dev-contrib

[surreal veldt]

Can asyncpraw be added to the docs command?

[vale ibex]

!docs setdoc asyncpraw https://asyncpraw.readthedocs.io/en/stable/objects.inv

[stable mountain]

Added the package asyncpraw to the database and updated the inventories.

[vale ibex]

They don't have the top level module documented

!d asyncpraw.Reddit

[stable mountain]

asyncpraw.Reddit

class asyncpraw.Reddit(site_name: Optional[str] = None, config_interpolation: Optional[str] = None, requestor_class: Optional[Type[asyncprawcore.requestor.Requestor]] = None, ...)```
The Reddit class provides convenient access to Reddit’s API.

Instances of this class are the gateway to interacting with Reddit’s API through Async PRAW. The canonical way to obtain an instance of this class is via...

[vale ibex]

using it on an actual class/method will work

[surreal veldt]

Oh okay thanks

[vale ibex]

can you find the class in the docs?

[brazen charm]

I think it may be, but I can't really get anything to work. Documenting the typevar with either a comment or a docstring under didn't change the error. When I let it emit all warnings it also can't resolve the Callable uses, not sure if that needs to be brought into scope for sphinx or something

[clever wraith]

Remove write-protected folder (y/n)?

[gritty wind]

I’ll look into it, but no promises yet. We might just have to disable symbols for that typevar which wouldn’t be great but

[gritty wind]

This is wonderful

Running sphinx with -v complains about a different not found symbol than running with -vv

How does verbosity affect the problem, well...

haha

[brazen charm]

It looks like it works if they're public an exported, thought I tried that but apparently I didn't correctly. Though it's having problems with the paramspec as it's trying to use it as a normal typevar

and callable is still messed up

[tawdry vapor]

Maybe sphinx doesn't work with ParamSpec yet?

[brazen charm]

Yeah, it looks like it's happy if I make it a typevar before the TYPE_CHECKING if so it can find it, and then patch an attribute it tries to access from the paramspec

How should a decorator be documented to expose what the wrapper raises? Should I just add a raises to the function that creates the decorator?

[tawdry vapor]

No I don't think that would be correct.

Maybe the best way is to just write it out as a sentence that's part of the summary.

Or description/body. Whatever the proper term is.

[gritty wind]

I've done some more testing, and I think this is actually not at all related to TypeVars

We just don't have a proper handler for unexported internal references

Sphinx has support for adding one

I'll draft something up

Minimum repro:

class A:
    ...

class B(A):
    """Abc."""
    ...

With B exported. Substitute A for generic, and you're at the same point we started

[brazen charm]

__all__ = ["CommandOnCooldown", "block_duplicate_invocations", "P", "R"]

_ArgsTuple = tuple[object]

P = typing.TypeVar("P")
"""The command's signature."""

if typing.TYPE_CHECKING:
    from botcore import BotBase
    import typing_extensions
    P = typing_extensions.ParamSpec("P")
    P.__constraints__ = ()

R = typing.TypeVar("R")
"""The command's return type."""

builds for me, though it only sees the typevars. Haven't really been able to get it to have proper typing while satisfying whatever sphinx wants as it doesn't like the P TypeVar under an else for TYPE_CHECKING

[tawdry vapor]

Thoughts on removing the extra space before the code block for eval results?

It throws me off sometimes. Looks weird. I dunno

[fallen patrol]

I tried it with monty, didn't like it

it's too compressed on some devices, and doesn't give the codeblock room to breathe

[cold island]

Hey @vale ibex do you remember when we updated dpy we had that weird bug where pressing a help embed button would cause it to change the embed, but then still load and say the interaction failed?

Do you remember what that was about?

[vale ibex]

they changed something about not being able to modify the context after creation

[cold island]

hmmm that's really weird. All I did was fix the return button and now it's back

[vale ibex]

it was this that was reverted https://github.com/Rapptz/discord.py/pull/7685

specifically this line is what was causing the problem with our help embed https://github.com/Rapptz/discord.py/pull/7685/files#diff-6afaab6ebc84984f398353e5b84ed690bc1cc12b6f03f8a9971a6580cc15d94dR284

[cold island]

Oh hmm

We didn't fix it actually

Try !help i and then clicking the search button

[vale ibex]

yea, not sure

all I saw was that this was reverted, so not sure what else is causing the issue

[cold island]

The interaction might be hanging

[vale ibex]

https://github.com/Rapptz/discord.py/commit/29668c51aebfe429b386db32396753e7ae4d44e5

this was the revert

it wasn't a strict revert, so might be something causing issues still

[cold island]

Ok, I'll investigate, thanks

[whole forge]

ImportError: cannot import name 'music_cog' from 'music_cog'

Anyone knows how I can fix this?

[cold island]

fwiw the issue was that I had to defer the response. The odd thing is that from the Discord API, editing the original message should be its own type of response, but I wasn't able to do it with dpy so I ended up with making two API calls

[fallen patrol]

you were using inter.response.edit_message, I presume?

[cold island]

I was trying interaction.edit_original_message

[cold island]

Is this what I'm supposed to be trying?

[fallen patrol]

yes

[cold island]

Hmm. This is confusing af

[fallen patrol]

this edits the response to the interaction

[cold island]

😂

I hope you see why these naming choices seem problematic to me

[fallen patrol]

tldr to respond you need to use one of the messages on the InteractionResponse object

then the other methods are usable

[cold island]

Alr, thanks

[cold island]

interaction.edit_original_message -> edits the response message
interaction.response.edit_message -> edits the original message

[full fractal]

a wa

[vocal wolf]

a wa

[severe tangle]

what the

[solemn vine]

what's absolute referencing ?

relative imports are erroneous afaik right ?

[gritty wind]

Is this part of this server’s docs?

But no relative imports are valid

A lot of people struggle with the import system and tend to use hacky methods to avoid actually understanding and fixing their systems, but in a well configured system it’s fine

[solemn vine]

i see

ll look at it more

[solemn vine]

yeah its part of style guide referenced in contributing guide

[gritty wind]

That sounds about right

[solemn vine]

hmm

[gritty wind]

I see I also missed a part of your question, absolute imports will use the name of the package in the import itself

[solemn vine]

do you know any source where one can learn about imports in depth

[gritty wind]

In my experience, it’s best to get hands on experience with it to see what does and doesn’t work, but let me see

[solemn vine]

is it also possible to import without package name ?

oh maybe like import *

or importing a already imported library(in file one) from file one

right ?

[gritty wind]

We don't allow import * as it creates a mess in terms of scoping for the file

But that's not what's relative and absolute

An example would better demonstrate this:

In our bots, we have all our source code under one bot pacakge (folder)

# To import bot/utils.py
from bot import utils

# utils.py might have a function such as send_message
# we can call it as
utils.send_message()

# We could also do:
from bot.utils import send_message
send_message()

This is all absolute because we start the path from bot.
A relative import would look more like:

If you have a file bot/logic.py, and wanted to import the utils file, you could also do:

from . import utils
# or
from .utils import send_message

We almost never use the latter in our codebase though

[solemn vine]

nice

is it not good way of doing it or just a convention for project

[gritty wind]

Are you referring to relative imports, or star (*) imports?

[solemn vine]

and what if the package name is something like "sir lancebot"

[solemn vine]

relative imports

[gritty wind]

sir lancebot would technically be a valid package name, but you can't import it because the import statement wouldn't parse it because of the space. You have a few solutions in that case:

1. Rename the package. Anything that's a valid variable name would work. For instance: sir_lancebot.
2. Use relative imports
3. Directly import it using importlib which can accept a string, but this has it's own mess and issues, so ultimately option 1 is preferable.

We just call it bot for sir-lancebot though

[solemn vine]

oh ok cool

[gritty wind]

Relative imports are fine. We usually don't use them because it can make things harder to understand and track, especially since things can move around.

They do have their uses though, and they are used here and there. I personally only find them useful if you have a deeply nested structure that would make it difficult to type out and read the entire path

I've just finished skimming the real-python article on imports. They are usually pretty good, if a little long (this one is really long), and I can't see anything that stands out here. https://realpython.com/python-import/

For all the technical and in-depth aspects, I suggest:
https://docs.python.org/3/reference/import.html
Specifically, 5.2.1, 5.3 first section, 5.4 first section, 5.7, and 5.8 would cover most of what you need to know as a user.

If you're already familiar with packages in other languages, the docs are much shorter and straight to the point than the realpython article. The python import system is pretty similar to other languages.

[solemn vine]

thanks you so much for helping out

[cedar frigate]

quick question .. I am pressing on with the idea of doing a Sir Lancebot lyrics command. is it better to use an API (with the downsides of a new API credential to manage and another quota) to fetch lyrics, or is it preferred to scrape a web page (assuming such a thing exists) so no additional API keys/tokens are required? still deciding what to use.

..or option 3 - scrap the idea 😄

[tawdry vapor]

If the API credentials don't expire then I do not think it's much of a burden to manage - there really wouldn't be anything to manage (just create creds once for production and forget about them).

[cedar frigate]

ok 👍 thanks

[tawdry vapor]

If all the APIs have quotas then it's probably not going to have much effect on users since I doubt the command will be used too much.

If anything it would affect the implementation since you'd need to respect rate limits and whatnot. Whether you're willing to deal with that is up to you.

Or if you are really keen on doing something with webscraping then that's fine too.

Web scraping just tends to be less stable.

Anyone else have an opinion?

It's sir lancebot so as far as I'm concerned you can do whatever you want within reason.

[cedar frigate]

Thanks for the pointers. Based on that, and real world constraints encounteted while exploring the Genius API, I think I found a way forward:
Use the Genius API to make a search query. This gives back some metadata along with a URI pointing to the song lyrics on their web site. This must then be scraped, as the API doesn't provide an endpoint to get the actual text. But luckily it is a simple scrape one request and bs4 can easily handle.
I'm working on it on my own fork of sir lancebot, though I it's not yet PR-ready.

[gritty wind]

I’ve looked into this myself a while back

If memory serves they don’t allow you to scrape them as per their ToS

In fact, I basically found no free APIs

With the recruitment of staff, we did manage to stumble upon one that was undocumented, and the only reference on the internet to it was one SO post comment

So I’m willing to bet that’s some shady stuff

But you can just link the lyrics link returned by the genius API

You dont need to get the lyrics all the way to discord

[placid ermine]

hahahha i remember

[gritty wind]

A few weeks later Spotify literally just implemented the exact same thing

One of y’all is a spy and I want my cut

[surreal veldt]

what are the open sir lancebot issues or PR's that one can contribute to?

also, were all the #discord-bots pins moved to the website guides yet?

[hoary haven]

not yet, that project is on hold for a bit

[hoary haven]

i would search for approved issues and start with ones that are unassigned or that are up for grabs
https://github.com/python-discord/sir-lancebot/labels/status%3A approved

be sure to check out the same for our other repos as well :)

if you see one that interests you add a comment asking to be assigned to it (or ask any questions you may have about it) and a core dev will respond in due time

[surreal veldt]

What happened to it?

[hoary haven]

i'll be looking for a new lead, frontogenesis had to step down due to personal life stuff

but at the moment we have some more time-sensitive things that require planning, such as code jam prep

[surreal veldt]

Who can become project leads?

[hoary haven]

from staff members, sometimes we do an open call asking for someone who is interested, other times we have specific staff members in mind that we recruit/ask to become project lead
i haven't decided yet what to do for this specific project

[surreal veldt]

Is there any update on Python bot using slash commands and time soon?

Specifically the docs command

[vocal wolf]

@fallen patrol oi m8 status on these PRs? sir-lancebot#982 sir-lancebot#981

[dusky shore]

GitHub

Interaction tic tac toe
fix: subcommands inherit their parent's whitelist

[fallen patrol]

!remind 3d are you free again

[stable mountain]

Can do!

Your reminder will arrive on <t:1655868894:F>!

ID: 4650

[vocal wolf]

Very well

[fallen patrol]

that's a question for me

basically busy rn

[vocal wolf]

ty for response

[austere hornet]

@fallen patrol I believe you were also working on sir-lancebot#968. Are you still planning to? Would love to see it implemented.

[dusky shore]

GitHub

github command enhancements

[fallen patrol]

see above reminder but for that command in specific i want to do something like this https://github.com/onerandomusername/monty-python/blob/a96d151fdb34392eceb8306eb65b0fa02ce10185/monty/exts/info/github_info.py

[austere hornet]

Oh that looks nice, yeah I saw the reminder

[surreal veldt]

bump

[cold island]

Haven't really discussed it. Also dpy is still undergoing breaking changes (especially slash commands) so we're not in too much of a hurry

[surreal veldt]

Is this a good suggestion:
Allow users to send two documentation embeds at once by doing like
!d x, y will send two embeds

[surreal veldt]

I think it's pretty cool, if anyone agrees, am I allowed to make a PR?

[brisk brook]

Wouldn't a space be enough because there cannot be any spaces in names

[brazen charm]

!d not in

[stable mountain]

not in

6.10.2. Membership test operations¶

The operators in and not in test for membership. x in s evaluates to True if x is a member of s, and False otherwise. x not in s returns the negation of x in s. All built-in sequences and set types support this as well as dictionary, for which in tests whether the dictionary has a given key. For container types such as list, tuple, set, frozenset, dict, or collections.deque, the expression x in y is equivalent to any(x is e or x == e for e in y).

For the string and bytes types, x in y is True if and only if x is a substring of y. An equivalent test is y.find(x) != -1. Empty strings are always considered to be a substring of any other string, so "" in "abc" will return True.

[brazen charm]

I don't think parsing multiple symbols is worth it, as compatibility with sending a message after the symbol should also be kept and would complicate things more for not that much of a benefit

[surreal veldt]

Well, I think an example of where it'll be useful can be #discord-bots

Although I don't get

compatibility with sending a message after the symbol should also be kept

Oh nevermind I get it

Can't something like how the rule command works be done?

You can have multiple rule indices with a text after the indices

[brazen charm]

rules don't need to deal with the ambiguity of what's a word from the message, and what is a doc symbol. It can definitely be done, but I'm just not of the opinion that it'd be worth it considering you can wait a bit and invoke the command again if it's necessary to send multiple embeds

[cold island]

Looking for one more approval on https://github.com/python-discord/site/pull/726

[severe tangle]

#roles message
Fore instead of For

PyWeek Champions Embed

[vale ibex]

hah thanks, that one has been in there a while apparently

[surreal veldt]

the count function is not in the !docs?

[vocal prairie]

most list methods are not in the objects.inv file (what we get our docs from) for some reason (I'm afraid I do not remember what it was)

[tawdry vapor]

It's just the way the Python docs are structured

[vale ibex]

Yup, there's a bug for it https://github.com/python/cpython/issues/89923

[surreal veldt]

is it something that can be fixed, or due to what Mark said, it can't?

[vale ibex]

It's something that needs to be fixed in the CPython codebase itself

hence the CPython bug issue

We could special case it in our docs command with some amount of effort, but we haven't wanted to

[cold island]

head scratch does anyone have an idea what's going on here? I copied it from last year's page. It works there, here it gives this error

[vale ibex]

Is there a {% load staticfiles%} in the file?

I think just {% load static %} works too

needs to be after any extends tags

[cold island]

Aha!

Thanks

[surreal veldt]

why does the !eval not support the help() method?

[gritty wind]

So python has a site module which includes a couple things, one of which is help. You can read more about that here:
https://docs.python.org/3/library/site.html

Snekbox disables it as a consequence of passing the -S option to avoid exposing snekbox’s dependencies (dependencies of the API calling snekbox not the evaluation layer, such as falcon or gunicorn)

You can still manually load it and execute it to get it if you need it

[surreal veldt]

what does this mean? You can still manually load it and execute it to get it if you need it?

[vale ibex]

you can do import site in your snippet

[gritty wind]

import site; site.main()

This at the top of your script

[surreal veldt]

site.main() does what?

[gritty wind]

It’s in the docs. It adds all the standard search paths removed by the S option, which adds back what was removed

Chris was more right, you don’t need to call main yourself. It’s called when importing

Wait no

You do need to call it

[surreal veldt]

so i did it in #bot-commands but it error'd again

[vale ibex]

yea you need to call .main() since we use -S

[surreal veldt]

oh okay thanks

what's the limit of lines in the eval command output?

[gritty wind]

Ten lines or 1000 characters

Whichever comes first

[surreal veldt]

how'd I get 11 in #bot-commands?

[gritty wind]

It counts the number of \n which is one off from actual lines

Meaning you get > 11

[cold island]

Reviews welcome https://github.com/python-discord/site/pull/727

Also, why did the static preview fail 😦

[gritty wind]

GitHub is ratelimiting us

But their API does not allow us to get a read-only token, so the only way around the limits is giving netlify a write token, so anyone that modifies the build script can write anything to our main branch on all our projects 😄

[cold island]

wonderful

[gritty wind]

I’m thinking of having an intermediate which uploads the build off GitHub so we can pull it more easily

It’ll simplify like 3-4 steps in the script

I just don’t know where

Dropbox lmao

[vale ibex]

could rsync it to netcup servers

we've got a shit load of storage that we will likely never fill

[gritty wind]

Would that pose a security risk for PRs

[vale ibex]

depends how much we can lock down rsync

[subtle kraken]

what about uploading it to random discord server and pulling from there?

[vale ibex]

lol

[gritty wind]

Actually

Github apps might work for this

[vale ibex]

what would be the attack vector though?

you cna place a file that gets built from an action to our servers?

and then have a github builder do something with it

[gritty wind]

I don't know, my jank detector just fired up automatically when it detected "Warning: Uploading untrusted content to prod servers"

[vale ibex]

hah yea

[gritty wind]

Well you can rewrite the action to upload anything

[vale ibex]

it doesn't get ran nativly though right

[gritty wind]

The vector is basically unlimited access to a rsync to our servers

[vale ibex]

unless you can also change the path to overwrite something like /etc lol

[gritty wind]

This is where my lack of knowledge of rsync limits me

[vale ibex]

it should be fine if
1- we hardcode the path
2- we require devops approval to change the action

[gritty wind]

@patent pivot could you look into github apps? Can we get one with just read access to our public repos, and nothing else

2. Is that something github can do

Specifically limit it to changes

[vale ibex]

we have it via policybot atm

[gritty wind]

I don't follow

I mean you can cause the exact same damage on a PR or master

You don't need to merge for the vector to apply

[vale ibex]

actions are always run from main

and changes to the actions in main require a PR with devops approval

[gritty wind]

I still don't follow

Like you have the action, which runs on PRs, and has access to a token

Modifying it does change what the PR runs

[vale ibex]

Yea, but all the action does is push a build artifact to the server

on a hardcoded path

[gritty wind]

Right, so 2 is not a real possibility then

Just 1

[vale ibex]

wdym not a possibility, we already have it?

[gritty wind]

I mean you can do the exact same thing on a PR that you can on main wrt the concerns from having an rsync anyone can access

[vale ibex]

if you change an action in a PR, the actions that run are still the ones that are in main

[gritty wind]

That isn't true though

I mean, I can pull up any of my PRs which add or change actions

They run too

Sphinx-multiversion PR added a build step to the action, which was run on the PR
https://github.com/python-discord/bot-core/pull/79

It added a whole new action as well

[cold island]

Hello two wonderful core devs. Would you like a nice content-only PR to review to boost your stats?

[gritty wind]

Diff is too small, won't make a dent

Get gud

I'll try to review it once netlify works again

I'll rerun

Nope, still 403

[vale ibex]

ahhhh, ok I see why

any PR that uses pull_request as a target, and then checks out the head branch uses the action from that branch

So yea, it would open us up to any staff member being able to change things

[gritty wind]

Wouldn't it be any non-first time collaborator?

Actually, we don't have that limit enabled, so any collaborator

[vale ibex]

apparently ```
Only users with write access to your repo can push to and create new branches in your repo, so it is assumed that they are trustworthy.

If the head branch is in a forked repo (external contributor), there is no access to secrets in the workflow. They will simply be not set. This applies to on: pull_request

https://github.community/t/can-pull-requests-change-actions/143120/2

If that's the case then external contribs just wouldn't get previews then

[vale ibex]

😢

this isn't from your changeset though

no idea when it was added

also inconsistent capitalisation

[gritty wind]

I think apps might be our best bet for this

I'm experimenting right now

But you can get very specific with what you enable and disable

And everything can be set to read-only, or read/write

So far I've figure the app will need read-only to: Metadata (mandatory), actions, pull requests

What each of those enables is listed here:
https://docs.github.com/en/rest/overview/permissions-required-for-github-apps

That will be installed per-repo

[vale ibex]

That does sound promising

[cold island]

Oh hey the preview succeeded this time

[gritty wind]

Alright, I think I found a blocker for apps

To use the API with them, you need to generate a JWT, which is only allowed to last a max of 10 minutes. Generating this means keeping the app's private key accessible to the netlify build, or hosting an independent signing server to generate the JWTs (this is a no go).

If the worst that can be done is generating a JWT with the read only scopes, then we can live with that. I'm not sure if this exposes us to anything else though, that's what I'm figuring out now

Yeah this whole thing seems like a dead-end

That's a shame

[gritty wind]

@vale ibex I've documented all the things I've thought of/tried so far. Would appreciate if you could give it a look and add your thoughts

site#728

[dusky shore]

GitHub

Deploy Preview Failures

[vale ibex]

Sure I'll give it a read tomorrow

[gritty wind]

Thanks

[stable mountain]

It has arrived!

Here's your reminder: are you free again
[Jump back to when you created the reminder](#dev-contrib message)

[timber meadow]

so i wanted to update the discord logo in the navbar for python-discord/site but ive hit a roadblock.
if i use the discord logo as-is, the new blurple color doesnt match with the rest of the site's colors, so it looks a bit off, and discord.com/branding says

Please do not edit, change, distort, recolor, or reconfigure the Discord logo.
i guess this means i cant change the color to match with the rest of the site

further, i couldnt find the new discord logo font to use for the "Join Us" text

so, this leaves 4 options:

1. dont change anything
2. update the logo and every other bit of branding with discord's new blurple color
3. just update the logo without modifying it as per discord's requirements
4. update the logo and change the color
   (the question of the font still remains)

personally i think option 1 would be the easiest and safest. only after spending a good amount of time updating the svg did i realize that the new logo doesnt match the colors haha (was about to open a pr too). i guess this is why no one has bothered to update it in the past year, and my apologies if this has already been discussed. i would like to get some higher-ups input on this, thanks.

[tawdry vapor]

When Discord changed their branding, I think it was decided that our site would not follow suit.

[timber meadow]

very well, thank you

[gritty wind]

We don’t need to use the background from the discord logo, we can just use the actual icon, and place it on our own background

Aren’t we already using the latest icon tho

[vale ibex]

the website is using the old Discord icon in the join us button

https://github.com/python-discord/site/blob/main/pydis_site/static/images/navbar/discord.svg

[vale ibex]

fwiw, in the brand guidelines they say they use two fonts Ginto Nord and
Whitney. They also included a download link

https://www.dropbox.com/sh/j2eq5qobyqm1v6i/AACt6OBMB0TvI0ZO8D5wTJi9a?dl=0

page 41 of the guidelines

[stable mountain]

Hey @timber meadow!

It looks like you tried to attach file type(s) that we do not allow (.svg). We currently allow the following file types: .gif, .jpg, .jpeg, .mov, .mp4, .mpg, .png, .mp3, .wav, .ogg, .webm, .webp, .flac, .m4a, .csv, .json.

Feel free to ask in #community-meta if you think this is a mistake.

[timber meadow]

heres my updated svg. i recolored the logo to match the site though which probably isnt ideal (screenshot because bot didnt let me upload svg)

[gritty wind]

We’re using the grayscale icon on mobile, so I assumed we were doing that on desktop

Don’t see why we couldn’t

It would be inline with their policies, and won’t require clashing colors

[vale ibex]

Yea, that's the fontawesome icon

which is updated for free

[gritty wind]

That’s probably why we’re on the latest for mobile (and forms 💪)

[vale ibex]

we wanted to be fancy on desktop

a nice animated, sliding svg

[vale ibex]

Cool, and I guess for the transparent part of the svg you're just using the discord full white logo?

[atomic ivy]

new help channels sometimes show up at the top of the list. #help-peanut is a newly claimed one now, for example

[gritty wind]

The placement logic is as simple as it gets. Request the category, read the number of entries, and set the position to that +1

The only way that fails is when discord returns wrong data or just does not respect our request

I don’t see a way to fix it

[atomic ivy]

I see. usually people don't check channels at the top, so the helpee may not get the help, but yeah, I understand

[gritty wind]

It might also be caused by our cache not being correct, but that’s managed by d.py so there’s no clear action there either

[timber meadow]

yes, same logo but white

[vale ibex]

this one?

[timber meadow]

i didnt change the text at first

i can try that and see how it turns out now that i know the font name

[vale ibex]

you can just use the svg directly in their press pack

it has the correct padding etc

[timber meadow]

what about the "join us" text?

[vale ibex]

For the colour one, you can use the Mark only logo, might be worth seeing what that looks like with the discord font

for the white side, I'd just use the full white logo

[timber meadow]

i think if i change one half i should also change the other for consistency

alright i'll do that right after breakfast

[vale ibex]

Cool cool, thanks :D

[timber meadow]

not final

grey background is just for visibility

this is with the old text

[vale ibex]

I think we definitely want the new text on the right

I'm more interested to see what the new text looks on the left

[vale ibex]

I like that

[timber meadow]

so what about the color?

[vale ibex]

I'd need to see it in-app to make a call on that

preferably we use the discord colours as stated in the branding guidelines

[timber meadow]

agreed

[vale ibex]

Are you familiar with git/github?

[timber meadow]

yes

[vale ibex]

alright awesome, if you open a PR to the site repo, replacing pydis_site/static/images/navbar/discord.svg our CI will create a deployment preview

we can see what it looks like in the site then

[timber meadow]

perfect

[timber meadow]

will it create a preview for draft PRs as well?

[vale ibex]

I think so

[timber meadow]

guess we'll find out

[vale ibex]

hah yea

[timber meadow]

looks like its building something https://github.com/python-discord/site/pull/729

[vale ibex]

yup

😎 Deploy Preview deploy-preview-729--pydis-static.netlify.app

from the comment

on hover seems to be falling back to a default font

[timber meadow]

works fine for me

oh probably cause i have the font installed

[vale ibex]

I'm guessing you have the font installed on your PC :P

[timber meadow]

maybe i can fix that in inkscape somehow

[vale ibex]

I think we either need to serve a .ttf file, or we use a fallback font in the svg itseelf

from the branding guides it says Ginto Nord could be replaced with Poppins Black (all caps). Whitney could be replaced with Roboto Normal.

as fallback fonts

[timber meadow]

i think i can just convert the text to paths in the svg

[vale ibex]

I am really unfamiliar with typefacing in svgs

@crude gyro @exotic ember I see you're both online do either of you have time to assist?

Picking on you since I know you speak svg

[timber meadow]

pff it was as simple as clicking one button 🤦‍♂️

[vale ibex]

oh lol

[timber meadow]

and running one google search to know which button to click of course

[crude gyro]

yeah convert them to paths.

[crude gyro]

I'm not as online as I look.

[vale ibex]

that looks real nice

https://deploy-preview-729--pydis-static.netlify.app/

[crude gyro]

screenshots? on mobe

[vale ibex]

white is on hover

[timber meadow]

the color still bugs me

[vale ibex]

Yea, Discord updated their colours, but we stuck with our own to be more distinct from Discord themselves

[crude gyro]

their new font is so so ugly.

[timber meadow]

i agree 100%

[vale ibex]

Yea, I dislike it too lol, but it their brand

[crude gyro]

Discords whole new branding is godawful, honestly, but it's nice to update this so it uses it anyway.

[vale ibex]

we could change the font for join us though

since that's not technically part of their branding, but it may make it look worse

[timber meadow]

we could also replace the blurple with black as Scaleios suggested since that would be inline with their policies

[vale ibex]

Yea, that's a good point, might look good

[vale ibex]

hmmm, I'm not too sure about that to be honest, it does lose something without the splash of colour

[timber meadow]

true

[vale ibex]

I think the best version is the one currently in the PR

[exotic ember]

also not as online as I look, but give me 10 mins

[timber meadow]

just for fun, recolored with the old blurple color

[vale ibex]

Could you add those 3 screenshots to the PR for ease of access for people?

[timber meadow]

sure thing

[vale ibex]

Well, the most recent 2 screenshots and then a similar one for the "correct" blurple colour

[vale ibex]

no worries kosa, if you're busy it's fine

we've figured out the blocker, and just looking for colour opinions now

[exotic ember]

this looks the best to me, but its hard to tell without seeing it on the site

[timber meadow]

https://deploy-preview-729--pydis-static.netlify.app/
this is with the new blurple color. i agree that the old color looks better, but the question then is, are we fine with violating discord's branding policy by recoloring the logo?

[full fractal]

*gasp*, review has been requested from me 👀

[vocal wolf]

yes.

[crude gyro]

It is a step up from what we've got today, compliance-wise. and it's a pretty subtle difference. I think it's probably fine. if anyone complains, we can change it.

I do feel like the capital U in Join Us bothers me though.

why would that letter be uppercase

I can't think of any justification for it

[gritty wind]

Not a big fan of title case I take it

[crude gyro]

Not When Used For Things That Aren't Titles, No

[gritty wind]

Hah

[crude gyro]

just looks like a motivational poster from grandma

Your A Star And Believe In Urself!! 🌟

thanks grandma

[timber meadow]

better?

[crude gyro]

yeees

now I like it a lot.

[timber meadow]

i'll commit this and un-draft the pr

[vale ibex]

It's merged and live! https://www.pythondiscord.com/ Thanks again :D

[rapid swallow]

I'm able to run the Python Bot on docker using cellular data but not using my university wifi. I have to connect to their proxy to access internet.
Docker daemon works FWICT ( docker login worked after I followed https://docs.docker.com/config/daemon/systemd/#httphttps-proxy ).

I tried https://docs.docker.com/network/proxy/#configure-the-docker-client but the container doesn't connect to the site (or even attempt to? idk, I'll send the screenshot)

left: when I connect to uni wifi through their proxy
right: when I connect the laptop to mobile hotspot

oh, and when i type http://0.0.0.0:8000/ in my browser, the website loads when I'm connected to mobile hotspot but not when I'm connected to the uni proxy

ERROR
The requested URL could not be retrieved

The following error was encountered while trying to retrieve the URL: http://0.0.0.0/8000

    Access Denied.

Access control configuration prevents your request from being allowed at this time. Please contact your service provider if you feel this is incorrect.

Your cache administrator is *****

this is what I get

I havent tried running docker in windows yet, this is in linux

[rapid swallow]

I tried https://docs.docker.com/network/proxy/#configure-the-docker-client

ping me when replying \🥺
I'm gonna sleep ✌️

Scaleios, didn't you have to wrestle with your uni's proxy?

[vale ibex]

0.0.0.0:8000 is internal to the docker container. It will be available on 127.0.0.1:8000 on your local machine

[vale ibex]

Also, from this output, it seems the bot connected to the site just fine, what makes you think it's not working?

[full fractal]

https://github.com/python-discord/sir-lancebot/pull/1049 to be honest I feel unqualified to review this given the core devs don't seem to agree this feature in its current design is a good idea

[fallen patrol]

Hi where can i report a snekbox exploit pls kthx

@patent pivot already knows about it but its been quite a few months and I see no progress so I want to get it officially on the books somewhere so it can be patched

[patent pivot]

uhhh

DM me it again to remind me what it is

[full fractal]

Hey core devs, I'd like to submit a PR to sir-lancebot without an issue for this formatting quirk (the double spaces after the parent command) nvm, filed an issue https://github.com/python-discord/sir-lancebot/issues/1065

Subject: [PATCH] Help: remove redundant space in subcommand aliases

`parent` already has a trailing space so let's not add *another* one.
---
 bot/exts/core/help.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/bot/exts/core/help.py b/bot/exts/core/help.py
index db3c2aa67..f9b3513fe 100644
--- a/bot/exts/core/help.py
+++ b/bot/exts/core/help.py
@@ -306,7 +306,7 @@ async def _add_command_signature(self, paginator: LinePaginator) -> None:
         signature = self._get_command_params(self.query)
         parent = self.query.full_parent_name + " " if self.query.parent else ""
         paginator.add_line(f"**\`\`\`\n{prefix}{parent}{signature}\n\`\`\`**")
-        aliases = [f"`{alias}`" if not parent else f"`{parent} {alias}`" for alias in self.query.aliases]
+        aliases = [f"`{alias}`" if not parent else f"`{parent}{alias}`" for alias in self.query.aliases]
         aliases += [f"`{alias}`" for alias in getattr(self.query, "root_aliases", ())]
         aliases = ", ".join(sorted(aliases))
         if aliases:

Fixing this should only (hopefully) require this tiny change.

[outer oasis]

What's wrong with creating an issue?

(just curious)

[full fractal]

this felt like lower overhead i.e. less work

but I guess in hindsight this is more work since I have to wait

I'll just file an issue instead.

[rapid swallow]

"attempting site connect" doesn't show up

And cogs aren't loaded

[rapid swallow]

Okie

But the bot doesn't come online 😩

I'll have to connect to my phone's hotspot everytime I use docker then

[rapid swallow]

127.0.0.1:8000 works, ty

but the bot still doesn't come online on discord

(when I'm using cellular internet/mobile data/mobile internet by connecting to my phone's hotspot)

back to sleep I go

for in sleep, uni's proxy cannot fuck with me

[gritty wind]

I didn’t have issues with the bot no

But it seems the site itself is up, it’s just the bot that isn’t starting

[vale ibex]

What command are you using to start it?

[rapid swallow]

Docker compose up

[cold island]

docker compose or docker-compose?

[gritty wind]

It’s the same on windows

Could you pull the logs for just bot from the docker desktop app

(When it fails to start)

[fallen patrol]

how does pydis do zero downtime updates with k8s?

specifically with the bots here

[cold island]

Wasn't the same for me when I tried it

[gritty wind]

Ez, we don’t

[gritty wind]

It should be according to the docs

[cold island]

Should be

[gritty wind]

I mean

What was different for you

[cold island]

Not sure, it just failed to connect to site

[gritty wind]

Sounds unrelated or like a misconfiguration. Right now it should basically just be an alias

[cold island]

Well it definitely didn't work with one and worked with another

[vale ibex]

Yea, docker compose and docker-compose are different apps, even on windows

docker compose doesn't work for me on the bot

[vale ibex]

yea, as scale said we don't for the bots. It's non-trivial to figure out when the new bot is "ready" and cut off the old bot, without having overlap where commands may get double responses

[fallen patrol]

ah

[gritty wind]

They are different executables, but they should just work as a drop in replacement. Strange it doesn’t work for y’all though, I suggest trying to figure it out because the docker-compose command still uses the old version, and I imagine it’ll be dropped all together eventually

I haven’t had any problems using the one with a space tho

[rapid swallow]

The former

[cold island]

Then yeah try docker-compose

[rapid swallow]

I'm not on windows

[gritty wind]

Its on other OSes too as part of the normal cli, just most people haven’t updated docker in 20 years

But the same holds, could you pull the logs from the docker logs command

[rapid swallow]

Yeah i will

2022-06-26 03:25:32 | root | INFO | Cog loaded: DocCog
2022-06-26 03:25:32 | root | INFO | Cog loaded: ThreadBumper
2022-06-26 03:25:32 | bot.exts.backend.sync._syncers | INFO | role syncer finished: created `0`, updated `0`, deleted `0`.
2022-06-26 03:25:32 | bot.exts.backend.sync._syncers | INFO | Starting user syncer.
2022-06-26 03:25:32 | root | INFO | Cog loaded: Reminders
2022-06-26 03:25:32 | root | INFO | Cog loaded: Infractions
2022-06-26 03:25:32 | root | INFO | Cog loaded: Filtering
2022-06-26 03:25:32 | bot.exts.backend.sync._syncers | INFO | user syncer finished: created `0`, updated `0`.
2022-06-26 03:25:32 | root | INFO | Cog loaded: Sync
2022-06-26 03:25:32 | root | INFO | Cog loaded: DuckPond
2022-06-26 03:25:32 | bot.exts.moderation.incidents | DEBUG | Crawl task finished!
2022-06-26 03:25:32 | bot.exts.moderation.defcon | INFO | DEFCON synchronized: -
2022-06-26 03:25:32 | bot.exts.help_channels._cog | INFO | Cog is ready!
2022-06-26 03:25:32 | root | INFO | Cog loaded: HelpChannels
2022-06-26 03:25:32 | bot.exts.info.pep | INFO | Successfully refreshed PEP URLs listing.
2022-06-26 03:25:32 | root | INFO | Cog loaded: PythonEnhancementProposals
2022-06-26 03:25:33 | root | INFO | Cog loaded: Big Brother
2022-06-26 03:25:34 | root | INFO | Cog loaded: PythonNews
Found `config.yml` file, loading constants from it.
2022-06-26 12:50:50 | botcore.utils._monkey_patches | DEBUG | Patching send_typing, which should fix things breaking when Discord disables typing events. Stay safe!
2022-06-26 12:50:50 | discord.client | WARNING | PyNaCl is not installed, voice will NOT be supported

[gritty wind]

Is this on the uni network?

[rapid swallow]

yep

[gritty wind]

I thought you didn’t get any bot logs on the network

Oh I see, these are from two separate boots

[rapid swallow]

nah, it's on uni network
i thought it's an earlier boot but i forgot the logs were in UTC

[gritty wind]

You have logs from 03:00 and 12:50

[rapid swallow]

oh

[gritty wind]

If that’s all in one boot, that’s quite the startup time haha

That last log line is from when we init the client

That happens on line 55 of __main__.py

Could you add some debug print statements around there to figure out if it gets through that part

[rapid swallow]

2022-06-26 13:10:45 | botcore.utils._monkey_patches | DEBUG | Patching send_typing, which should fix things breaking when Discord disables typing events. Stay safe!
2022-06-26 13:10:45 | discord.client | WARNING | PyNaCl is not installed, voice will NOT be supported

so this is pretty much what i got

[rapid swallow]

If i leave it running for a long time I get

Traceback (most recent call last):
  File "/bot/bot/__main__.py", line 73, in main
    await _bot.start(constants.Bot.token)
  File "/usr/local/lib/python3.9/site-packages/discord/client.py", line 682, in start
    await self.login(token)
  File "/usr/local/lib/python3.9/site-packages/discord/client.py", line 543, in login
    data = await self.http.static_login(token.strip())
  File "/usr/local/lib/python3.9/site-packages/discord/http.py", line 557, in static_login
    data = await self.request(Route('GET', '/users/@me'))
  File "/usr/local/lib/python3.9/site-packages/discord/http.py", line 439, in request
    async with self.__session.request(method, url, **kwargs) as response:
  File "/usr/local/lib/python3.9/site-packages/aiohttp/client.py", line 1138, in __aenter__
    self._resp = await self._coro
  File "/usr/local/lib/python3.9/site-packages/aiohttp/client.py", line 634, in _request
    break
  File "/usr/local/lib/python3.9/site-packages/aiohttp/helpers.py", line 721, in __exit__
    raise asyncio.TimeoutError from None
asyncio.exceptions.TimeoutError

[rapid swallow]

nope, doesn't work

[gritty wind]

Yeah there’s no reason for that to work

What zig and Chris experienced is most likely unrelated to this

Anyway, that just shows it’s failing to log into discord

Could you create an empty d.py project which just calls bot.run

With the same intents as bot

[rapid swallow]

https://github.com/python-discord/bot/blob/main/bot/__main__.py#L71
it gets through till here

[stable mountain]

bot/__main__.py line 71

)```

[rapid swallow]

)
😎

[gritty wind]

It's just the under the hood call to discord login

Through d.py

(hence the tb)

Is there a good typehint for "pass this particular function", instead of a generic (args), return

[cold island]

Why do you need an argument if it's a constant value?

[gritty wind]

It's not, it's a choice between two funcs

[cold island]

Hmm

[gritty wind]

It's either discord.RemoveRole or discord.AddRole

Right now it's documented as a callable, and the docstring says pass one of these two

[cold island]

Literal?

Yeah the docs don't limit its use to strings

[brazen charm]

Literal would only work with an enum for something like that

just document it in the docstring

[gritty wind]

I've asked in #type-hinting

I'll see if something pops up there

[rapid swallow]

I'll try it later

[timid sentinel]

Should the (bot or sir-lancebot) help command with no arguments display

• All top level commands and their subcommands
• Only top level commands
• Top level commands and subcommands with root aliases
• Something else

I think our current behavour is option 3 but i'm not sure if that's on purpose or not. I'm thinking we probably just want option 2, would be nice to try and make it more concise.

[gritty wind]

Top level commands and aliases seems best for discoverability, but also 2 sounds fine if it's easier

[timid sentinel]

I think all should be doable, was just thinking that the current 11-page output is a bit of a monster that hurts discoverability more than removing some subcommands would.

It's possible it could do with more of a redesign to try and make it shorter anyway though, idk

[cold island]

Keep in mind that regular users see the commands they can run, which is much more limited

[gritty wind]

@vale ibex Any idea why this doesn't have an await? Or how it's working haha
https://github.com/python-discord/bot-core/blob/e28b42f9b1e085df6961d195abe6a36210d25242/botcore/_bot.py#L200

[stable mountain]

botcore/_bot.py line 200

self.log_to_dev_log(msg)```

[vale ibex]

it probably doesn't work lol

but that condition rarely ever gets hit

[gritty wind]

Fair enough

I'm fixing a couple misc bugs like that

[vale ibex]

nice nice

[cold island]

@timid sentinel for non-staff the help command gives 3 pages

[gritty wind]

This is new

The request to nightly.link is timing out

It's usually GH that does that

It downloads fine for me hmm

Ah they change the url

[gritty wind]

I've opened an issue there since it seems to be an actual issue

[gritty wind]

Hey @brazen charm what's the generic equivalent for typing.Type? Seems to just be type, but that produces a warning Class 'type' does not define '__getitem__', so the '[]' operator cannot be used on its instances

It also seems to not like collections.abc

Not sure what's up with that

bot-core\botcore\utils\caching.py:docstring of botcore.utils.caching.AsyncCache.__call__:6:py:data reference target not found: collections.abc.Callable

[brazen charm]

type should work

and Callable from collections.abc is unfortunately bugged in the doc typehint package

[gritty wind]

Ahhh

Well we do have the ignore for this reason then

The type thing is probably just pycharm being funny

[brazen charm]

would it be referenced correctly with an ignore?

[gritty wind]

It wouldn't, but the warning will be silenced

[brazen charm]

It seemed to be an easy fix and I posted an issue, but I didn't get a reply there

[gritty wind]

Heh

Could you link it?

[brazen charm]

tox-dev/sphinx-autodoc-typehints#237

[dusky shore]

GitHub

data role used for collections.abc.Callable while class is required

[brazen charm]

Callable uses a different role in typing for some reason, and it forces that different role for collections.abc.Callable too in which case it's wrong

[gritty wind]

Thanks

Hope you get a response soon

It's kind of unfortunate right now

I might just migrate everything else and leave this

[brazen charm]

I just used callable from typing in the PR I have open for now

[gritty wind]

That's fine

Speaking of

I wrote something to complement your solution

I added a hook for missing-reference which pulls anything from our project and gives a raw link instead of leaving it hanging

I've gotta remember to throw that up some time

[timber meadow]

looks like fontawesome fa-duck and fa-alien-monster aren't showing up on pythondiscord.com
any idea why? i searched for both on fontawesome.com under 6.1.1 and it found fa-duck but not fa-alien-monster

[exotic ember]

thanks for letting us know, I'll work on a fix

[timber meadow]

also the alignment of the icons is a little off, i think a value of 30px instead of 33px for line-height would be better
https://github.com/python-discord/site/blob/main/pydis_site/static/css/home/index.css#L130

[stable mountain]

pydis_site/static/css/home/index.css line 130

line-height: 33px;```

[exotic ember]

PR opened here: site#731

[dusky shore]

GitHub

Fix interactive timeline icons.

[placid ermine]

known issue with pycharm

[cold moon]

Anyone using GitHub Codespaces? If I create codespace from one of PyDis repos, I get DNS_PROBE_FINISHED_NXDOMAIN (MS Edge on Mac).

[outer oasis]

Hmm
Are you able to access the base github.dev domain?
What's the domain for your codespace?

[cold moon]

Somehow, magically, it now started working.

[outer oasis]

🎉

[gritty wind]

@brazen charm re: bot-core#91

Have you had luck with setting the typehint in the docstring? It can't resolve the type var (different error from the other one you had a while back)

[dusky shore]

GitHub

Miscellaneous Bug Fixes

[gritty wind]

bot-core\botcore\utils\scheduling.py:docstring of botcore.utils.scheduling.create_task::py:class reference target not found: TASK_RETURN

[gritty wind]

For some reason just putting TASK_RETURN evaluates as an absolute

Gotta give it the full qualified path

If I give it the full path, my resolver can link it's source code

But sphinx itself does not find it

[brazen charm]

you could document and export the typevar for it to find it, but I think normally for ones that are only in a function's signature they're not linked to anything so it should be fine to ignore

Maybe also wrap it in typing.TypeVar(), not really sure if better but it's consistent with what it generates elsewhere

[gritty wind]

The missing-reference handler will catch it either way, so I'm thinking I can add a special condition for type vars to be ignored. What I'm wondering though is if we want to ignore it, or link source code

Both would be trivial

for reference:

I changed source code to source to match the one at the top

[gritty wind]

I've decided to shorten typevars, so now they'll look like this:

or that without the source link

[brazen charm]

I'd ignore them, the only thing the typevar can have other than its name is a docstring under it, and I'd assume if it has a docstring it's being properly done for sphinx too so it shouldn't be a missing ref

[vocal prairie]

!pep 0

[stable mountain]

Sorry, an unexpected error occurred. Please let us know!

AttributeError: 'Context' object has no attribute 'trigger_typing'

[vocal prairie]

^ this has happened for some reason

[outer oasis]

!pep 621

[stable mountain]

Sorry, an unexpected error occurred. Please let us know!

AttributeError: 'Context' object has no attribute 'trigger_typing'

[outer oasis]

huh

[placid ermine]

https://github.com/Rapptz/discord.py/pull/7979

[vale ibex]

wtf why

[outer oasis]

Only three results
Doesn't look terrible

[vale ibex]

bot#2203

[dusky shore]

GitHub

Use the new Messageable.typing instead of the removed Messageable.tri…

[outer oasis]

Four*
I'm dumb

[outer oasis]

Damn you're fast

[vale ibex]

ctrl_shift_f lol

[white light]

python hastebin is bias against typescript? It redirects .ts and .tsx to .py

[placid ermine]

i think we redirect everything to py

[vale ibex]

it redirects everything to .py, you can add ?noredirect query param to stop it

[white light]

.js works

[outer oasis]

I thought it would allow other extensions if you specified them

[vale ibex]

since we're a Python server, and the hastebin lang detection isn't the best, that's the most correct

[outer oasis]

Replace across files?
What editor?

[vale ibex]

oh, that's just project wide search

in vscode

[outer oasis]

Ah
Does it support regex?

[vale ibex]

yea

It's got match case and match whole word options

or regex if you want to be specific

[outer oasis]

Damn
It's really catching up to PyCharm these days

[vocal prairie]

you can change the extension to .ext?noredirect if you don't want to redirect to .py

[outer oasis]

I've been playing with CodeSpaces, I'll have to give it a more serious look

[gritty wind]

Pycharm does have the ctrl+shift+f for the same purpose

Actually, most editors do (atom, notepad++, etc) 😛

[vale ibex]

Yea

huh

github isn't triggering the lint test action

ah

yup

https://www.githubstatus.com/

github actions down

[thorny obsidian]

😔 is it ever up at this point?

[vale ibex]

It's what you get for using Oracle products lol

[outer oasis]

... GH uses Oracle?

[vale ibex]

MySQL

for all things non-git

[outer oasis]

Oracle made that?
Don't they have their own database product?

[vale ibex]

it was bought by Sun microsystems

so oracle got that when they acquired sun in 2010

Around the same time was when it was forked to make mariadb

[outer oasis]

Isn't that Java

Oh wait Oracle owns Java

[severe tangle]

Lol

So basically Oracle owns the most used Android app making language

[green oriole]

Hey, form unit test code didn't seem that broken

I got scared for a sec that I committed a french there but thankfully it was only @gritty wind
https://github.com/python-discord/forms-backend/blob/main/backend/routes/forms/unittesting.py#L46

[stable mountain]

backend/routes/forms/unittesting.py line 46

# Unite code```

[gritty wind]

hahaha

You know I wrote a transpiler for this year's qualifier

Fun stuff, fun stuff

[vale ibex]

it's very United

[green oriole]

you WHAT

[gritty wind]

You know the programmer mantra

Why spend an hour converting tests by hand

When I can spend 7 hours writing a program to do it

[green oriole]

oh no you didn't

did you write a script to convert a normal unit test to the JSON format?

[gritty wind]

It even gives you the JSON ready to go 🙃

[green oriole]

that is very funny and very stupid at the same time, and I'm all for it

[gritty wind]

Exactly haha

[green oriole]

what even is that global madness

do you have a sample of what it outputs?

[gritty wind]

Yeah one sec

The alternative was adding it to every test, and CF was already rejecting my payloads

The JSON, and the file that gets sent to snekbox

I don't think I leaked any state secrets

This should be from the public test suite

[green oriole]

truly amazing code

[gritty wind]

I'm unironically very proud

[green oriole]

honestly that's very good

[pseudo igloo]

👋

pleased to meet you all, I'm a bot dev & admin at libera.chat, interested in branching out in to discord

[vale ibex]

@gritty wind mostly because I can't think of a nice way to clear that state in the cog_command_error without impacting other calls

so it could become messy

[gritty wind]

Don't we have the cog that failed to unload in the error handler

If we have a dict of cog_name: state, I think it could work

But it's really not a big deal outside of help_channels

[vale ibex]

yea

[vale ibex]

TIL about git --force-with-lease

pretty neat

[outer oasis]

What does that actually do?
I've never understood it

[vale ibex]

it's like --force, but fails if it would overwrite commits from other authors

at least afaik

[outer oasis]

Sounds right
The part I get stuck on is if it supports doing partial stuffs
Does it fail to update any commits if just one of them is "wrong"

[vale ibex]

Yea, my understanding is the push entirely fails if it would discard another authors work

so it would make you reconsider, or use --force to actually do that

[outer oasis]

Sounds reasonable

[green oriole]

hi

the sentry error on form backend is me

I may have been messing with unit testing

it is a non issue really

[vale ibex]

I'm not seeing any recent sentry errors 😅

oh wait, there's a 500 axios error in front-end

guessing it's that one from 9 min ago?

[green oriole]

yeah

[vale ibex]

cool cool

[green oriole]

I got the unittesting to 99

[vale ibex]

I will ping joe and scale to fix immediately

thank

[green oriole]

lmao

also the backend gives much more information about failed tests than the frontend does

is that just a "we ran out of time" thing?

[thorny obsidian]

We could only do a css hack because of issues with state. If I tried to show more information it would clear all that was entered, which is non-ideal. We didn't have time to add in proper state handling

[green oriole]

that's fair

wait is the text below added by CSS

[vocal prairie]

completely separate, but could we add typing_extensions to snekbox?

[thorny obsidian]

Yup

[green oriole]

lmao

[gritty wind]

Technically yes, but the bigger question is if we want to, and that’s a question for the events team

So ask in #code-jams-planning

[vocal prairie]

not for code jam, for @stable mountain

[vale ibex]

FWIW snekbox for !eval , and the snekbox for forms are separate

[gritty wind]

In that case, yeah we can add it

[green oriole]

makes sense

being able to select what packages to use inside each snekbox eval when

the worst part is it is really not hard to do, just add a bind mount for each package

[vale ibex]

it'll be adding to the list here https://github.com/python-discord/snekbox/blob/main/deployment.yaml#L53

[stable mountain]

deployment.yaml line 53

yarl~=1.7```

[vocal prairie]

could I open a PR 🥺

[vale ibex]

go for it

I'm fine with adding it

[vale ibex]

so is the snake

[green oriole]

is it a snake if it has legs

[vocal prairie]

do I need to like, test it, or can I just trust.

[gritty wind]

I mean that just adds a pop install

Nothing to test there

[vale ibex]

just make sure the version number you add exists on pypi and you're good

[green oriole]

you'd have to start a k8s cluster to test that

[vocal prairie]

yeah figured

[green oriole]

have those been recently upgraded?

[gritty wind]

No

Things in that list are just added/modified when there's a need

[vale ibex]

should be live now, no idea how to test since I haven't used that package before :P

[vocal prairie]

!e

import typing_extensions

[stable mountain]

@vocal prairie :warning: Your eval job has completed with return code 0.

[No output]

[vocal prairie]

yayy

thanks chris

[vale ibex]

bot-core#96 nice small PR, a simple fix for clients that don't want to use statsd

[dusky shore]

GitHub

Don't try to connect to statsd_url if it's None

[vale ibex]

also @gritty wind is bot-core#91 in a state to be merged?

[dusky shore]

GitHub

Miscellaneous Bug Fixes

[vale ibex]

I see discussions around type hints, but I don't want that to block the bug fixes you have in there

[gritty wind]

They aren't blocking for this PR

we can work on it in another PR

[vale ibex]

very good very good

I'll roll a release once #96 is merged

unblocked sir lance work for now at least :D

[vale ibex]

thanks @cold island for some reason it didn't come up in #dev-log

[cedar frigate]

can somebody look at my command and give some feedback

also should i open an issue (it's for the lyric command @tranquil topaz and i were talking about) or is it ok to just link to our discussion about it

here it is https://github.com/greyblue9/sir-lancebot/blob/544834eeca877027ae107ee36c69a64ad0625a54/bot/exts/fun/lyrics.py

thanks ❤️

[green oriole]

Genius' ToS forbid automation

You agree to not use the Service to engage in any prohibited, illegal, or harmful activity, including without limitation:
4. harvesting or collecting, through use of automated scripts or otherwise, the contents of the Service or email addresses, contact information or other private information of other Users from the Service for any purpose, including without limitation for the purposes of sending unsolicited emails or other unsolicited communications to Users or reproducing the content of the Service;
https://genius.com/static/terms

You usually want to make an issue before working on it, for reasons such as this

If the stance of the core devs hasn't changed, this implementation cannot be accepted

[atomic ivy]

they do have an API though
https://docs.genius.com/

[subtle kraken]

they dont provide lyrics though afaik

[atomic ivy]

this implies they do,
https://docs.genius.com/#/songs-h2

A song is a document hosted on Genius. It's usually music lyrics.

[subtle kraken]

except it does not

all I see from their response is link to lyrics, not the lyrics themselves

['album', 'annotation_count', 'api_path', 'apple_music_id', 'apple_music_player_url', 'artist_names', 'current_user_metadata', 'custom_performances', 'description', 'description_annotation', 'embed_content', 'featured_artists', 'featured_video', 'full_title', 'header_image_thumbnail_url', 'header_image_url', 'id', 'lyrics_marked_complete_by', 'lyrics_marked_staff_approved_by', 'lyrics_owner_id', 'lyrics_placeholder_reason', 'lyrics_state', 'media', 'path', 'primary_artist', 'producer_artists', 'pyongs_count', 'recording_location', 'relationships_index_url', 'release_date', 'release_date_for_display', 'song_art_image_thumbnail_url', 'song_art_image_url', 'song_relationships', 'stats', 'title', 'title_with_featured', 'url', 'verified_annotations_by', 'verified_contributors', 'verified_lyrics_by', 'writer_artists']``` is all they provide (in terms of keys that you can access)

[gritty wind]

This was brought up a couple weeks ago here

[gritty wind]

And my personal opinion was this, due to the limitation Akarys mentioned above

And as they've mentioned, this is why we do issues usually, to avoid miscommunication, and clarify details such as this

In it's current state, we can not accept it

[green oriole]

I got a 400 missing_discord_data when submitting a form, despite having a submit button available

cleared the cookies and it went fine (the login button actually appeared)

maybe expired?

[cold island]

yeah it happened to us a few times, trying to figure out what might cause it

[green oriole]

by the way, I can submit a patch for what mina forwarded in the dev channel

that's indeed very weird

considering

do tokens have an expiration

yes

actually no

it is discord expiration, not a standard exp

[gritty wind]

They do expire yeah, at the same time the discord data expires

(Doesn’t make sense to keep it once its expired)

[gritty wind]

But yes, it’s quite strange

I can’t get down a solid repro tho, and relogging seems to fix it

[green oriole]

There is a token refresh endpoint, but is it used correctly?

If the token is expired I guess you'd want to redirect to the refresh endpoint, but that isn't ideal, ideally the authentication should do that in the background if necessary

[gritty wind]

It does do it in the background

But who keeps forms open for 7 days

[cold island]

Oops

I never close tabs lol

[gritty wind]

It’s not the reason fwiw

I managed to do it twice in like 3 days

[green oriole]

Is expirity correctly set then

[gritty wind]

Set correctly for what?

[green oriole]

Is actually set to a week in the future?

[gritty wind]

The token definitely is, but you might be onto something

The cookie the frontend can actually see is out of sync

The automatic refresh probably doesn’t set it does it

That would explain things

[tawdry vapor]

You're in the wrong channel. See #❓｜how-to-get-help

[green oriole]

Actually no, that'd be the other way around

The frontend thinks you are logged in, but not the backend

[gritty wind]

Yeah you’re right

It’s a non-consequential bug

But I did do more investigation in the mean time

One thing the backend does is delete cookies when it runs into issues

It doesn’t delete the frontend cookie tho

In fact, backend never sets the frontend cookies despite being able to

[vale ibex]

I've had a very naïve look at the auth code

how come this sets the expiry to the one already on the bearer_token, rather than the recently updated token_expiry? https://github.com/python-discord/forms-backend/blob/main/backend/routes/auth/authorize.py#L71

[stable mountain]

backend/routes/auth/authorize.py line 71

await set_response_token(response, request, token, bearer_token["expires_in"])```

[vale ibex]

that should be the same though IG

[gritty wind]

Bearer token is the new token

This just avoids having to switch back and forth from datetime to ints

Also to be pedantic, token_expiry is a point in time, while this is a duration (7 days)

[vale ibex]

ahh right, that makes sense

another question, isn't expiry here a unix time stamp? https://github.com/python-discord/forms-frontend/blob/main/src/api/auth.ts#L206-L207

[stable mountain]

src/api/auth.ts lines 206 to 207

const expiry = Date.parse(response.data.expiry);
setTimeout(refreshBackendJWT, (expiry * 0.9));```

[vale ibex]

So basing the expiry on that, rather than a diff from the current time seems off

[gritty wind]

No it is a time diff

Same 7 days you linked above

[vale ibex]

ahhh yea, mixed up expires and expires_in

[gritty wind]

Tbh

It’s a mess

Could use a rewrite

In fact, if we do away with this JWT nonsense and use the DB

It becomes trivial

[green oriole]

and just use an opaque token?

Yeah that could make sense

[gritty wind]

I meant more the stateless stuff

It’s only so complicated because both the backend and frontend are needed to coordinate refreshes

We could still keep JWTs themselves

[vale ibex]

wait, i don't follow this ```py
interaction_start = datetime.datetime.now()
max_age = datetime.timedelta(seconds=int(bearer_token["expires_in"]))
token_expiry = interaction_start + max_age
data = {
...
"expiry": token_expiry.isoformat()
}

...

const expiry = Date.parse(response.data.expiry);
setTimeout(refreshBackendJWT, (expiry * 0.9));

in the backend the expiry is the iso date of when the token expires

So wouldn't passing that through Date.parse() return a unix timestamp of that date?

[gritty wind]

Oh I see what you’re saying

One sec

[vale ibex]

afaik the fronend should be (and excuse my js)

[gritty wind]

Parse returns a Unix time stamp

[vale ibex]

const expiry = Date.parse(response.data.expiry);
setTimeout(refreshBackendJWT, ((expiry-Date.parse(Date.now())) * 0.9));

[gritty wind]

We pass the td (7 days) to the frontend I think, so you can cut out the math entirely

[vale ibex]

it's in the max_age of the cookie

unl;ess the frontend has direct access ot the bearer_token

[gritty wind]

Honestly we just need to dump all this from the frontend

[gritty wind]

It’s in the token, so no

[vale ibex]

The only thing returned from the auth is ```
response = responses.JSONResponse({
"username": user.display_name,
"expiry": token_expiry.isoformat()
})

so we could include bearer_token["expires_in"] too

cut out the math as you say

then it would just be ```js
setTimeout(refreshBackendJWT, (response.data.expires_in/1000 * 0.9));

I can PR that now if you think it's correct

ah looks like the frontend already does ```
const expiry = Date.parse(response.data.expiry);
return {username: response.data.username, maxAge: (expiry - Date.now()) / 1000};

that's in requestBackendJWT, which is used when getting the token initially

[gritty wind]

The refresh and first login are very strange

[vale ibex]

the refresh logic doesn'nt use this

Yea

[gritty wind]

I don’t know what idiot designed this

[vale ibex]

I could update the refresh logic to do the same as this, just *.9, rather than updating the backend too

[gritty wind]

Don’t git blame

[vale ibex]

lol

vsc has git blame built-in

I always know who to blame

and sometimes it's not me

[gritty wind]

Go for it if you wanna PR

Either is fine because I’ll hopefully rip it out lol

[vale ibex]

@@ -204,7 +204,7 @@ export async function refreshBackendJWT(): Promise<boolean> {
         cookies.set(CookieNames.Username, response.data.username, {sameSite: "strict", secure: PRODUCTION, path: "/", expires: new Date(3000, 1)});

         const expiry = Date.parse(response.data.expiry);
-        setTimeout(refreshBackendJWT, (expiry * 0.9));
+        setTimeout(refreshBackendJWT, ((expiry - Date.now()) / 1000 * 0.9));
     }).catch(() => {
         pass = false;
         cookies.remove(CookieNames.Scopes);

just gonna be this anyway

to share the logic with the initial auth

[green oriole]

What is that 0.9

This looks cursed

Ah I see

Isn't there a very small window of 0.7 days to refresh the token then?

[green oriole]

Hey @patent pivot data man, do you have CF analytics set up for pydis?

[vale ibex]

What sort of analytics are you thinking of?

We don't have Web traffic analysis, since we don't have pro

We have the other cf analytics though, like requests, cached vs uncached etc

[green oriole]

Yeaaaaah

What does the traffic between subdomains look like?

This is a totally random curiosity

[gritty wind]

I don't know how discord handles expired refresh tokens (it would've required me sitting and waiting for 7 days to test), so I added in a little buffer for us. Doing it a little earlier means we don't get too close to the expiration time, and it gives the user a grace period, because as indicated earlier, it's unlikely we'll always have the frontend running to refresh things

[green oriole]

Question

Do we actually need to refresh anything?

Surely we just need to know the user ID (and that the user owns the account)

Everything else can be pulled from discord

[gritty wind]

Well you can't pull user info any other way

It's intended to power features such as the username and PFP, which haven't been implemented yet

And it means anything that tries to access user data will not have to account for the condition when it's logged in but no longer accessible

Such as scope elevation

The system is only as complicated as it is because everything needs to be done by the frontend

#StatelessWasAMistake

[vale ibex]

We don't seem to have any of that, might be under the pro analytics I mentioned

Can only see total request numbers data saved, dns queries per response code etc

[green oriole]

It should be available under web analytics

But it is something you have to turn on

[vale ibex]

This one?

[green oriole]

Hmmmmmmmmm

[vale ibex]

This is the one I was talking about not having access too

[green oriole]

Why do we have access to that

[vale ibex]

Not sure

[green oriole]

You are looking here, right ?

Oh wait, if you go to a site in particular it asks you to upgrade

But if you go to the account home and select web analytics here, you don't have to

[vale ibex]

account home just shows this

(on pc now)

[green oriole]

If you select the first one, do you actually have no data?

[subtle cargo]

Why is the blog bypassing CF 🤔

[vale ibex]

yup, no data

We're probably just not serving the snippet

[green oriole]

The snippet is supposed to be autoinjected

Yet I've looked, it isn't

That's so weird

[green oriole]

You can use the bot to fetch the member here

You'll have to do that anyway to check for admin

[gritty wind]

No, not for the features forms needs

Such as email

[green oriole]

Right, email is an issue

[gritty wind]

You can obtain member info, but a lot of user info requires oauth

[green oriole]

Emails and connections are the two things you cannot get

Hmm

[gritty wind]

What’s the problem with oauth anyway

Why are we trying to get rid of it

[green oriole]

Considering the issues refreshing caused, it could have been interesting to just use OAuth to verify ownership then forget about it

But yeah, emails are a thing

[surreal veldt]

Can the amount of memory that the code took be added to the timeit command, maybe?

[gritty wind]

I don't think it's trivial to implement

[brazen charm]

That's quite a bit more complex, and I don't think anything in the stdlib can do it either

[gritty wind]

The timeit command uses, well, timeit

[brazen charm]

talking about timeit, bot#2201 's a simple no code review

[dusky shore]

GitHub

Improve help of eval and timeit

[gritty wind]

Oh wow that's cool

[surreal veldt]

Can't you use sys.getsizeof?

[brazen charm]

no

There's really not much you can do internally without significantly influencing the timing, so it would probably also need to be separate command

[gritty wind]

I imagine it'd also require injecting some boilerplate around user code unless we write a CLI utility for it

I still have nightmares from doing that with forms

!help eval

[stable mountain]

Command Help

!eval <code, ...>
Can also use: e

*Run Python code and get the results.

This command supports multiple lines of code, including code wrapped inside a formatted code block. Code can be re-evaluated by editing the original message within 10 seconds and clicking the reaction that subsequently appears.

If multiple codeblocks are in a message, all of them will be joined and evaluated, ignoring the text outside of them.

We've done our best to make this sandboxed, but do let us know if you manage to find an issue with it!*

[gritty wind]

very cool

[green oriole]

Hmmmm

Hmmmmmmmm

The system call wait4() can be used to retrieve that information

Is forking allowed in snekbox yet?

Yes

So it is possible to make a binary that will just call python and return the memory usage

Or go the lazy way and use time -v and parse that information

So yeah, it is doable

The question is whenever it is relevant

Don't forget that Python is an interpreted language, its memory usage would probably not fit your expectations

[broken topaz]

https://github.com/python-discord/bot/tree/main/bot

1. why is the bot structured the way it is? (you will run the folder instead of running a file?)
2. why not use pipenv for env management?
   3.1) in here https://github.com/python-discord/bot/blob/main/bot/constants.py why do we read a yaml file instead of a .env
   3.2) why are we reading instead of just defining it?

[thorny obsidian]

Running it as a module is pretty standard for large bots. It makes imports easier to contend with.

Pipenv takes forever We swapped to poetry because it's better overall for our primary developers

Yaml allows for a more complex config file vs an env. So we use yaml because our server setup for Python is pretty complex. Not sure what you're asking about reading vs defining

[broken topaz]

Fix question #3.2
Why do you put it in a yml file instead of just a python class or dictionary?

4. where can I find resources for standards on these kind of stuff? I really need it because I am very conscious about my project structure that I sometimes end up doing nothing.

Thanks for helping!

[gritty wind]

The reason it’s in yaml instead of a python file is that yaml supposedly is easier to parse and use especially when configs get larger

I’m not convinced, but not enough to change it

We do use python files for most other projects

(Though not dictionaries, we’d use things like data classes and what not. Look into pydantic, which does ship with a bunch of utilities that make this easier)

As for resources, I don’t really have anything. You can gather what is common by looking through large open source projects, but that’s far from the best options out there. Ultimately whatever works for you is most likely fine

[surreal veldt]

May I ask a question about github in general?
Let's say I fork a repository, and make changes to it and make a PR which is accepted.
Then, the owner of the repo makes changes to the files. How can I update the folder (VSCODE) to show the changes? I tried pressing the Fetch upstream from github but it didn't work

[gritty wind]

It didn’t work as in your fork on GitHub is still not updated, or only your local files?

I assume it’s the latter, in which case you need to make sure that you’re on the same branch on both (your fork on GitHub, and locally. Usually best to be on main). After that run git fetch <remote> and git pull <remote>, where remote is the remote you set when cloning. This is probably something like origin.

[surreal veldt]

oh yeah it's my local files. On github my fork is updated

should i do the above?

[gritty wind]

Yup

[surreal veldt]

where remote is the remote you set when cloning
What does this mean?

Just used origin. And it worked. Thanks!

[gritty wind]

Git remotes are basically a human-readable form of URLs. You can tell it that upstream binds to https://GitHub.com/repo, and origin binds to https://GitHub.com/fork, for instance. If you run fetch origin, it will pull from a different url than fetch upstream

The origin/upstream names are common in the forking model

[tawdry vapor]

Need a review https://github.com/python-discord/snekbox/pull/147

[wild prism]

I'm trying to redis on sir-robin working but it's like get or set operations never return

cachey = RedisCache(namespace="scoff")

class Ping(commands.Cog):
    """Send an embed about the bot's ping."""

    def __init__(self, bot: SirRobin):
        self.bot = bot

    @commands.command(name="hi")
    async def hi(self, ctx):
        log.debug("Attempting to set 'foo' to 'bar'")
        await cachey.set("foo", "bar")
        log.debug(f"Set op successful")
```log:

sir-robin | 2022-07-05 21:10:05,541 | bot.exts.ping | DEBUG | Attempting to set 'foo' to 'bar'
sir-robin | 2022-07-05 21:10:05,541 | async_rediscache.types.base | DEBUG | Creating NamespaceLock for namespace='sir-robin.scoff'.
sir-robin | 2022-07-05 21:10:05,542 | async_rediscache.types.base | DEBUG | Trying to acquire <NamespaceLock namespace='sir-robin.scoff' [unlocked]> for RedisCache.set
sir-robin | 2022-07-05 21:10:05,542 | async_rediscache.types.base | DEBUG | Acquired <NamespaceLock namespace='sir-robin.scoff' [locked]> for RedisCache.set
sir-robin | 2022-07-05 21:10:05,542 | async_rediscache.types.cache | DEBUG | Setting s|foo to s|bar.

127.0.0.1:6379> hgetall sir-robin.scoff

1. "s|foo"
2. "s|bar"

get calls do the same thing

anyone have an idea why?

running it docker-compose and real redis btw

[vale ibex]

What version of asyncrediscache and aioredis is installed?

[wild prism]

lol i've had that adventure, i think ultimately i ended duplicating sir-lancebot

let me double check

0.1.4 and 1.3.1

[vale ibex]

You can bump asyncrediscache up to 0.2.0, but that shouldn't be a problem

are they pinned to that version or ~= set?

[wild prism]

~=

[vale ibex]

try pinning aioredis to 1.3.1

[wild prism]

does that even matter if that's already in the lock file?

[vale ibex]

I remember having a similar issue with lance in my d.py migration branch

pining to 1.3.1 solved it

I think it was a sub dep not being pinned correctly

[wild prism]

hm nope, same behavior

[vale ibex]

did you also bump asyncrediscache version?

[wild prism]

no

[vale ibex]

alright lets try that

[wild prism]

are you saying that should be pinned as well?

[vale ibex]

it should be bumped to at least 0.2.0

We're slowly moving to exact version pinning with dependabot to tell us whn we need to update

I'm looking at the diff and can't think why this would fix your issue, but worth a try

[wild prism]

nope, nothing changed

[vale ibex]

the only thing after the hset in set is releasing the connection

so if that's going through, it suggests a problem there some how

actually, there is this call https://github.com/SebastiaanZ/async-rediscache/blob/6a68850e48cc36f0a60a3ddd12bbf3c908e10380/async_rediscache/types/base.py#L271 but I'm not sure why this would cause the hanging issue you're seeing

[stable mountain]

async_rediscache/types/base.py line 271

def _maybe_value_from_typestring(```

[vale ibex]

could you push your changes so I can attach a debugger?

[wild prism]

https://github.com/python-discord/sir-robin/tree/redis-debug

[vale ibex]

thanks

[wild prism]

when is the namespace lock released?

bc that's not happening either

[vale ibex]

that was deprecated afaik

[wild prism]

sir-robin              | 2022-07-05 21:54:17,551 | bot.exts.ping | DEBUG | Attempting to set 'foo' to 'bar'
sir-robin              | 2022-07-05 21:54:17,551 | async_rediscache.types.base | DEBUG | Creating NamespaceLock for namespace='sir-robin.scoff'.
sir-robin              | 2022-07-05 21:54:17,551 | async_rediscache.types.base | DEBUG | Trying to acquire <NamespaceLock namespace='sir-robin.scoff' [unlocked]> for RedisCache.set
sir-robin              | 2022-07-05 21:54:17,551 | async_rediscache.types.base | DEBUG | Acquired <NamespaceLock namespace='sir-robin.scoff' [locked]> for RedisCache.set
sir-robin              | 2022-07-05 21:54:17,551 | async_rediscache.types.cache | DEBUG | Setting s|foo to s|bar.
sir-robin              | 2022-07-05 21:54:37,456 | bot.exts.ping | DEBUG | Attempting to set 'foo' to 'bar'
sir-robin              | 2022-07-05 21:54:37,457 | async_rediscache.types.base | DEBUG | Trying to acquire <NamespaceLock namespace='sir-robin.scoff' [locked]> for RedisCache.set

[vale ibex]

oh

[wild prism]

was that an "oh" like "oops" or like "eureka"?

[tawdry vapor]

"oh" like "oh"