and see how far it can expand with this syntax

[brave badger]

!pep 614

[fallen slate]

**PEP 614 - Relaxing Grammar Restrictions On Decorators**

Link

Status

Final

Python-Version

3.9

Created

10-Feb-2020

Type

Standards Track

[brave badger]

Here's a fairly recent one

[sacred tinsel]

the [] syntax feels really strange to me, how would that work in a repl?

>>> [1, 2, 3]
[1, 2, 3]
>>> def fn(): ...
... 

[brave badger]

The proposals state on both the @ syntax and the list syntax does break on the REPL, but it seems like the latter one does break harder

[sacred tinsel]

ah true, I guess the @ needs special treatment in the repl already

it needs to wait for what the next line will be

though @deco without a following def is invalid syntax, whereas the list isn't, so I guess that makes it "worse"

[wide shuttle]

I've read that compilation document on proposed decorator syntax before, but I'm so used to the current syntax that I can't really be objective anymore

Most of the alternatives just seem less readable and/or more verbose to me, especially those trying to force it on the def line or adding a level of identation

[desert peak]

we have enough indentation problems in python :x

I build web apps, lots of with: indents

with integration_conn:
  with db_conn:
    with db_crsr:

[brazen jacinth]

with A() as X, B() as Y, C() as Z:
    do_something()

fyi

[desert peak]

@brazen jacinth would that work when one of the latter statements relies on the previous?

I always forget you can compound with statements

[radiant garden]

I assume there's also code between each statement

[desert peak]

sometimes yes, sometimes no @radiant garden

[brazen jacinth]

if there's code with each statement, i don't really see a problem

indentation merely seperates blocks

if you need one with the scope of a context manager, i don't see why that is a problem

[peak spoke]

They create the exact same bytecode

[desert peak]

!e

import sqlite3

with sqlite3.Connection() as a, a.cursor() as c:
  print(a)
  print(c)

[fallen slate]

You are not allowed to use that command here. Please use the #bot-commands channel instead.

[peak spoke]

Also got proper implicit line continuation in 3.9 for it

[desert peak]

mmm, my sqlite test is no good, it doesn't use a context manager

[true ridge]

Also, if you want to keep your code linear instead of separated across different blocks, you might want to give ExitStack a try.

[desert peak]

So I never noticed the Python logo is two snakes. That's clever.

[stone crescent]

Yea took me a while to notice as well

[desert peak]

!pep 638

[fallen slate]

**PEP 638 - Syntactic Macros**

Link

Status

Draft

Created

24-Sep-2020

Type

Standards Track

[desert peak]

Does anyone know what Macros would gain Python?

[dusty mango]

@desert peak i worry more confusing code

[desert peak]

well that's true of any feature, but my question is pointed.

[grave jolt]

@desert peak You mean, what advantages it might have?

[desert peak]

@grave jolt correct. what advantages does it give Python over its current state?

I read that it can enable features without being built into the language and that interests me a lot

[grave jolt]

yes, it's for trying out new syntax and semantics for features

for example, pattern matching

or 'do-notation' from haskell

the majority of people will probably hate your new favourite feature, but your personal project or your team might still benefit from it

[desert peak]

dreams of all the features

so basically it takes insert feature and generates python-valid code from the symbols to the macro?

like.. what macros do? lol

[grave jolt]

a macro is a function that runs at compile time

and it can transform the syntax tree

[desert peak]

hm, interesting. so would your source files always have a .pyc then?

[grave jolt]

what do you mean?

[desert peak]

well, just running a python script doesn't result in compilation

[grave jolt]

it does

how else would it run?

[desert peak]

I mean, a compiled file

[grave jolt]

Python source code is first compiled to bytecode, then that bytecode is executed

[radiant fulcrum]

its very common that your IDE will hide them from the file viewer

[grave jolt]

that bytecode may or may not be saved to a file

[desert peak]

where is logic that determines that? I don't ever see pyc files generated in my projects (though I do see __pycache__)

[grave jolt]

or 'do-notation' from haskell
when this:

do as result: 
    x := get_line
    y := get_line
    print_line("answer is:", x, y)
    pure(None)

is transformed to this:

result = get_line.flat_map(lambda x: get_line.flat_map(lambda y: print_line("answer is:", x, y).then(pure(None))))

[desert peak]

what is do notation, compiling a block to a single result?

[grave jolt]

"do-notation" allows you to write a chain like that in an imperative way

it's used in Haskell

[desert peak]

sounds like Rust's blocks

let result = {
  let x = get_line();
  let y = get_line();
  println!("the answer is: {} {}", x, y);
  None
};

[grave jolt]

no, that's a different thing

it (#internals-and-peps message) actually gets transformed into a chain like this

it's just building up a data structure

not performing all those operations

I guess do-notation is already implemented in some Rust crate

yep https://github.com/TeXitoi/rust-mdo

[grand crag]

I guess do-notation is already implemented in some Rust crate
wait there is this do-notation in some Rust crate ! Impressive 😂

[grave jolt]

I guess you can do something like

with do as result:
    do.x = get_line
    do.y = get_line
    do(print_line("answer is:", x, y))
    do(pure(None))

[desert peak]

interesting

[grave jolt]

oh, well, I've got an idea for a package

[desert peak]

that illustrates better for me

thanks

[grand crag]

oh, well, I've got an idea for a package
oh oh 😈

[crude turret]

in the shutil module, on linux specifically, shutil calls os.sendfile() for copying files, why do that when you could just do:

with open(file, "rb") as file:
    # write to new file

is it because they need a way to tell when the end of the file has been reached? i.e sent == 0

[peak spoke]

I'd guess that sendfile is more performant

[crude turret]

ah alright. ty

[plain spindle]

@deft pagoda if called as an instance method, the first argument will have a __func__ attribute pointing to the uninstantiated version. check if args[0].__func__ is f

🐍 class F:
••      def g(self):
••              pass
•• 
🐍 F().g.__func__
<function F.g at 0x7fca6637b3a0>
🐍 F().g.__func__ is F.g
True

caveat: this may be a CPython impl detail

[muted crane]

Please don't tell me I'm imagining a bunch of emoticons in your code

[radiant garden]

Just a custom repl skin

check out sys.ps1 and sys.ps2 ; )

export a script to PYTHONSTARTUP in your shell and enjoy

I personally have mine set to an ipython-like In [line]:

[radiant garden]

That being said... I wonder what's the wackiest repl prompt I could realistically work with.

Maybe this idea is closer to #esoteric-python in nature, lol.

[unkempt rock]

@grave jolt wouldn't it be this?

with result as do:
    do.x = get_line
    do.y = get_line
    do(print_line("answer is:", x, y))
    do(pure(None))

[plain spindle]

@radiant garden bruh i didn't even know about PYTHONSTARTUP

i just wrote a virtual env wrapper script which copies my sitecustomize.py file into every venv

[radiant garden]

lol

[flat gazelle]

@viral flame @dark granite this is a discussion channel, for questions of this kind I would suggest #❓｜how-to-get-help, this is not the proper channel.

[grave jolt]

@unkempt rock no, because do is the access point, and the resulting expression will be stored in result

actually, to make it threadsafe:

with do as result, h:
    h.x = get_line
    h.y = get_line
    h(print_line("answer is:", h.x, h.y))
    h(pure(None))

[unkempt rock]

why are sets left unordered when dicts are ordered?

it would make sense for them to have consistent behaviour right

[radiant garden]

I don't remember the rationale for making dicts ordered but one did exist and it was meaningful

[flat gazelle]

sets are optimized for fast in checks, and the current best way to do that is with using sets unordered

[sacred yew]

https://softwaremaniacs.org/blog/2020/02/05/dicts-ordered/

[flat gazelle]

dicts are ordered because it is an implementation detail of how they are written, and if it were not part of the spec, it would lead unportable code

[wide shuttle]

The reason is that sets work differently behind the scenes and it's not easy to make them remember insertion order while also keeping the flexibility of the set operations

[sacred yew]

idt the dict optimization applies to sets anyways

[wide shuttle]

Since sets, as a mathemacal construct, don't have an order, it's not a priority to make Python sets retain order

https://twitter.com/raymondh/status/944454031870607360

Raymond Hettinger (@raymondh)

@kunsv Sets use a different algorithm that isn't as amendable to retaining insertion order.
Set-to-set operations lose their flexibility and optimizations if order is required.
Set mathematics are defined in terms of unordered sets.
In short, set ordering isn't in the immedia...

Twitter · Dec 23, 2017, 6:26 AM

[flat gazelle]

since people would use dict ordering, and then it would not work outside cpython or when cpython decides to switch impls

[wide shuttle]

Although, arguably, I haven't followed later discussion on the topic in, say, the dev mailing list

Maybe there are some new plans now, but I wouldn't hold my breath, as sets being sets is the main priority

[safe hedge]

dicts are ordered because it is an implementation detail of how they are written, and if it were not part of the spec, it would lead unportable code
@flat gazelle That seems backwards. Like I'm not disputing this but I mean it feels wrong that python got a feature due to one particular implementation currently having that side-effect

[flat gazelle]

it is useful regardless, but it was orginally just an implementation that did not hold when keys get deleted

but it was easy enough to enforce

[wide shuttle]

It was introduced as an implementation detail in Python 3.6, with every warning that it wasn't a part of the specification

[safe hedge]

If users were opting to rely on a previously unenforced implementation side-effect that was on them surely

Again, I'm not complaining but it just seems a weird justification for adding as a feature

[wide shuttle]

Maybe, although it could also be considered a bit "tone deaf" if you break it in a few years now that is has been something that a lot of people started relying on

[safe hedge]

"A new implementation meant dicts were ordered, which some users started to rely on so now we are going to make it enforced"

[flat gazelle]

a language should not make it easy to do the wrong thing, and dict order made a lot of things much easier

[wide shuttle]

I think an earlier implementation even had deliberate scrambling of the order to prevent people from relying on that implementation detail

[safe hedge]

Surely that same logic could be used to justify making anything currently a CPython implementation detail the "standard"

[wide shuttle]

In part, although some things may be far more obscure and far less discussed than this implementation detail

[safe hedge]

a language should not make it easy to do the wrong thing, and dict order made a lot of things much easier
@flat gazelle Did people previously assume dicts were ordered? I feel like I learned very early dicts were unordered?

[wide shuttle]

The "why don't dictionaries retain order" has been a frequently asked question for a long time before 3.6 introduced it

[flat gazelle]

if you aren't sure, pop open a repl and check, and see that they are, you are not going to question it further

[safe hedge]

But there are lots of "why don't/doesn't..." questions which won't ever get changed

Just seemed a bit unnecessary to me and I do sort of find it more unintuitive for dicts/sets to have differing order behaviours than to have the same

[wide shuttle]

I think there was a point at which the decision was made that scrambling it on purpose just to not have that order show up was a bit weird as well. I think that was the decision that ultimately led to it being adapted into the language.

Dictionaries are so common and so many people really want their dictionaries to maintain order for some of their features, that once regular dicts did maintain order, people just started using it. And, at that point, you can go back and say "We told you it was an implementation detail" whenever it changes in the future (which could have been years from now), but I don't think that would have gone over well with a lot of users.

I agree that it's not a perfect reason, but it's a reason that makes sense once you're in the situation of having "remembering" dicts

[safe hedge]

I mean surely the answer to that is don't document it in the first place?

Like, if no-one mentioned they retained order even as an implementation detail would anyone have actually used it?

[unkempt rock]

ive never understood why people wanted dicts to be ordered anyways? its not like you can do dict[0] to pull out the first index

[radiant garden]

people will rely on the behavior regardless of documentation if it seems consistent

[wide shuttle]

It started to be discussed by a lot of people already, people who are intimately familiar with the implementation

[safe hedge]

people will rely on the behavior regardless of documentation if it seems consistent
@radiant garden Then people are dumb

[peak spoke]

It's nice to have and most major implementations did it anyway

[wide shuttle]

One of them even suggested in his Python talk something along the lines of "If you like this, like I do, just start using it! They can't change it then"

[flat gazelle]

that is a bad argument. You do not want magic behavior which is not actually guaranteed

programmers should not have to be smart to write python

[safe hedge]

I'm not sure who that is aimed at?