internals-and-peps

[gleaming rover]

No I don't want to log instances at creation. I just have a need to store child instances, in the parent, after they're defined.
@swift imp just to be clear

you want instances of the child class to be stored as a class attribute in the parent class?

[swift imp]

Yeah

but I'm realizing I think this is ill-thought, so I'm just going to capture the module and name of the child. @gleaming rover

[gleaming rover]

but I'm realizing I think this is ill-thought, so I'm just going to capture the module and name of the child. @gleaming rover
@swift imp but my esoteric Python 😦

[swift imp]

Hey man, I'm all ears

[gleaming rover]

@swift imp

class Parent:
    child_instances = []

    def __init_subclass__(cls, **kwargs):
        def init_hook(self, *args, **kwargs):
            old_init(self, *args, **kwargs)
            Parent.child_instances.append(self)

        super().__init_subclass__(**kwargs)
        old_init = cls.__init__
        cls.__init__ = init_hook

you could use weak references, I guess

[swift imp]

very very interesting my friend, I like this A LOT more than storing name and module of the child class.

[gleaming rover]

maybe old_init, cls.__init__ = cls.__init__, init_hook shrug

also if anyone has any ideas how to point to the current class without using the name Parent specifically...that'd be nice

[swift imp]

Well wait, this still has problems.

[gleaming rover]

Well wait, this still has problems.
@swift imp what?

[swift imp]

So like I'm using the **kwargs of __init_subclass__ as indicators to what child class to use for processing. Now I store those kwargs.values() in a frozenset and that becomes the key within an internal dict of the parent. What I want is for a child instances to be the values of that internal dict. Basically the top level module gets fired up, pulls some specific attributes from a file and then uses those attributes as a key to that internal dict of parent. So like child will never actually get instantiated on its own to make that init_hook go off.

[gleaming rover]

huh.

confused.

[swift imp]

right

[gleaming rover]

you want instances of the child class to be stored as a class attribute in the parent class?
@gleaming rover so you want to store the child classes themselves?

[swift imp]

hold on

[gleaming rover]

I thought you said you wanted instances of the child class

[swift imp]

Ideally, an instance would be perfect

but the way I'm storing the instances with your method

Like the instance wont get stored until its instantiated

[gleaming rover]

yup, because it wouldn't exist otherwise, right?

that's the earliest an instance can exist

[swift imp]

But in my case, child would never get instantiated by itself

Does that make any sense?

[gleaming rover]

So like I'm using the **kwargs of __init_subclass__ as indicators to what child class to use for processing. Now I store those kwargs.values() in a frozenset and that becomes the key within an internal dict of the parent. What I want is for a child instances to be the values of that internal dict. Basically the top level module gets fired up, pulls some specific attributes from a file and then uses those attributes as a key to that internal dict of parent. So like child will never actually get instantiated on its own to make that init_hook go off.
@swift imp I understand this

but it sounds like the values within said internal dict would be classes, right?

the said child classes above

i.e. not the instances

[swift imp]

@gleaming rover yeah it would be the cls and not an instance of cls, which is a problem because then @abstractmethod doesn't go off.

This is a work in progress but its my general idea

[gleaming rover]

@gleaming rover yeah it would be the cls and not an instance of cls, which is a problem because then @abstractmethod doesn't go off.
@swift imp any other problems with storing the classes?

because I think that's a more sensible approach

[swift imp]

@abstractmethod doesn't go off, if the property isn't defined

[gleaming rover]

other than that

[swift imp]

Other than that, nothing.

I think what I'm doing is essentially a class registry

[gleaming rover]

assert all(not getattr(cls, attribute).__isabstractmethod__ for attribute in dir(cls))

[swift imp]

right, because if its been redefined properly then __isabstractmethod__ should never be True

I didn't know you could do that!

[gleaming rover]

I didn't either

until 5 min ago

[swift imp]

Tyvm. Gonna delete that code bc its for work lol

[gleaming rover]

yw 🙂

[unkempt rock]

Hey guys, can someone help me with this https://dpaste.org/JmjH - not sure what im missing :`\

[gleaming rover]

Hey guys, can someone help me with this https://dpaste.org/JmjH - not sure what im missing :`
@unkempt rock this is a discussion channel; you might want to try #❓｜how-to-get-help

[unkempt rock]

oh thanks,

[heady mauve]

How are the stacks organized in the official CPython interpreter? I know there is a call stack that stores different objects, and that each frame has its own data stacks, but beyond that I have no idea.

[spark magnet]

@heady mauve the call stack is a linked list of frame object.

[heady mauve]

gotcha.

what other stacks are there?

[spark magnet]

and now that you mention it, i'm not sure each frame has its own data stack.

[heady mauve]

also no shock youre the one to respond given you wrote byterun XD

[spark magnet]

there's also a block stack, for each loop

(I just happened to peek in right after you asked 🙂 )

[heady mauve]

hehe fate works in strange ways

so what stacks are in each frame then?

[spark magnet]

in byterun, the Frame object has a stack attribute. but in CPython, the frame objects don't.

[heady mauve]

interesting...

[spark magnet]

maybe cpython has them but doesn't expose them?

[heady mauve]

why dont I ask a core dev? :P

[spark magnet]

(or read the source:) PyObject **f_valuestack; /* points after the last local */

[heady mauve]

fair

[spark magnet]

in frameobject.h

ok, so: call stack, data stack per frame, and block stack per frame

[heady mauve]

nice

you got that from https://github.com/python/cpython/blob/cb9879b948a19c9434316f8ab6aba9c4601a8173/Include/cpython/frameobject.h ?

[spark magnet]

the two "per frame" values are there, yes

[heady mauve]

thanks!

[unkempt rock]

@spark magnet mind gimme some of your thoughts on an issue I currently have? I saw different blogs, docs, etc. but I still can't figure out a piece of code

[gritty pebble]

#❓｜how-to-get-help @unkempt rock

[silk pawn]

so are we actually going to be getting match with pep 622 in 3.10? (https://www.python.org/dev/peps/pep-0622/#version-history)

can't tell if that's been accepted yet

the status says draft, but it specifies the version as 3.10

[boreal umbra]

@silk pawn the version number just indicates which version of Python the proposed functionality would be introduced into.

if it were approved right now (and it appears that it hasn't been), there wouldn't be enough time to include it in 3.9.

[raven ridge]

From the gossip I've heard, the steering committee is still pretty divided on it.

[boreal umbra]

I haven't really read how it works but the idea of adding soft keywords is a big red flag to me.

and I've never encountered a circumstance where chained if statements weren't perfectly fine

[raven ridge]

Switch statements have an advantage over chained if statements in that they can be compiled into a lookup table, so you don't need to check every condition... But, the proposed match statement won't do that.

(if you need a switch statement for its performance in Python, it's common to use a dict of functions as your lookup table)

[boreal umbra]

But, the proposed match statement won't do that.
@raven ridge so what's the point? syntactic nicety?

[raven ridge]

The point is to bring pattern matching to Python. It seems to be done in a way that's sort of crippled compared to pattern matching in other languages, though.

[boreal umbra]

I still really want function composition and except-with, but big changes are scary

[mint forge]

what is function composition

[sacred yew]

except with?

[paper echo]

https://stackoverflow.com/questions/64095346/pickle-how-does-it-pickle-a-function

[deft pagoda]

composition:

In [11]: f = 2 * x
    ...: g = 2 + x
    ...: f.subs({'x': g})
Out[11]: 2*x + 4

[boreal umbra]

what is function composition
@mint forge the idea is that (x @ y)(n) would be the same as x(y(n)). you could chain as many functions as you want as long as the value returned by each one is usable by the previous function in the chain.

[cloud crypt]

interesting

@deft pagoda what library/project is this

[deft pagoda]

sympy

[mint forge]

ohhhhhhhhhhhhh

[boreal umbra]

@ was introduced as an operator recently for matrix multiplication in numpy, but @ is already used for decorators and this is along the same lines.

@sacred yew except-with is just my proposed syntax for inline exception handling

[deft pagoda]

I do a lot of math-y stuff with python and even I don't think I'd use a composition operator that much

[boreal umbra]

var = my_list[8] except IndexError with 3.14

[deft pagoda]

in-line excepts might be ok, i'd rather have none-aware first

[boreal umbra]

so if my_list[8] raises IndexError, you replace it with the value after the with. There was actually a pep along the same lines but the proposed syntax was var = my_list[8] except IndexError: 3.14, though there was tons of bikeshedding.

no one proposed except-with and I'm not sure why. the only other existing keyword that I think could be sort of appropriate is as but that already has a specific meaning in the context of exception handling.

[deft pagoda]

should be else

[boreal umbra]

I don't like the use of the colon in the proposed syntax. Guido rejected it, but I think more because he was opposed to the concept of inline exception handling

why else?

[deft pagoda]

looks familiar

[boreal umbra]

I would interpret that to mean "this is the value you use if you don't get the given exception"

var = my_list[8] except IndexError else 3.14

"if you don't except IndexError (because you didn't get it), the value is 3.14"

[deft pagoda]

3.14 except IndexError else my_list[8]

[boreal umbra]

that treats the "excepted" value like that's what you expected it to be

[cloud crypt]

from types import FunctionType
from typing import Callable, TypeVar

from forbiddenfruit import curse

T = TypeVar("T")
U = TypeVar("U")
V = TypeVar("V")

def chainer(f: Callable[[U], T], g: Callable[[V], U]) -> Callable[[V], T]:
    def wrapper(value: V) -> T:
        return f(g(value))
    return wrapper

curse(FunctionType, "__matmul__", chainer)``` eeh

cursed but it works

[boreal umbra]

and the expression that's being tried comes after the exception that it might raise

[deft pagoda]

annotations makes it take twice as long for me to read

[raven ridge]

Hm, I kinda like except/with. I'm sure there are times I would have used it if it existed.

[deft pagoda]

x ??= something

[cloud crypt]

annotations makes it take twice as long for me to read
@deft pagoda oh, I am hugely into rust here, so... heh

[boreal umbra]

inline exception handling of some kind, func comp, none-aware

anything else we need before Python can have its final release?

[deft pagoda]

In [12]: from inspect import currentframe, getframeinfo
    ...:
    ...: def maybe(val):
    ...:     parent_frame = currentframe().f_back
    ...:     line = getframeinfo(parent_frame).code_context[0]
    ...:     var, sep, _ = line.partition('=')
    ...:
    ...:     if not sep:
    ...:         raise SyntaxError('maybe(val) needs to be called in an assignment expression')
    ...:
    ...:     glob = parent_frame.f_locals
    ...:     loc = {}
    ...:     exec(f'val = {var}', glob, loc)
    ...:
    ...:     return val if val is not None else loc['val']
    ...:
    ...: x = 3
    ...: x = maybe(None)
    ...: x
Out[12]: 3

[cloud crypt]

fn compose<F, G, T, U, V>(f: F, g: G) -> Fn(V) -> T
where
    F: Fn(U) -> T,
    G: Fn(V) -> U,
{
    move |value| f(g(value))
}``` just an example, haha

anything else we need before Python can have its final release?
syntactic macros and pattern matching, please

[deft pagoda]

pattern matching pep probably needs some revision before it gets accepted

[cloud crypt]

indeed

[deft pagoda]

i'd like dict-xor

[cloud crypt]

hm idk

>>> from iters import iter
>>> iterator = iter(range(10))
>>> dir(iterator)
['__class__', '__class_getitem__', '__delattr__', '__dict__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__getitem__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__iter__', '__le__', '__lt__', '__module__', '__ne__', '__new__', '__next__', '__orig_bases__', '__parameters__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__slots__', '__str__', '__subclasshook__', '__weakref__', '_is_protocol', '_iterator', 'all', 'any', 'append', 'chain', 'collapse', 'collect', 'compress', 'copy', 'copy_infinite', 'copy_safe', 'count', 'cycle', 'cycle_infinite', 'cycle_safe', 'drop', 'drop_while', 'enumerate', 'exhaust', 'filter', 'find', 'find_all', 'first', 'first_or', 'flatten', 'fold', 'for_each', 'get', 'get_all', 'group', 'group_longest', 'inspect', 'inspect_format', 'iter', 'iter_chunk', 'iterate', 'last', 'last_or', 'length', 'list', 'list_chunk', 'map', 'max', 'max_or', 'min', 'min_or', 'next', 'next_or', 'nth', 'nth_or', 'nth_or_last', 'partition', 'partition_infinite', 'partition_safe', 'prepend', 'product', 'reduce', 'repeat', 'reverse', 'reversed', 'set', 'side_effect', 'skip', 'skip_while', 'slice', 'star_map', 'step_by', 'sum', 'take', 'take_while', 'tuple', 'tuple_chunk', 'unwrap', 'with_iter', 'zip', 'zip_longest']```
I kinda liked the idea of OOP iterators, and so I made just that hahaha

>>> (
...     iter(range(10))
...     .filter(lambda n: n % 2)
...     .inspect_format("odd: {}")
...     .map(lambda n: n * 2)
...     .inspect_format("double: {}")
...     .sum()
... )
odd: 1
double: 2
odd: 3
double: 6
odd: 5
double: 10
odd: 7
double: 14
odd: 9
double: 18
50``` allows doing fun stuff

[boreal umbra]

@deft pagoda what would that do

[deft pagoda]

x = {'a': 1, 'b': 2}
y = {'b': 3, 'c': 4}
x ^ y  # {'a': 1, 'c': 4}

the keys not common to both

[boreal umbra]

ah

when would you use that?

[deft pagoda]

I would've used it a few times --- working with random data structures, I'm not sure I can pull a use-case out right now

[boreal umbra]

I wouldn't want dicts to use set operators in a way that isn't as close as possible to how the same operator is used for sets

so I guess I'd be fine if they added that

[deft pagoda]

but it's a natural operation on dicts unlike |

[boreal umbra]

why isn't union a natural operation?

[deft pagoda]

because there are two ways you can handle common keys

[boreal umbra]

ah

good point

[deft pagoda]

typically, the last-seen rule is what python uses

like update or |=

[teal yacht]

It feels very weird that they are adding intersection and union but not (symmetric) difference

[boreal umbra]

are they adding intersection?

I thought they were only adding union

[teal yacht]

Maybe I had dreamed it, can't check rn

Makes it worse

[boreal umbra]

I dreamed that global del was added as a compound keyword for forced garbage collection

that's the only python dream I remember tbh

I think I want to make a reddit thread to ask what forced garbage collection would entail. not because I think it's worth adding, I just wonder.

[teal yacht]

I think it's mostly a question of usability, what would be the purpose of manually setting the refcount of every variable to 0, you can just exit the program and run another

[raven ridge]

I don't think I understand what global del would do

[boreal umbra]

I don't know a whole lot about how languages are implemented but I figured you might be able to have a special value that you get an error if you try to access. So if you force delete something that you would have used later, you get an exception rather than a segfault. And then references to that special value would have the same reference count as the object they replace

[teal yacht]

No please don't add js' undefined

[sacred yew]

no that's haskell undefined

[boreal umbra]

@raven ridge the idea is that if you have global del x as a single statement, the object referred to by x is deleted before the next line is executed.

[sacred yew]

js undefined doesn't error

[boreal umbra]

and anything it contains references to have its reference count decremented, and those objects are also deleted right away if the count reaches zero. Again, not saying they should add that, that's just the idea.

[raven ridge]

In addition to being terrible, that would be very slow.

[sacred yew]

@boreal umbra that sounds like it should be a function

[teal yacht]

What about objects that depend on x

[sacred yew]

like on gc or smth

[teal yacht]

I'm not sure I see the point, still

[raven ridge]

Every time someone does any operation on any value, the operation would need to first check if the value has been flagged as already-destroyed and raise an exception if so. On every single operation.

[boreal umbra]

@teal yacht it would be the programmer's job to not mess it up. But I don't know if there's any way it could be implemented where you get an exception when trying to access a value you manually deleted.

[teal yacht]

We already get exceptions

Nameerror

[boreal umbra]

that's only for a specific name

this would be the object itself being deleted

[teal yacht]

So it's not really deleted?

[raven ridge]

In all cases where the user doesn't "screw it up", it would be equivalent to just del. The only case where it would be different is the case where other references exist and are now broken.

[boreal umbra]

!e

thing = 'hello world'
my_list = [thing, 2]
del thing
print(my_list[0])

[fallen slate]

@boreal umbra :white_check_mark: Your eval job has completed with return code 0.

hello world

[teal yacht]

How would you access a value that doesn't have any binding to it

[boreal umbra]

del thing doesn't delete 'hello world', it just removes thing as a name for that object.

[teal yacht]

Yes, and reduces the refcount of the object by one

[whole shore]

Oh nice, built-in python execution

[teal yacht]

(modulo the fact that's its a literal string)

[raven ridge]

But if it did delete that object, then my_list would be broken

[boreal umbra]

right

[raven ridge]

Which is bad.

[teal yacht]

I don't see the point

[boreal umbra]

there isn't necessarily a point

[teal yacht]

We'll get disguised segfaults everywhere

[raven ridge]

I mean, you already can do this, if you want to see just how disastrous it winds up being. Use ctypes to change the reference count of an object to 1 before you del it.

Leaving existing, broken references to an already destroyed object is a very common bug in Python C extensions.

Letting people introduce the same bug in plain Python code seems exceptionally horrible

[boreal umbra]

doesn't the Python api for C handle that as long as you use the decref function correctly?

or is that the problem?

[whole shore]

Python C extensions are bae

[raven ridge]

doesn't the Python api for C handle that as long as you use the decref function correctly?
@boreal umbra right, the bug is keeping a reference to an object that you forget to incref. Or, alternatively, accidentally decref it twice.

Either way, when its reference count drops to 0 it will be destroyed, but its reference count is off by 1 from what it should be, and so something is still referring to it even though it's already been destroyed.

[boreal umbra]

I should go to sleep so that my life is somewhat in harmony with my school schedule.

[stable cobalt]

hey folks

if there is one book you had time to read on python

what would you go for

[whole shore]

I personally quite liked "Automate the Boring Stuff" myself

Thought that was back when I was much more of a beginner

[mint forge]

how many years back may I ask, @whole shore

[stable cobalt]

man

I just checked it out

It looks like a fun read

any "python under the hood" kind of books?

[cloud crypt]

Py_INCREF and Py_DECREF surely give me chills, haha

[safe hedge]

If you want to get the full dotted import path of a class is it safe and correct to do f'{theclass.__module__}.{theclass.__name__}?

[flat gazelle]

There is __qualname__ iirc

[safe hedge]

Hmm so looking at it seems like to be robust it should be f'{theclass.__module__}.{theclass.__qualname__}?

[raven ridge]

That looks right to me. With the caveat that the full path to a module may not be the canonical path to it. That is, the class pkg._private_submodule.Klass may be reexported from pkg, so that users are meant to do pkg.Klass, and this approach would find pkg._private_submodule.Klass

Which isn't incorrect, but isn't exactly correct either.

Also with the caveat that I'm not sure __module__ is always defined. It appears to be writable for user defined classes, so someone could set it to something incorrect. I'm not sure if it's guaranteed to be set at all for classes defined in extension modules

[safe hedge]

So basically what I want to do is determine the dot-path of a class from within that class. The next part of this is I quite want to define it on a baseclass and then have it work properly for any subclasses defined elsewhere (i.e. return the dotpath of the subclass) - but I think that's potentially difficult?

[raven ridge]

That should work exactly the same way. The subclass has its own __module__ and __qualname__, so if you create a method that accesses self.__module__ and self.__qualname__, it will return the appropriate path for objects of whichever type self is, whether base class or subclass

[safe hedge]

Ok, that's pretty much what I thought I'd have to do. Since this doesn't actually require instance level values I was hoping I might be able to create something at the class level? Is there such a thing as a classproperty?

[raven ridge]

No, though you could make one if you were really determined. The descriptor protocol should allow that, I believe.

[safe hedge]

Meh a classmethod is basically fine I was just being picky

[ivory gorge]

Is there anyone here who knows a lot about python that could be a mentor and help me it on my beginner level ?

[muted crane]

Head to #python-discussion or #❓｜how-to-get-help , this channel is solely based on discussing about the language and not helping on any topics @ivory gorge

[unkempt rock]

Is there any way to serialiaze/pickle a file that uses selenium?

Apparently multiprocessing with selenium is (almost?) not possible because it can't be serialized

[shadow ridge]

https://medium.com/@daanyaalkapadia13/send-email-using-python-30fc1f203505?sk=3b8fa70bd633b73fd77d0e9a60f9cf61
Please go and check out my first article🗒️ on medium. It's the first time I am writing an article so your support will be appreciated🙌.

[gloomy rain]

@unkempt rock This is not a help channel. Please see #❓｜how-to-get-help.

@cinder vale This is not a help channel. Please see #❓｜how-to-get-help.

[heady mauve]

when bytecodes reference TOS (top of stack), are they referencing the top item of the data stack of the top frame of the call stack?

or are they referencing the top frame itself?

[silk pawn]

@boreal umbra thanks!

[boreal umbra]

You're welcome! What did I do?

if there is one book you had time to read on python
@stable cobalt Fluent Python

[stable cobalt]

You're welcome! What did I do?
@boreal umbra thanks!! I'm going to get that right now

[silk pawn]

oh you answered my question about the match statement

https://discordapp.com/channels/267624335836053506/709904092280914030/759984492013813770

[boreal umbra]

Ah

@stable cobalt we have some recommend books on our website and I'm pretty sure FP is one of them, if that makes you feel any more confident about purchasing it. However it was written when 3.4 was the latest version and a newer edition is coming out within the next year.

[stable cobalt]

I just looked at book depository

and realised how expensive everything got

[boreal umbra]

😭

[stable cobalt]

I'm going to go skulk to genesis

like seriously, books like doubled in price

has to be brexit

[boreal umbra]

it's all Brexit

always has been

[granite moon]

is brexit the place in uk?

[radiant scroll]

Is there a way to automatically run super().__init__() in the __init__ of some class which inherits from a class that requires this?
I'm building a library where people will need to make a custom class that inherits from mine and I don't want to bother them with having to use super().__init__() first every time they use it. I was wondering how to automate this.

[peak spoke]

It's how every other class will work, doing it automatically will just confuse people

[radiant scroll]

hm, I suppose you're right

out of curiosity though, would it be even possible?

[half wolf]

Probably yes, with some meta class magic for example. But yea.. Don't do that.

[wide shuttle]

you could probably hack something together with __init_subclass__ in that base class as well (by patching __init__ with a new function that calls the base init and the actual init). I'd be pretty flabbergasted if I were confronted with that, though.

[boreal umbra]

@tropic magnet this channel is for discussion. Take a look at #❓｜how-to-get-help so you can open a help session.

[tropic magnet]

i need help with the above problem statement

sorry im new around here

[unborn raven]

Whats the best module to use for proxies?

[grave jolt]

@unborn raven you should ask in #networks

[deft pagoda]

i've actually done this before, adding super().__init__ to subclasses

!e

from inspect import signature
from textwrap import dedent

def parse_signature(signature):
    return str(signature)[1:-1].split(', ', 1)[1]

def additive_init(cls):
    def __init_subclass__(cls):
        def add_parent_init(init):
            self_args = parse_signature(signature(init))
            parent_args = parse_signature(signature(cls.__base__.__init__))

            globals_ = {'init': init, 'cls':cls}
            locals_ = {}
            wrapper = f"""
            def wrapper(self, {parent_args}, {self_args}):
                super(cls, self).__init__({parent_args})
                init(self, {self_args})
            """
            exec(dedent(wrapper), globals_, locals_)

            return locals_['wrapper']

        cls.__init__ = add_parent_init(cls.__init__)

    cls.__init_subclass__ = classmethod(__init_subclass__)
    return cls

@additive_init
class Person:
    def __init__(self, name):
        self.name = name

class Cyborg(Person):
    def __init__(self, version):
        self.version = version

class Android(Cyborg):
    def __init__(self, model):
        self.model = model

print(signature(Android))

[fallen slate]

@deft pagoda :white_check_mark: Your eval job has completed with return code 0.

(name, version, model)

[deft pagoda]

@radiant scroll

[frozen abyss]

Hello,
I found this little explaining text on a python project and I wonder how it works:

clients never write anything to the disk - not even temporary files (zero IO system calls are made) because remote imports allow arbitrary code to be dynamically loaded into memory and directly imported into the currently running process

Anyone knows which module is used ?

[paper echo]

@frozen abyss can you post a link to the project?

[frozen abyss]

@paper echo
Found this text on this project: https://github.com/malwaredllc/byob

[paper echo]

it looks like it's more than just a single module that enables that functionality, maybe more of a clever design than some specific module

this is quite an impressive project

[unkempt rock]

so this is a little off topic it is java what does this mean

**** FAILED TO BIND TO PORT!
[13:36:00] [Server thread/WARN]: The exception was: java.net.BindException: Can't assign requested address
[13:36:00] [Server thread/WARN]: Perhaps a server is already running on that port?
[13:36:00] [Server thread/INFO]: Stopping server
[13:36:00] [Server thread/INFO]: Saving worlds
[13:36:00] [Server thread/ERROR]: Exception stopping the server
java.lang.NullPointerException: null

[boreal umbra]

@unkempt rock this is a discussion channel, so Python questions would also be off-topic for this particular channel. We don't really have a platform to answer Java questions here, but you could join the Together Java discord.

[sacred yew]

@heady mauve top of data stack

[heady mauve]

thanks @sacred yew

[swift imp]

I'm just now discovering how awesome dataclasses are, I recommend to all.

[tulip linden]

@swift imp that's actually really useful

[still crater]

@unkempt rock - It means there is already a Minecraft server running on that port, or you've selected a port that requires root access to bind to.

[grave jolt]

@swift imp If you like dataclasses, you can look into namedtuple and its awesome typed version -- typing.NamedTuple

it's like a frozen dataclass

(dataclasses can be frozen as well)

[peak pollen]

oh very interesting, I'm familiar with namedtuple but not the typed version

[grave jolt]

yeah, that can be useful if you want to build an ADT

from typing import NamedTuple, Union, Tuple, Callable


class Log(NamedTuple):
    severity: int
    location: Tuple[str, ...]
    message: str
    args: Tuple[Any, ...] = ()

class SpawnTask(NamedTuple):
    client_uid: int
    task_path: Tuple[str, ...]
    task_uid: str

class GetTaskResult(NamedTuple):
    task_uid: str

class GetConfigValue(NamedTuple):
    key: str

Command = Union[Log, SpawnTask, GetTaskResult, GetConfigValues]

Command is an 'algebraic data type' because it's a sum of products

(int * Tuple[str, ...] * str * Tuple[Any, ...]) + (int * Tuple[str, ...] * str) + (str) + (str)

[swift imp]

I'm super lost here, why on Earth does that __get__ not get entered??

class PullFrom(object): 
  def __init__(self, table:str): 
    print("storing table") 
    self.table = table 
  def __call__(self, *args): 
    print("storing column") 
    f, = args 
    self.column = f(self) 
  def __get__(self, instance, owner): 
    print(f"getting table name {self.table} and column name {self.column} to pull value from") 
    return  
class Test(object):     
  @PullFrom("A") 
  def value(self): 
    return "foo" 

storing table
storing column

>> t = Test()
>> t.value

[boreal umbra]

As inspired by the current name of ot1, why exactly is eval so evil?

[grave jolt]

why are you inheriting from object? 🤔

[boreal umbra]

I get that it can evaluate malicious code

[grave jolt]

well... that's it 🙂

[boreal umbra]

what if the user can't change what's being evaled without editing the file?

[spark magnet]

and it's overkill for some things people use it for

[sacred yew]

if the user can't change whats being evaled, then why are you using eval in the first place

[boreal umbra]

I've used it (or it might have been exec that I used) to create some dataclass instances that were written to a log file

since the __str__ of a dataclass is typically the constructor for that instance.

[swift imp]

It seems like inheriting from object or not doesnt matter. Even if you don't do it, you still inherit from it.

[sacred yew]

unless you're on py2

[deft pagoda]

you might want __call__ to return self at the end

[grave jolt]

@sacred yew it's used in namedtuple, I think

[spark magnet]

@boreal umbra why did you need exec/eval for that?

[swift imp]

@deft pagoda hot damn, smarty-pants, I'm still trying to learn all this stuff .🙂

[sacred yew]

yeah ik

but namedtuple uses it to dynamically create the class

[boreal umbra]

@spark magnet my program creates a lot of dataclass instances during its execution and I wanted to get some stats about what it outputs

[spark magnet]

hmm, i still don't see where the exec comes in

[boreal umbra]

so the easiest way to do that was to just resurrect all the instances and count how many items appeared in various list comps.

[deft pagoda]

I use exec quite a bit for code generators

[boreal umbra]

it might have been eval that I used. I can't remember which does which.

[deft pagoda]

unapologetic about it, cause i'm not about to build an ast and compile it -- yet

[grave jolt]

@swift imp __call__ returns None, so Test.value is None

[spark magnet]

exec is for statements, eval is for expressions

[boreal umbra]

was probably eval then

[raven ridge]

It's a good idea to steer people hard away from exec and eval because both seem like really elegant solutions to lots of problems, but are actually Python injection attacks in disguise

[grave jolt]

ast.literal_eval might be more useful, I suppose?

[swift imp]

@grave jolt ty ty

[grave jolt]

is that train sounds?

[deft pagoda]

exec is where all the meta magic happens

https://github.com/dabeaz/cluegen/blob/master/cluegen.py#L31 , https://github.com/python-attrs/attrs/blob/6aca10f2a7ae7d692502707bf8bffecb78e9ecbc/src/attr/_make.py#L297, https://github.com/python/cpython/blob/3.8/Lib/dataclasses.py#L398

behold the power of exec

[raven ridge]

it's certainly not that exec and eval are never useful - they're still in the language for a reason, and they're the best solution to some problems. It's just that it's an order of magnitude or two easier to use them than it is to use them safely, and drilling in to people's heads that it's generally unsafe and only safe in specific exceptional circumstances is a good starting point.

[grave jolt]

I see many people use it to implement a calculator for some reason

[deft pagoda]

gross

[grave jolt]

Maybe it's counterintuitive, but making a reverse polish notaiton calculator is much easier 😛

[raven ridge]

SQL injection attacks are still one of the most common types of attacks, and they're much easier to prevent than Python injection attacks are - there's usually just a "sanitize this input" function you can call on your database driver.

[grave jolt]

Lisp parser is also fun

[deft pagoda]

https://paste.pythondiscord.com/icugubiyan.rb I did a normal parser for a calc as an example once

[grave jolt]

I knew it, you've been a ruby programmer all along

[gleaming rover]

the main problem I have with exec/eval is that they're builtins in the global namespace

seems a bit of a bad decision to me

[spark magnet]

even worse: id

a great name, taken up by a nearly useless function

[deft pagoda]

fair

[gleaming rover]

yeah, I don't like that too

but for a different reason

[spark magnet]

what's the reason?

[grave jolt]

exec/eval shouldn't be in the global scope because they're almost never the best solution (like reduce?), and they create a security hole
id isn't harmful, it's just not very useful

[gleaming rover]

as in, I agree that exec, eval and id all shouldn't be in the global namespace; the first two because they have the abovementioned pitfalls that new users won't be aware of, and id for the reason you gave

I like reduce...

😦

and the really irritating thing?

we have comprehensions, and map/filter, which are alternatives

but reduce, which has no alternative other than a loop, has to be imported

🥴

[deft pagoda]

i would use reduce a lot more than filter

i almost never use filter

[grave jolt]

I wish there was a map that would return a list/tuple/string, not an iterator

[spark magnet]

there is, it's called list(map

[deft pagoda]

lol

[gleaming rover]

I think @grave jolt means like a method that returns an instance of the same class

[grave jolt]

yeah

[gleaming rover]

like Scala has

[grave jolt]

Well, I would rather use a list comprehension than list ∘ map

[gleaming rover]

scala> List(1, 2, 3).map(_.toString)
res0: List[String] = List(1, 2, 3)
scala> Vector(1, 2, 3).map(_.toString)
res1: scala.collection.immutable.Vector[String] = Vector(1, 2, 3)

[raven ridge]

I think you could easily look at Python's builtins and name 50% of them that shouldn't be in the builtin namespace easily.

[gleaming rover]

I think you could easily look at Python's builtins and name 50% of them that shouldn't be in the builtin namespace easily.
@raven ridge I think we did that a month back or so

I have literally never used oct

[raven ridge]

id is bad, sure - but ord and chr have no business being builtin, and I've literally never used slice. memoryview is pretty rarely used... help is only used in a repl and could be handled like quit is...

[gleaming rover]

oh, I've used slice a fair bit

compile should be with eval/exec

[raven ridge]

TIL that ascii exists 😄

[grave jolt]

slice and memoryview should be in some library like extra_builtin_types.py, maybe frozenset as well

[gleaming rover]

I would like it if frozenset got the f-string treatment

f{1, 2, 3}

also not really sure about basic math functions like abs being part of global namespace

[grave jolt]

...maybe just add those things in types and rename it?

[gleaming rover]

I think it's an interesting design choice

not clearly bad, but it might surprise some people

because these things aren't part of the global namespace in most languages

[raven ridge]

callable was removed and then added back. 😦

I've literally never used complex, either.

[deft pagoda]

stay away from complex, mate

[spark magnet]

i didn't understand why callable was removed. They said, "If you want to know if it's callable, just call it" which seemed really naive.

[grave jolt]

uh... yeah

[deft pagoda]

complex is my built-in 2-vec class

[raven ridge]

well, we do have inspect, though

[gleaming rover]

i didn't understand why callable was removed. They said, "If you want to know if it's callable, just call it" which seemed really naive.
@spark magnet sounds quite EAFP

[grave jolt]

"Hey, there's this funny do_segfault object, let's poke it and see if it's callable"

[spark magnet]

it does, but sometimes you need to know without actually calling it

[gleaming rover]

yeah, precisely

[grave jolt]

Well, checking if it has __call__ might be ok

[boreal umbra]

exec/eval shouldn't be in the global scope because they're almost never the best solution (like reduce?), and they create a security hole
id isn't harmful, it's just not very useful
@grave jolt what's wrong with reduce?

[gleaming rover]

one of the cases where the EAFP idea breaks down

[raven ridge]

note that it's not possible to check if a class has an attribute without fetching that attribute.

hasattr calls getattr and swallows the AttributeError.

[grave jolt]

@boreal umbra Well, it goes in the same place as the other things like memoryview and id and slice. They can be useful, but not in the global namespace

[boreal umbra]

I thought reduce was in functools

[grave jolt]

Yes, it is

[gleaming rover]

I thought reduce was in functools
@boreal umbra it is

that's the problem

why? 😡

[deft pagoda]

i've used slice enough

[boreal umbra]

I've only used reduce to chain __or__ calls

[gleaming rover]

yeah, me too

slice is good

[grave jolt]

@gleaming rover I think I used it reduce very few times. Can you show examples of where it's better than a loop?

[deft pagoda]

slices are neat when you abuse them for things they aren't meant for

[gleaming rover]

don't think it's ever better per se...?

[grave jolt]

@raven ridge Well... that makes sense, kind of. Because many magical proxies/frameworks/etc. use __getattr__/__getattibute__

But maybe __hasattr__/__hasattribute__ could be a thing as well?..

[gleaming rover]

but it is more concise

and generally fits better with the way I think

I just don't really like loops

[grave jolt]

Maybe I'm just too used to loops

[gleaming rover]

or rather, I don't like loops when there aren't side effects

which is a bit of me being a purist but well

like I think I go quite out of the way not to rebind variable names

[grave jolt]

well, if the function already exists, it could be better

[gleaming rover]

I do agree that reduce is generally less comprehensible for the average person than loops

[raven ridge]

like I think I go quite out of the way not to rebind variable names
I think I'm the opposite, I greatly enjoy rebinding variables, almost whenever I can.

[grave jolt]

can you use just one variable name for the entire program?

no, this is getting #esoteric-python

[raven ridge]

mypy drives me crazy by assuming that a variable is only gonna have one type, heh

[boreal umbra]

@raven ridge does it assume that you're never going to reassign instance.__class__?

[raven ridge]

I mean, other people do need to maintain my code 😄

but I don't think twice about doing things like

if not isinstance(name, str):
    name = name.decode()

and mypy doesn't like that at all.

[unkempt rock]

anybody knw how to downlaod libraries in python

[unkempt rock]

pip install (library name)

[unkempt rock]

I'm trying to retrieve my ip address in Sockets SERVER = socket.gethostbyname(socket.gethostname()) but its throwing the local ip 127.0.1.1

please ping me if anyone knows what I'm doing wrong

[elder hawk]

Hey there! is there anyone who is able to help me out on my assignment? I have some questions and doubts I wanted to ask for my assignment

shoot me a Private message because I really need help and hopefully i can get going from there

[brave badger]

@elder hawk Hello, just a fair reminder that this is not a help channel. Feel free to check out #❓｜how-to-get-help though.

[elder hawk]

ok

[loud summit]

This is not a help channel, if you need help with python you'd be better to check #❓｜how-to-get-help
Also keep in mind pretty much any music service prohibits automation of their services. @scenic flare

[gloomy rain]

@unkempt rock This is not a help channel.

[unkempt rock]

@gloomy rain I didn't ask for help

[gloomy rain]

No, but you provided it.

Which is still off-topic.

[unkempt rock]

I was just saying that

Also keep in mind pretty much any music service prohibits automation of their services.
this is not true

[gloomy rain]

Which is still not the topic of this channel.

[unkempt rock]

okay if my two lines made the whole discussion off-topic I am sorry.

[gloomy rain]

Do not continue or prolong off-topic conversations. Thank you.

[boreal umbra]

Is dataclass a type of metaclass?

[peak spoke]

There may be some meta in how it works, but the dataclass decorator calls a function that does the magic

[boreal umbra]

@peak spoke well that answers that. Thanks!

what's next?

[peak spoke]

Looks like they got rid of the only metaclass usage a while ago because class getitem came around. The function works directly on the current cls object with modifying its attributes https://github.com/python/cpython/blob/master/Lib/dataclasses.py#L809

[gleaming rover]

I think I'm the opposite, I greatly enjoy rebinding variables, almost whenever I can.
@raven ridge why, though?

my PoV is that a name should have a consistent meaning while it's in scope

and if it's rebound, then that's not true

[soft bone]

Hi, i have a question in Cpython 3.8 api.
i am trying to do the same thing as

vars(str)
``` but in C. the equivalent of `vars()` in Cpython-api is `PyModule_GetDict`, and the equivalent of `str` in Cpython-api is `PyUnicode_Type`.

So i thought that i should do `PyModule_GetDict(PyUnicode_Type)` , but that doesn't work because `PyModule_GetDict` should get `PyObject` as its parameter and  `PyUnicode_Type` isn't `PyObject`, it's `PyUnicode_Type`

i am getting that error:
```python
SystemError: Objects/moduleobject.c:477: bad argument to internal function```

i will be happy to have some help, Thanks.

[raven ridge]

@raven ridge why, though?
@gleaming rover because when something semantically represents the same piece of data in a different format, I prefer using the same name. Given the choice between:

def execute(query):
    if isinstance(query, bytes):
        query_bytes = query
    else:
        query_bytes = query.encode()

And:

def execute(query):
    if not isinstance(query, bytes):
        query = query.encode()

I much prefer the latter

@soft bone PyModule_GetDict requires a module, not just any Python object, and str/PyUnicode_Type is a type, not a module

[soft bone]

@soft bone PyModule_GetDict requires a module, not just any Python object, and str/PyUnicode_Type is a type, not a module
@raven ridge so how can i do vars(str) in c-api? can you please send me a link or something to take a look at?

[raven ridge]

@soft bone PyObject_Dir((PyObject*)&PyUnicodeType) should do the trick, I think

[soft bone]

@soft bone PyObject_Dir((PyObject*)&PyUnicodeType) should do the trick, I think
@raven ridge Thank you, apreaciating!

[boreal umbra]

what is the default metaclass for user-defined types? do int, str, dict, and list each have their own metaclass since they each have their own built-in syntax?

[peak spoke]

The type of the builtin classes is type, so that should be their metaclass; same with an user defined class where you don't specify it

They don't really need metaclasses since you can do much more at the C level

[boreal umbra]

so are there any stdlib examples of a non-type metaclass?

[peak spoke]

enums use metaclasses extensively for example

[boreal umbra]

!e

class A: pass
thing = 'hello world'
thing.__class__ = A

[fallen slate]

@boreal umbra :x: Your eval job has completed with return code 1.

001 | Traceback (most recent call last):
002 |   File "<string>", line 3, in <module>
003 | TypeError: __class__ assignment only supported for heap types or ModuleType subclasses

[boreal umbra]

so A and str share the same metaclass, but I guess user-defined types belong to "heap types"?

[peak spoke]

I believe you'd get the same error with an instance of A

[boreal umbra]

maybe if you try to replace an A instance's class with str but it does work in general.

!e

class A: pass
thing = A()
thing.__class__ = str

[fallen slate]

@boreal umbra :x: Your eval job has completed with return code 1.

001 | Traceback (most recent call last):
002 |   File "<string>", line 3, in <module>
003 | TypeError: __class__ assignment only supported for heap types or ModuleType subclasses

[unkempt rock]

IDK if it's advanced, but what is my error? The terminal doesn't show nothing, but the bot isn't removing the roles. [discord.py]

@client.command()
async def det(ctx, member: discord.Member, role: discord.Role):
    print(member)
    print(role)

    roles = [r.name for r in member.roles]

    r_testadores = discord.utils.get(member.guild.roles, name = 'testadores')
    r_da = discord.utils.get(member.guild.roles, name = 'Da')
    r_db = discord.utils.get(member.guild.roles, name = 'Db')
    r_dc = discord.utils.get(member.guild.roles, name = 'Dc')
    r_dd = discord.utils.get(member.guild.roles, name = 'Dd')
    
    if r_testadores in roles:
        await member.remove_roles(r_testadores)
    if r_da in roles:
        await member.remove_roles(r_da)
    if r_db in roles:
        await member.remove_roles(r_db)
    if r_dc in roles:
        await member.remove_roles(r_dc)
    if r_dd in roles:
        await member.remove_roles(r_dd)

    await member.add_roles(role)```

[peak spoke]

After looking around, yeah it does look like a heap type refers to a class that was defined with the class statement (and type I guess) in python

[zenith topaz]

@unkempt rock I don't know the library, but it looks like roles is a list of role names, instead of the actual roles.

[unkempt rock]

@unkempt rock I don't know the library, but it looks like roles is a list of role names, instead of the actual roles.
@zenith topaz yes! roles give me the roles name.

[zenith topaz]

It doesn't really fit in this channel though you can open a help channel in #❓｜how-to-get-help or use the #discord-bots channel

[unkempt rock]

It doesn't really fit in this channel though you can open a help channel in #❓｜how-to-get-help or use the #discord-bots channel
@zenith topaz Ohhh, thx!

[unkempt rock]

Hello has anyone tried DataCamp or DataQuest?

[boreal umbra]

I haven't, but we do have some suggested learning material

!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

[unkempt rock]

@boreal umbra Those are just python-language oriented
The things I linked are career/skill oriented, such as using python with pandas, matplotlib, tensorflow, sql and so-on to do computer science, math and data science etc

[boreal umbra]

you could ask in #career-advice how people got the training for the jobs they have, I suppose.

[unkempt rock]

Heyg

[unkempt rock]

I think this channel is not that active

So I wanna ask something so reply to me personally or something

Or jus reply here fast or something

So is there a code to convert IP to CO ordinate?

Longitude and latitude

[boreal umbra]

@unkempt rock if you ever want to know if there's a python package for something, check the Python Package Index (PyPI--which is always pronounced "pie pee eye" and not "pie pie"). Be sure to read the topic for different channels because we always want to stick to them (especially this one) even if an on-topic discussion isn't ongoing.

[unkempt rock]

Oh ok

@unkempt rock There are APIs for that I believe

A quick Google search returned this for example: https://ip-api.com/

Also just an fyi, you won't get absolute cords with any IP service, most they'll give you is the cords of the city the IP is located in, for privacy reasons, plus the ISP of said IP only really knows exactly where that IP is located at any given time

[swift imp]

Can you define __eq__ on a dataclass and still have it create __hash__ for you automatically? I can't tell from the description at all

For it to generate __hash__ you need frozen=True and eq=True but if you define a custom __eq__ then eq arg is ignored so like

What does that mean wrt to __hash__

[teal yacht]

It's not possible, if it does generate it for you, it's a bug

because there's no way to know what the eq function does to create the hash automatically while keeping the property if x == y: hash(x) == hash(y)

[swift imp]

So I'd have to implement __hash__ myself

[late jay]

Hi, not sure if this is the right place to ask, but is anyone here familiar with pybind11? I'm working on a project that uses python and c++.

[raven ridge]

@late jay #c-extensions would be better

[late jay]

Awesome - thanks!

[raven ridge]

@swift imp if any member variable used by __eq__ can be changed, it shouldn't be hashable at all. Otherwise, you'll need to define you own __hash__ based only on the fields you included in __eq__, being careful to preserve the invariant that object equality implies hash equality

[unkempt rock]

what are good repl.it alternatives?

[boreal umbra]

@unkempt rock #tools-and-devops would be the channel to discuss that.

[hot oracle]

is there a good way to parse parts of a file as a CSV file?

[boreal umbra]

@hot oracle you could use regular expressions to parse out the parts that you want and send them over to pandas.

if you have any more questions about that, I'd check out #❓｜how-to-get-help

[hot oracle]

as in, there's a specific topical chat I should ask in, or I should use what's described there to get my question asked in one of the open help channels?

[boreal umbra]

the latter, though #data-science-and-ml would be the most relevant topical channel if you wanted to go the pandas route.

[wet obsidian]

Anyone else that enjoys reading the PEPs? :)

[spark magnet]

someone must

[boreal umbra]

@wet obsidian what is your favorite pep?

[wet obsidian]

Pattern matchmaking thing, already hyping

[brave badger]

I like the pattern matching PEP but I'd rather have null-aware operators in the meantime

[wet obsidian]

👀 how does that work?

something that has been suggested too?

[brave badger]

https://www.python.org/dev/peps/pep-0505/

[wet obsidian]

yeah, I found it myself

I like it!

It says 3.8??

[brave badger]

It's an old PEP that was deferred

[wet obsidian]

oh...

for how long? and why?

[brave badger]

I'm still looking as to why, but I asssume it's indefinite

[wet obsidian]

On top of my head, I often do
if car and car.fast: car.brake
which could be replaced by
if car?.fast: car.brake i think? hope I understood it well

In particular I don't like the first form, because I feel like it's sorta counter-intuitive because it's "abusing" short circuit evaluation

[spark magnet]

@wet obsidian how is it abusing? Isn't it exactly what short-circuiting is for?

[wet obsidian]

Dunno...? Figured it was for performance or something?

but since we're talking Python here, you're probably right lmao

[brave badger]

I mean it's more of a

# Coalescense (roughly)
if a is None:
    return b
else:
    return None

# Maybe access
if hasattr(obj, "attr"):
    return obj.attr
else:
    return None

[wet obsidian]

hmmm?

[brave badger]

or I misunderstood coalescense

[wet obsidian]

Yeah well, doesn't matter, if we won't be getting it anyway

🤣

[brave badger]

Still, having short-circuting attribute access is fairly neat

[wet obsidian]

For sure

It's just not very consistent with other operators

and sorta adds an extra layer of complexity:" what's gonna get evaluated and what won't"

I'm nitpicking ofc

Also.... if they didn't short circuit, is it possible we could overload them?

[deft pagoda]

there's a dunder proposed i think at the bottom of the pep

i don't remember what it was

[wet obsidian]

sorry, I meant the or and and ones

[deft pagoda]

I played with attribute chaining not long ago, it's not difficult to implement a version https://github.com/salt-die/Snippets/blob/master/maybe.py

[wet obsidian]

Oh, like in the PEP?

but, home-brew

[brave badger]

In Haskell there exists the Maybe datatype (e.g. Option in Rust if you're familiar) that basically allows you to short-circuit evaluation

getInfractionMessage = do
    infraction <- getInfraction 42 infractions  -- If this fails
    reason <- getMessage infraction             -- this never runs
    return message                              -- nor does this

I mean, it's useful for a language like Haskell because it can be repesented in an abstract manner plus you don't really want exceptions in the first place; there's also the fact that there's a million other combinator functions that you can compose

A special mapping type that can do a great job at error reporting might be a better candidate for the task to be honest

[deft pagoda]

wouldn't be hard to add error reporting to the class above, but it's maybe unweild-y, I dunno -- it was more academic

[brave badger]

Maybe something like:

try:
    Maybe(infractions)[42]["reason"]
except MaybeKeyError as e:
    print(f"Failed to access {e.last} at {e.depth} depth.")
```?

[deft pagoda]

i mean, sort of the point is that you don't have to catch errors though

[brave badger]

Hm, fair

[deft pagoda]

also to avoid the nested elifs when checking deep

[brave badger]

To be honest I don't really see features like pattern matching or none-aware operators making it to the language in the near future e.g. 3.10, 3.11

[deft pagoda]

probably not, i think pattern matching is too controversial --- though I'm really not sure why none-aware has been deferred so long

[wet obsidian]

pattern matching was approved though? Whoever we've enslaved to work on python just needs to work faster

[brave badger]

PEP 622 is still a draft but there's an implementation

[deft pagoda]

switch case itself could be revisited at some later time, if it ever became more popular

[brave badger]

Was the walrus operator really the reason why GvR stepped down as BDFL?

[wet obsidian]

PEP 622 is still a draft but there's an implementation
oh my bad, dunno how I got it mixed up

[brave badger]

or it's just some tall tale

[wet obsidian]

probably scored a job at google

[deft pagoda]

no, it was part of the reason

there was a lot of vitriol over walrus

[wet obsidian]

why is it so bad?

[deft pagoda]

I don't think it's bad

I like the operator

[brave badger]

It's alright but I really do underuse it

I'd like to keep compatibility with 3.7 as much as possible

[wet obsidian]

huh?

why?

[brave badger]

Some packages that I'm currently writing I'll go and publish to PyPI

[wet obsidian]

PyPI doesn't work with 3.8?

[north root]

i'd love a none-aware operator, but i probably wouldn't use it very much in Python

[brave badger]

No, I just like having packages be compatible with a previous release, especially if they're tools

[north root]

i haven't ran into many situations where you chain lots of function calls

[brave badger]

Cries in fluent interface

[wet obsidian]

tbh, all the low hanging fruit is probably already taken. Hard to come up with new stuff

No, I just like having packages be compatible with a previous release, especially if they're tools
Okay, I see

[brave badger]

RegExp literals

[wet obsidian]

oh yeah, I don't see why not

maybe something like
re"$hiya.*"

[sacred yew]

no

that still looks like a string literal

and the whole point is that regex literals aren't strings

just use slashes like other langs

[livid mason]

does this question belong here

[boreal umbra]

@livid mason you can ask questions in this channel if they're meant to spur discussion about the design of the language. Most questions belong in a help channel.

[livid mason]

okay thank

[prisma cypress]

Hey guys, I'm trying to move away from pipenv. But I do like the virtual environment setup.

I was wondering if pip-tools + venv will work about the same way as pipenv?

[boreal umbra]

@prisma cypress I've only ever used venv. In what way do you feel that it's lacking?

[prisma cypress]

I remember coming across with pipenv having issues installing packages.

I've also read that pipenv stopped development. I figured I might as well take the opportunity to try something new.

Though to be fair, I think the issues while installing packages were from pipenv dependencies instead of pipenv itself.

[boreal umbra]

What does pipenv do that isn't easy to do with venv?

I honestly don't know.

[prisma cypress]

Oh I've never used venv before. hahaha

[boreal umbra]

Do you have a question about how to use it?

[prisma cypress]

I'll probably try it out myself later and if I have further questions I'll come back.

I was just wondering about the ease of use and what managing venvs are like.

Such as listing the virtual environments.

[livid mason]

lister = ["hey", ["hello", "hi"]]
for word,the_list in lister:
    print(word)
    print(the_list)``` error: ```py
    for word,the_list in lister:
ValueError: too many values to unpack (expected 2)``` why doesnt the library allow me to iterate through a sting and a list within a list?

any alternatives to this?

[prisma cypress]

You can do nested loop.

[boreal umbra]

@livid mason that question isn't on topic for this channel. You can use a nested for loop though.

[livid mason]

can u give me links to what nested loop is? thanks, i wont post q like this again here

[boreal umbra]

for string in lister:
For char in string:
Do stuff

[gleaming rover]

can u give me links to what nested loop is? thanks, i wont post q like this again here
@livid mason it's something you can easily Google IMO

[boreal umbra]

I'm on my phone so I can't do better than that.

@prisma cypress one trick with virtual environments is that you can't move them, I don't think

They manifest as a directory with a few sub directories, one of which has the python executable.

[tawdry gulch]

Wait, why doesn't that unpack tho

there are 2 vals

[boreal umbra]

@tawdry gulch every value in a list has to be a tuple of the same length for unpacking to work.

[gleaming rover]

@tawdry gulch every value in a list has to be a tuple of the same length for unpacking to work.
@boreal umbra every value in an iterable has to be an iterable of the same length, but yes

[tawdry gulch]

Oh, yeah, my bad I am being stupid

For some reason I thought he was trying to unpack the list itself, woops

[boreal umbra]

Hmm, I didn't think it worked for non iterables.

Sorry, non tuples

[gleaming rover]

it would work if it was lister = [['hey', ['hello', 'hi']]]; the loop would run for one iteration with word = 'hey' and the_list = ['hello', 'hi']

[boreal umbra]

I'm in bed with one eye open, don't at me.

[prisma cypress]

Ahh okay okay. Thanks Stelercus.

From what I've gathered, it seems like just manually dealing with the venv instead of the automating part from pipenv.

I think I can deal with it.

My other option was using poetry.

[half wolf]

Pipenv is pretty much dead too.

[boreal umbra]

I stick with just using venv over conda and such because if I run into an environment related issue, the solution will probably assume I'm using a virtual environment.

And I don't want to have to think about if another thing like that is going to affect the applicability of an online answer

Because how else am I going to code?

[half wolf]

Staying on the beaten path is a good default, unless you have strong reason to be out in the woods :)

[boreal umbra]

Now I'm thinking of lost in the woods from frozen two. I liked that movie.

Another limitation of venv that I can think of is that you can't upgrade python versions, I don't think

[prisma cypress]

Quick question about venv. I can have the directory made first and then linking the virtual environment to it right?

Let's say I pull from github.

Oh as in you can't upgrade the python version after the virtual env is made?

[boreal umbra]

I don't believe so

But you can easily make a new one.

[prisma cypress]

Ahh okay okay.

[boreal umbra]

Also you don't want the virtual environment in github or in a git repo

It would be part of your git ignore

[prisma cypress]

Ahh. Where would you recommend setting up the virtual environments?

[boreal umbra]

If you're making an installable package, you would have the dependencies in the setup file

And you would install the package in editable mode

Otherwise you have a requirements file

Which is just a text file that lists the dependencies

[half wolf]

You can't upgrade python and keep your installed packages without venv either, so no difference there.

[boreal umbra]

And pip can automatically install everything listed in a file with the -r flag

Ahh. Where would you recommend setting up the virtual environments?
@prisma cypress virtual environment generally live in the top folder of the project they serve

They just aren't checked into git

[prisma cypress]

To clarify, then we do add it to git ignore?

[boreal umbra]

Git ignore templates for python projects will always have venv

[prisma cypress]

Oh! Ty ty.

I've tried out venv thus far, more typing. As such with source .../bin/activate.

But it seems pretty smooth.

[boreal umbra]

That's the worst part

Source venv bin activate

But you'll get used to it 👍

[prisma cypress]

To remove the venv, can I just delete it?

[boreal umbra]

Yes

[prisma cypress]

cool cool cool.

I think it'll be alright if it lives at the top of the project folder.

[boreal umbra]

Ye

Any other questions about that?

[prisma cypress]

Nope, thank you so much! Right now I have to shop for a router because a friend's router is dying.

[half wolf]

It's not much typing if you use a good shell that auto completes for you. I recommend fish.

[radiant garden]

Aliases are also useful. I have vv bound to source .venv/bin/activate and vvv to python3 -m venv .venv && [[ -f requirements.txt ]] && pip install -r requirements.txt

[unkempt rock]

or use autocomplete feature of zsh shell

[cinder nacelle]

i need help

File "main.py", line 1, in <module>
    from imageai.Prediction.Custom import ModelTraining
  File "C:\Users\arnav\AppData\Roaming\Python\Python37\site-packages\imageai\Prediction\Custom\__init__.py", line 4, in <module>
    from ..DenseNet.densenet import DenseNetImageNet121
  File "C:\Users\arnav\AppData\Roaming\Python\Python37\site-packages\imageai\Prediction\DenseNet\densenet.py", line 21, in <module>
    from tensorflow.python.keras.utils import convert_all_kernels_in_model
ImportError: cannot import name 'convert_all_kernels_in_model' from 'tensorflow.python.keras.utils' (C:\Users\arnav\AppData\Roaming\Python\Python37\site-packages\tensorflow\python\keras\utils\__init__.py)

error

[grave jolt]

@cinder nacelle This is not a help channel, this is a discussion channel, see the description. For help, check out #❓｜how-to-get-help

[grave jolt]

@pallid crescent Let's not post that here

[unkempt rock]

Hi

[gloomy rain]

@unkempt rock This is not a help channel. Please see #❓｜how-to-get-help.

[uncut sage]

@prisma cypress actually, you can upgrade a venv after it's been created, but it's often just as easy to create a new one in the same project folder and re-apply the requirements.

[frozen abyss]

hello,
Anyone knows how to execute code after having compiled a python program in bin ?

thanks in advance

[grave jolt]

How much is Python influenced by Smalltalk?
I'm working through the GNU Smalltalk tutorial, and I am sensing some slight similarities.

Like, for example, classes (which are objects as well) have a new class method, which is responsible for making a new instance, and instances have an init method which is called by new.

Object subclass: #Account
    instanceVariableNames: 'balance'
    classVariableNames: ''
    poolDictionaries: '' !
Account comment: 'I represent a place to deposit and withdraw money' !

!Account class methodsFor: 'instance creation'!
new
    | r |
    r := super new.
    r init.
    ^r
!!

!Account methodsFor: 'instance initialization'!
init
    balance := 0
!!

[wispy holly]

Hi ! Has anyone here worked on GIS maps ... like NASA Earth data in Python ?

I have a few questions regarding it ...

[grave jolt]

@wispy holly This is not a help channel, it's a discussion channel. See #❓｜how-to-get-help

[wet obsidian]

and I am sensing some slight similarities.
Dunno, I'm "sensing some slight similarities" between all languages lmao

[grave jolt]

Well, C and Python don't have many similarities 🤔

I mean, compared to haskell they do

[wet obsidian]

Perhaps

[boreal umbra]

@grave jolt this code looks like ruby to me

[brazen jacinth]

Any structural advice for basically wrapping an external api, one that requires a lot of config options (basicly handles all interaction with a google static maps api) and basically it needs around 5 mandatory args, and around 15 optional ones, im having trouble deciding between some kind of enums for specific optional param choices or just sticking everything in kwargs?

for the latter a pattern has emerged, there's gotta be a better way...

if kwargs.get('optional_param'):
    do something()
else:
    something_else()

Any suggestions?
Also sorry if it's not the right channel, perhaps more of a software design question

[boreal umbra]

@brazen jacinth I suppose it's on topic in the sense that it's "how can I use the language most expressively?"--if you're using repl.it, remember to change the settings to be four spaces per indent.

[manic hatch]

can anyone tell me how discord python bot works?

[brazen jacinth]

just a pasting issue ;P

[boreal umbra]

@manic hatch are you asking how our bot works, or how to use the discord api for Python? We have a channel to discuss the discord api, it's #discord-bots

[manic hatch]

i want to use this python bot in other apps

[boreal umbra]

@brazen jacinth you could use early exiting

[brazen jacinth]

not sure what exactly that is, but the problem is, this function is being used by cca 30 other functions, and has grown out of bounds to be used in many ways, so i would kinda like to retain "backwards-compatibility" by checking for all the arguments to still retain support for all these functions

[grave jolt]

@brazen jacinth how about 'optional param' in kwargs?

[manic hatch]

can anyone address my question?

[brazen jacinth]

that would work in general, but would have to refactor all of those functions in order to make it work, but yea, if i started anew that would seem like a good choice

im gueesing the fact that i have to retain this compatibility (or simply bite the bullet and refactor now), it makes things a bit more..difficult annoying

[boreal umbra]

@manic hatch here is the source code for the bot: https://github.com/python-discord/bot
You would need to adopt it for your server and deploy it, as it is only designed to work on this server.

[grave dust]

Hello, is anyone here good with django or http requests and routing?

[brazen jacinth]

you'll have more luck in #web-development

[grave dust]

thanks

[manic hatch]

Ty ty @boreal umbra

[grave jolt]

another pattern I've seen is is bib = whargs.pop("bob") -- to check whether there are unprocessed kwargs (in order to throw an error)

[brazen jacinth]

thanks, that popped some ideas actually

[grave jolt]

no joke
      no pun

no pun indented

[half wolf]

@brazen jacinth I've written about someething that sounds very similar: https://kodare.net/2020/09/14/transparent_apis.html

I might have explained it badly though, so feel free to tell me you don't understand at all what I'm on about :)

[heady mauve]

Is it possible to communicate with a printer by sending it an http request of some sort?

I see that my printer appears on my print queue for notepad with the location 'http://192.168..:***' (I added the stars), so I wonder if Windows just uses http to print to wifi printers.

[gleaming violet]

of course

@heady mauve Checkout the win32print library.

[hallow scroll]

@heady mauve no need to censor that ip by the way, it's a private ip.

[heady mauve]

ok thx guys

[wet obsidian]

Re: boxed. Seem interesting, but yeah, I at least didn't catch the final part >_>

[unkempt rock]

]

[modern frigate]

Why does true multi-threading not exist in python

[brave badger]

I'd assume because of the GIL

[modern frigate]

The what? excuse my stupidity 😅

[brave badger]

The GIL is the global interpreter lock, if I remember correctly, it makes sure that only one threads steps through Python code at a time

[feral cedar]

why do we need the GIL? i heard one of the reasons was to keep reference counts accurate?

[deft pagoda]

well, you don't need it, but if you remove it you have to put locks everywhere and python gets even slower

[raven ridge]

"true multithreading" isn't a very precise term, anyway.

[deft pagoda]

i imagine at some point far in the future python will be multi-threaded naturally

probably when having infinite locks doesn't outweigh the gains from running on 100 cores

or something

or someone figures something more clever out

[raven ridge]

"Python" the language doesn't have a GIL, necessarily. CPython the reference implementation of the language does, but it's an implementation detail. CPython does have "true multithreading" in the sense that it is possible to run code in multiple operating system threads.

[deft pagoda]

well, yes

[raven ridge]

What's not possible in CPython is having multiple threads executing Python bytecode or manipulating Python objects simultaneously.

[deft pagoda]

you can still use multiprocessing and universal queues

though, i haven't found many use cases for it -- but I don't do a lot of concurrent programming

[raven ridge]

and, things like numpy can often release the GIL while they do their stuff.

"true multithreading doesn't exist in Python" is only true if you make up a definition for "true multithreading" on the fly. There's a lot more nuance to the story.

[deft pagoda]

you ever seen this: https://github.com/dabeaz/curio/blob/master/curio/queue.py#L81

[raven ridge]

I'm vaguely familiar with curio, but not in any detail

that's pretty cool.

[deft pagoda]

he has talk somewhere where he uses threading and async together using these queues

[raven ridge]

It's a clever design, but I'm trying to wrap my head around the use case for it - presumably it's to make it so that the producer doesn't have to know where or how the consumer is running (or vice versa, perhaps) - but how often do you change your concurrency model? Is it often enough to go through the overhead and extra abstraction of using a different queue library right from the start to enable the flexibility to switch later?

[deft pagoda]

maybe, I think the nice thing about these Queues is the abstraction isn't really that deep, it's not too hard to grok the entire curio codebase really -- makes it way nicer than trying to learn async from scratch, say

[raven ridge]

I mean, I get that it's a pretty thin abstraction, but you need to know that you want that abstraction from the start, or you need to always use this queue whenever you would use any queue...

[manic hatch]

can anyone help me with bs4

i am learning

[deft pagoda]

it's probably pretty easy to just use it as a drop in whenever you need it

[raven ridge]

#❓｜how-to-get-help @manic hatch

[manic hatch]

#help-candy is the channel. Join me if you can guide me with bs4

[raven ridge]

eh - it's got a consistent interface for all concurrency models, but that interface is different from all the standard queues. it's not API compatible with a queue.Queue - and if you need to make changes to your code to switch away from queue.Queue to something else, you could just as easily switch to multiprocessing.Queue as curio.UniversalQueue

[deft pagoda]

is it that different? it's got a get and a put

[raven ridge]

it lacks options and knobs that you can twiddle for a queue.Queue, but fair enough that if all you need is a blocking get and put, it could just be swapped in

[manic hatch]

can u guys help me with bs4?

i need to learn to scrape

[raven ridge]

I haven't used that library in ages, and it's off-topic for this channel.

[safe hedge]

@manic hatch please don't keep asking in here - it's off topic. Be patient with your help channel, everyone here is a volunteer giving up their time freely. If it goes unanswered you can try asking again at a later time, when someone else who can help might see it.

[hearty monolith]

Can Python benefit from branchless programming?

[raven ridge]

Basically, no.

[modern frigate]

Not really.

Afaik it would just make python slower :/

[sacred yew]

if you need to use hacks like that just write it in c/c++

[hearty monolith]

Why can't Python benefit from branchless?

[raven ridge]

Because the interpreter itself needs to branch a lot in the process of executing your code.

Avoiding branches in your Python code likely actually means performing more CPU branches, because avoiding branches in your Python code generally means executing more bytecode instructions, and executing more bytecode instructions requires more CPU branches in the interpreter.

[hearty monolith]

TIL

Thanks!

[undone hare]

This is a cool idea, but I’m not sure if it feels really pythonic, what do ya’ll think? https://mail.python.org/archives/list/[e-mail]/thread/QAU3URTLBVZDQA7TWECUXCG6FENRR7VZ/

[quick snow]

Unpythonic. What if replace is also defined?

[brave badger]

I'm not really sure to be honest,

x = "foo"
x .= upper()
```feels unintuitive as I personally mentally parse `upper()` not as a method of `x` but a separate function.

[quick snow]

def foo():
    pass

x .= foo()

Exactly.

This will have the IMO surprising effect that those two foo() s are unrelated

[undone hare]

Yeah, that’s a good point

[cloud lion]

https://del.dog/psutilerr
Why try execpt block is not catching error while importing psutil ?

[flat gazelle]

The bigger issue imo is what would foo.=6 do

[quick snow]

@flat gazelle foo = getattr(foo, "6") ?

[spice pecan]

Force-casting the value to a string before the attribute lookup?

[quick snow]

Why force-casting? foo .= bar also "force-casts" the identifier bar into the string "bar" when doing foo = getattr(foo, "bar")

[spice pecan]

6 is not an identifier though

[quick snow]

So? My point was: whether it's an identifier, or a number, or whatever, it always gets converted into a string.

Attributes are always strings

foo .= bar isn't foo = getattr(foo, bar)

[deft pagoda]

i don't find it too unintuitive, but I can't think of much use for it either

[spice pecan]

It's passable I guess in cases like string .= upper(), but IMO it's really, really iffy

I don't think the convenience is worth the questions it starts

[deft pagoda]

i just can't think of any time i've ever wanted the convenience

i'm trying to think of one

[spice pecan]

String manipulation is the only thing I can come up with

[flat gazelle]

Raku does have it, Nim too most likely

[deft pagoda]

would that be a new dunder

[flat gazelle]

I could see it as an alternative to the . operator, but then it has to be an expression like :=

[deft pagoda]

__igetattr__

[flat gazelle]

Returning a function from that to make a.=b() work sounds not great

And, well, what about every other operator, such as a.=b[0]

[quick snow]

What about a []= 0 , meaning a = a[0] ? 😄

[flat gazelle]

That I would be more ok with, since you put an expression into getitem anyway

Would conflict with the kwargs in getitem pep, but whatever

But at the same time, that happens rarely

[quick snow]

But then we'd also need a ()= foo, bar=23 , meaning a = a(foo, bar=23)

[flat gazelle]

That seems utterly useless tbh.

[gloomy rain]

@unkempt rock This is not a help channel. Please see #❓｜how-to-get-help.

[hard adder]

Hey everybody, I am trying to advance my knowledge on metaclasses and actual usage of them, could somebody point me to an actual project that has used metaclasses properly and extensively so I can do a little reading/inspecting on it ? 🙂

[quick snow]

Django?

[hard adder]

I am not very familiar with Django it self so I find it a bit hard to understand what's going on there exactly

but yeah, guess I will have to stick with it for a bit ^^

[quick snow]

I think SQLAlchemy also uses them for declarative models, that might be a bit easier to understand, if that is a thing when it comes to metaclasses 🙂

[hard adder]

Oh, awesome, SQLAlchemy is thing I have some experience with, not very extensive, but at least some

Thanks a lot! ❤️

[quick snow]

It feels like with every release of Python metaclasses are needed less and less. Most things can be done with class decorators or new dunder methods nowadays..

[hard adder]

I am just interested in understanding them, I really see no reason of using them except for maybe Singleton implementation, but just want to get the hang of them, for my own sake 🙂

[magic python]

is stating that an input value requires __iter__ a way of saying that it's fine to pass a list, a tuple, a dictionary ( just the keys will be used ), or not. I feel like these can often be used for the same purpose - ie, iterating over them and carrying out some process. I'm not sure what would go wrong by enforcing that the input type has __iter__ instead of using isinstance( obj, (list, tuple, dict))

[brave badger]

__iter__ would allow for iterables to be passed so that's that, if you have a way to manage infinite collections on whatever procedure you're running then not much would go wrong.

[magic python]

so i guess using hasattr(obj , "__iter__") vs isinstance( obj, (tuple, list, dict)) 🤔

[sacred tinsel]

you can use typing.Iterable

>>> from typing import Iterable
>>> 
>>> isinstance([], Iterable)
True
>>> isinstance((), Iterable)
True
>>> isinstance({}, Iterable)
True
>>> isinstance(123, Iterable)
False

[magic python]

i don't use typing at all 😬

[brave badger]

>>> from collections.abc import Iterable
>>> isinstance(obj, Iterable)

But yeah, the re-exports from typing are nice

[magic python]

they're making it more straightforward in 3.9 aren't they ?

maybe I'll jump on then

[brave badger]

PEP 585: Builtin Generic Types
In type annotations you can now use built-in collection types such as list and dict as generic types instead of importing the corresponding capitalized types (e.g. List or Dict) from typing. Some other types in the standard library are also now generic, for example queue.Queue.

[magic python]

that's it, i guess

[brave badger]

It's pretty neat

[magic python]

right - as i understand it'll be more straightforward from 3.9, and save me doing things twice if i wait a bit

i might just be being lazy tho and looking for excuses

[sacred tinsel]

the docs mention a caveat with __getitem__: https://docs.python.org/3/library/collections.abc.html#collections.abc.Iterable

well I think the best way to express that is to annotate that param as typing.Iterable, at least that's the way I'd approach it

if someone decides to pass a non-iterable, let it fail 🤷‍♂️

[magic python]

and you'll do that after 3.9 as well?

[sacred tinsel]

Iterable is an abc

I'm not sure if 3.9 changes anything w.r.t. that

you no longer have to import typing.List to annotate something as List[int]

you can just do list[int]

[magic python]

yeah , maybe that's all there is 😄

[gloomy rain]

@gusty relic This is not a help channel. See #❓｜how-to-get-help.

[south token]

plz help me at #help-apple I need more advanced ppl

[amber nexus]

This isn't a help channel

And you don't need to plug your help channel anywhere

[dreamy adder]

muted?

[gloomy rain]

!mute 453358610953535500 12h The message literally right above is telling you that this is not a help channel. When you come back, stop blatantly disregarding staff instructions.

[fallen slate]

:incoming_envelope: :ok_hand: applied mute to @south token until 2020-10-02 01:06 (11 hours and 59 minutes).

[boreal umbra]

you can use typing.Iterable
@sacred tinsel I thought nothing in typing had any effect at runtime

but you're saying that it works for isinstance?

!e

import typing as t
my_type = t.List[int]
my_list = [1, 2, 4]
print(isinstance(my_list, my_type))

[fallen slate]

@boreal umbra :x: Your eval job has completed with return code 1.

001 | Traceback (most recent call last):
002 |   File "<string>", line 4, in <module>
003 |   File "/usr/local/lib/python3.8/typing.py", line 766, in __instancecheck__
004 |     return self.__subclasscheck__(type(obj))
005 |   File "/usr/local/lib/python3.8/typing.py", line 774, in __subclasscheck__
006 |     raise TypeError("Subscripted generics cannot be used with"
007 | TypeError: Subscripted generics cannot be used with class and instance checks

[boreal umbra]

guess it's only "compound" types.

[modern night]

that wont work, you can't use isinstance with typed generics afaik

[boreal umbra]

:((((((((((

will we ever get that?

[modern night]

ig its cuz its very costly to evaluate as data strucutes like lists can contain any (and a mix) of types, so it would need to look at every item in that container

[boreal umbra]

it should be up to the programmer if they think it's worth spending computation time on that.

[modern night]

i think it would be nice if it could evaluate generators through isinstance by their return annotation (if it has one), so for example py isinstance(itertools.count(), typing.Iterable[int])would return true (assuming itertools.count() has annotations saying that it will return int only

its not fool proof, you could lie about the types returned but I feel for the most part that could be nice in some cases

[boreal umbra]

it could just check that everything it returns it an int.

/s

[modern night]

but that wouldn't work for infinite iterators like itertools.count(), it can't evaluate every item

if thats what you mean

[boreal umbra]

the /s is for "end sarcasm" 😄

[modern night]

oh ok

lol

[boreal umbra]

but that wouldn't work for infinite iterators like itertools.count(), it can't evaluate every item
just run it in parallel
/s

[cloud crypt]

it can’t and shouldn’t evaluate iterators at all

[cloud crypt]

Okay a weird question, not sure if #internals-and-peps or #esoteric-python — has anyone ever heard about any project that can connect several python files (say, in a package) into one?

[unkempt rock]

wrong channel

Hello, I'm looking for a feedback based on experience regarding ML forecasting technics issues in training the model in a small sample size and moving it to whole set of data, for example take a small sample of data that is very representative of the whole set test different models select one then go to the selected model train it in the whole set of data to see if that's the right model to use for our data ? will this be a correct approach ?

[low lagoon]

@proper remnant please read the description of channels before posting.

[sacred tinsel]

but you're saying that it works for isinstance?
@boreal umbra isinstance(x, typing.Iterable) works yea, but it cannot be parametrized with the subscript

but unless you need to branch based off whether something is iterable or not (or a similar use case), it's probably better to not do the check

if you try to iterate over something that is not iterable, you already get a very nice type error out of it

so it should be considered whether raising your own type error even adds anything of value

[unkempt rock]

@unkempt rock Yes that will be a good method. If your original dataset is very large or you have limited computation power.
Random Sampling generally works fine in most cases but if your dataset is special like Time Series then random sampling is not a good choice.
Otherwise if you do a good sampling you should be able to get a good model.

[sacred tinsel]

on the other hand, if you have a function that is annotated to take a e.g. a list but can in fact take any iterable (I find that's often the case), you may want to consider relaxing the annotation

[boreal umbra]

@unkempt rock that would be a question for #data-science-and-ml

so how does it know if something is iterable? hasattr(x, '__iter__')?

I also try to use t.Iterator any time it doesn't have to be a list.

sorry, iterable

[wide shuttle]

That wouldn't work

You can create an iterable object if the class just implements __getitem__

!e

class MyIterable:
    def __init__(self, data):
        self.data = data

    def __getitem__(self, i):
        return self.data[i]


for i in MyIterable([1, 2, 3]):
    print(i)

[fallen slate]

@wide shuttle :white_check_mark: Your eval job has completed with return code 0.

001 | 1
002 | 2
003 | 3

[wide shuttle]

In this case, it relies on the first item being able to be fetched using 0 and it raising an IndexError when you run out of items. (I think StopIteration would also work.)

[boreal umbra]

I don't like that

[teal yacht]

    @classmethod
    def __subclasshook__(cls, C):
        if cls is Iterable:
            return _check_methods(C, "__iter__")
        return NotImplemented```

def _check_methods(C, *methods):
    mro = C.__mro__
    for method in methods:
        for B in mro:
            if method in B.__dict__:
                if B.__dict__[method] is None:
                    return NotImplemented
                break
        else:
            return NotImplemented
    return True```

[wide shuttle]

This is sometimes referred to as the "old" iteration protocol

[boreal umbra]

glad that it's old

[wide shuttle]

I think the docs somewhere say that just trying iter(object) is the best way to see if something is iterable

[sacred tinsel]

I guess if you really need to check, you could:

>>> try:
...     iter(x)
... except TypeError:
...     ...

yea I think the docs recommend it

The only reliable way to determine whether an object is iterable is to call iter(obj).

[boreal umbra]

I have to take a practice test, so I have to tune out interesting stuff.

[wide shuttle]

This SO answer seems to suggest that iter(object) is not the best way anymore: https://stackoverflow.com/a/61139278

[teal yacht]

isinstance(MyIterable([]), Iterable) in your example yields False btw

despite being actually iterable

[wide shuttle]

Yeah

[teal yacht]

a function that doesn't block the current thread of execution

[unkempt rock]

anyone know what

root.wm_overrideredirect

means

in tkinter

[wide shuttle]

I think these two questions, about learning asynchronous functions and tkinter, would fit a help channel better

[unkempt rock]

sure

how do i do a help channel

Oh.... didn't saw the channel name...my bad ..sorry

[wide shuttle]

This is a discussion channel to discuss Python itself, from a higher-level and more abstract perspective. We do have up to 32 help channels. (See #❓｜how-to-get-help for info on how to open a help channel.)

[unkempt rock]

@wide shuttle yo how do i see if someone answered my question

[boreal umbra]

@unkempt rock going to the channel where you asked the question. You can use the Discord client to search for messages from yourself and then go to the one where you asked your question.

[unkempt rock]

yes

channels = *channels```
How do I unpack a list in a variable? The above is wrong but shows what im trying to accomplish

[flat gazelle]

if you just want the first element, channels, = channels

[unkempt rock]

no

i want to unpack the whole list

[flat gazelle]

what should be the result then?

[unkempt rock]

[[['a', 'b', 'c', 'd', 'e']], [['f', 'b', 'c', 'd', 'e']], [['f', 'b', 'c', 'd', 'e']], [['f', 'b', 'c', 'd', 'e']], [['f', 'b', 'c', 'd', 'e', 'f']]]
to
[['a', 'b', 'c', 'd', 'e']] [['f', 'b', 'c', 'd', 'e']] [['f', 'b', 'c', 'd', 'e']] [['f', 'b', 'c', 'd', 'e']] [['f', 'b', 'c', 'd', 'e', 'f']]

[flat gazelle]

channels = [*channels]
``` will unpack the list into another list

that is not a data structure

[unkempt rock]

thats the before and after

[flat gazelle]

a single variable always holds a single value

[gilded storm]

Can I make my own web scraper in python as I can in Perl? and where can I find the documentation?

[flat gazelle]

you can, probably using beatifulsoup4 or selenium

[unkempt rock]

@flat gazelle what does that have to do with my problem?

[gilded storm]

Thanks @flat gazelle have used Beautifulsoup4 before, with IMDBPy

[flat gazelle]

you apparently want multiple values in a single variable. Or do you want something such that when you do print(a) it does print(*original_a)

[unkempt rock]

Look

I have [[['a', 'b', 'c', 'd', 'e']], [['f', 'b', 'c', 'd', 'e']], [['f', 'b', 'c', 'd', 'e']], [['f', 'b', 'c', 'd', 'e']], [['f', 'b', 'c', 'd', 'e', 'f']]]

There are 5 lists in a list, in a list

[flat gazelle]

oh wait, I am in #internals-and-peps, lets move to a help channel

mb

[unkempt rock]

I want to just have 5 lists in a list

@flat gazelle

[boreal umbra]

My friend says that you should always explicitly inherit from object

because explicit is better than implicit

but I don't think that applies to things you should know about the language.

[quick snow]

I think it was a good idea in 2-to-3 transition times, but that should hopefully be over in 2020

[boreal umbra]

also I see inheritance from object as an implementation detail and not much else

[quick snow]

It's not an implementation detail, but nowadays I would see class Foo(object): pass similar to foo = int(3) . Superfluous, obvious.

[peak spoke]

I just see it as overly verbose syntax, anyone familiar with the language knows that everything's an object at the base so inheriting from it just feels overly verbose. No point in doing it for python3 only projects, and python 2 projects are hopefully migrated or migrating to 3

[grave jolt]

Maybe it's just that some resources/schools are still teaching python2, and that gets stuck

[radiant fulcrum]

nothing like schools teaching depreciated versions 🙃

[unkempt rock]

I know this isn’t the right server

But does anyone know what I’m doing wrong here

I’m Lowkey fuming that it’s not working

Shouldn’t be that hard

[grave jolt]

@unkempt rock This is not a help channel. See #❓｜how-to-get-help or #discord-bots

[hollow crane]

@unkempt rock you just want mylist[0]

[swift imp]

Yo there's this blog that has this awesome picture of the order of how __init__, __new__ and __call__ are executed during class creation and I can't remember it but I'm curious if anyone here remembers

https://blog.ionelmc.ro/2015/02/09/understanding-python-metaclasses/#id28

[peak spoke]

call of the class shouldn't be called as a part of the constructor; the main things that happen is that new creates an object and then init of that object is called if it's an instance or a subclass of the cls passed to new

[raven ridge]

I think it was a good idea in 2-to-3 transition times, but that should hopefully be over in 2020
laughs in corporate

[radiant scroll]

How could I define a dataclass from a dictionary like this:

dct = {"param1": [int, 0], "param2": [str], "param3": [str, "default value"]}

which should represent this dataclass:

from dataclasses import dataclass

@dataclass
class Entry:
  param1: int = 0
  param2: str
  param3: str = "default value"

[raven ridge]

Probably easiest to use exec, unfortunately

[radiant scroll]

oh

[raven ridge]

Actually scratch that

I didn't know about https://docs.python.org/3/library/dataclasses.html#dataclasses.make_dataclass

That may help

[radiant scroll]

oh, that's handy, thanks

[raven ridge]

That should be able to do the trick pretty easily. How did I never know about that! 😅

[hollow crane]

why do you need to

woops

why do you need to specify the type if there's a value

casting?

[tropic gale]

Hello everyone, I'm tryna send an email using python's SMTP lib and stuff. Mail is working fine and sending. The mail message is in HTML format. Is there any way I could format the message, like so I personalize the mail. For instance, each mail addresses the user's name, like hello {username}. I was able to do this using plain text message, but I'm tryna implement it in HTML. Thanks guys, I could snap the code if you want.

[dim harbor]

Hi @tropic gale, you would just add a <p> element with your greeting text and template field like you show, and replace with the user name however you are doing it for plain text.

HTML is just plain text anyway.

[grave jolt]

@tropic gale This is not a help channel. See #❓｜how-to-get-help

[swift imp]

Can someone explain to me why 'foobar' in Class.__dict__ happens twice

Can't find it in the class and can't find it in the instance therefore it looks in the class again?

[spark magnet]

where did this drawing come from?

[swift imp]

https://blog.ionelmc.ro/2015/02/09/understanding-python-metaclasses/#id28

[spark magnet]

are you sure it's right? At the top is just Class, no instance, so where does instance come from?

ionelmc usually knows what he's doing, but it doesn't look right to me

[swift imp]

I don't understand why it looks in class first either

[spark magnet]

the text clarifies it some: the first box is tp_getattro, not a native attribute lookup

it's news to me that that happens, but there's a lot of subtlety here

[boreal umbra]

how do you think IDE creators decide what should be what color?

It looks like self (or whatever you call the first parameter of a non-static method) has its own color in PyCharm

built-ins also have their own color. Keywords share a color with objects that get a non-reassign-able name.

[gleaming rover]

It looks like self (or whatever you call the first parameter of a non-static method) has its own color in PyCharm
@boreal umbra it does?

mine is the same hm

a nice healthy orange

[boreal umbra]

self is orange for you?

[gleaming rover]

all parameters are orange

[boreal umbra]

what version of pycharm?

for me the other parameters are white (default color)

also I'm using the dark theme for pycharm

[gleaming rover]

2020.2, but it's been like that since forever

think it's just my theme

[boreal umbra]

ah

I like dark themes for everything

[feral cedar]

i think in pycharm there's a setting for self and cls

[dusky crescent]

Hey does anyone know how to code a custom stream bongo cat?

[feral cedar]

it's special in pycharm

[boreal umbra]

@dusky crescent I'm not sure what that is but you'll want to see #❓｜how-to-get-help to learn about the help system. This is a discussion channel.

[unkempt rock]

anyone can help me with if else functions?

its just one question

[boreal umbra]

@unkempt rock as I said, this is a discussion channel

however

if you have an if statement, you'll have at least one line of code indented under that if statement. but there will probably be a few lines

these line(s) are your if block

after that block, you have the option of having a line that says else:, and after that you can have another block of lines that are indented under the else: line

that's your else block, and it's optional, but the else block will be entered only if the if block is not entered.

if you have any questions about that, please open a help channel and I'll explain it in more detail.

[unkempt rock]

i did

i opened a help chanel

[unkempt rock]

am I able to interact with another program (python script with interactive shell) with subprocess but keeping the state and continue sending commands to the 2nd program while parsing on the first? subprocess can keep the state of my 2nd program or I should use something else?

[sacred yew]

yes?

use subprocess.Popen, but this is wrong channel

[sand musk]

hi posted related on devops channel but wanted to ask along here, i'm trying to setup python auto linting check upon any pull requests to github repo, is this something doable only from github side? or should this involve build server like jenkins?

[gleaming rover]

what do you all use objects for?

like object()

[charred wagon]

Sentinels is the only use case I can think of off the topic of my head

[exotic pendant]

hey guys is pygames on 3.8.5?

[gleaming rover]

hey guys is pygames on 3.8.5?
@exotic pendant try #game-development, this is for language discussion

[exotic pendant]

oh my bad

[gleaming rover]

is pyaudio available on 3.8
@scarlet ingot this is a discussion channel...

[scarlet ingot]

oh

where do i ask this

[gleaming rover]

general?

[scarlet ingot]

k