#dev-contrib
sleek steppe
Tells the linter what type the object will be
placid ermine
or you could just do from typing import List and it'd be pretty much same lol
green oriole
Riiight, I see
sleek steppe
or you could just do `from typing import List` and it'd be pretty much same lol
Nah this looks cleaner
placid ermine
yes
fallen patrol
!d typing
gritty wind
why is it a dict of the same times what
It tells your IDE what the return type of the function is
stable mountainBOT
sleek steppe
I mean lowercase looks better
green oriole
Typing is useless during runtime, except in specific cases
gritty wind
Like d.py cases, where it chooses to analyze the signature for some god forsaken reason
sleek steppe
In 3.10 it's completely ignored until you call get_type_hints or something
fallen patrol
Typing is useless during runtime, except in specific cases
pydantic
green oriole
d,py makes a really bad use case of typing haha
fallen patrol
its actually becoming increasing more important as argument validation
brazen charm
It is a bit more reasonable with the introduction of Annotated
fallen patrol
sleek steppe
d,py makes a really bad use case of typing haha
It hurts when I see people doing x: str = None
green oriole
Yikes
fervent sage
It hurts when I see people doing `x: str = None`
👀
gritty wind
converters*
green oriole
You typehint one thing, but get another
Yeah, that’s the real issue
I was about to say that decorators are good 😄
gritty wind
lol, brain fart
sleek steppe
Type hinting to functions are the worst
placid ermine
even pygen reminds me of latex pain now 😔
gritty wind
@green oriole migrated the command to a new file, and removed unnecessary imports
Currently looking at
import pathlib
import resource
import signal
import sys
import types
import typing
from functools import partial
from io import BytesIO
import matplotlib.pyplot as plt```still need 400mb though
any ideas why
gritty wind
Hmm
let me test that
disclaimer: I'm not actually counting, I'm just seeing the minimum needed to not get an error
Ok, if I remove nothing except the call to the render
I can go down as low as 0.0001mb
probably lower
I haven't tested
Ok, new idea
I'll also use matplotlib.Figure
gritty wind
Sure, the 0.0001 figure was everything other than the render call
gritty wind
ask matplotlib lol
placid ermine
it barely crossed 200mb without changes for me
vale ibex
iirc matplotlib uses a gui backend for every plot
so I'm not surprised on the memory reqs
gritty wind
plt.savefig(rendered_image, bbox_inches="tight", dpi=600)
everything before this line
placid ermine
that line spikes about 42mb for me
thorny obsidian
I don't remember which rendering backend is the lowest in mem usage... You can specify different ones though
vale ibex
You could try matplotilb.use('agg')
gritty wind
jumping up to 100, it still fails
green oriole
Interesting
gritty wind
and I don't think we'll give one command even 100mb in prod
vale ibex
using agg, it'll remove the need to teardown and gui objects
gritty wind
You could try `matplotilb.use('agg')`
is this before or after creating it
vale ibex
before
gritty wind
ok, still nothing on 10mb
jumping to 100
still a no
why is savefig taking so much ram
and what can we do about it
ok better question, what does it actually do
we write to file after that
green oriole
Can you pipe it to stdout?
gritty wind
Can you pipe it to stdout?
pipe what?
gritty wind
yeah, but we never get past the savefig
so my question is what does that do
Documentation says it "saves the figure"
green oriole
Wow
gritty wind
so... what does the writing to file do
green oriole
what a documentation
green oriole
600 dpi is enormous dear god
gritty wind
32 is still too high for 100mb
green oriole
Looking at some backend source code, savefig triggers the render (?)
green oriole
i- what is the image size?
gritty wind
I wonder if it's because of windows
I just dropped to 32dpi
no other changes
I'll deploy this on my server
green oriole
There is no image size setting?
gritty wind
matplotlib.use("agg")
fig = plt.figure()
rendered_image = BytesIO()
fig.text(0, 1, text, horizontalalignment="left", verticalalignment="top")
plt.savefig(rendered_image, bbox_inches="tight", dpi=32)
rendered_image.seek(0)
with open(filepath, "wb") as f:
f.write(rendered_image.getbuffer())
entire code right now
placid ermine
There is no image size setting?
it fits to the text size
and the font size is set in rcParams.update
i think i set it to 16
plt.figure() isn't reference counted properly apparently, that's why I switched to figure.Figure
gritty wind
that still has the same problem
it takes too much mem to save
even for really short inputs
placid ermine
i don't think the input size affects it much at all
green oriole
Yeah it shouldn’t
gritty wind
it was definitely enough to cause the initial memory trips that caused this issue to be raised in the first place
gritty wind
matplotlib.use("tk")?
vale ibex
agg is the raw backend
not sure how much better tk would be
worth trying at least though 
gritty wind
'tk' is not a valid value for backend; supported values are ['GTK3Agg', 'GTK3Cairo', 'MacOSX', 'nbAgg', 'Qt4Agg', 'Qt4Cairo', 'Qt5Agg', 'Qt5Cairo', 'TkAgg', 'TkCairo', 'WebAgg', 'WX', 'WXAgg', 'WXCairo', 'agg', 'cairo', 'pdf', 'pgf', 'ps', 'svg', 'template']
take your pick lol
TkAgg?
placid ermine
TkAgg probably
gritty wind
that's not happy with 100mb either, but on 500:
Traceback (most recent call last):
File "/usr/local/lib/python3.8/site-packages/discord/ext/commands/core.py", line 85, in wrapped
ret = await coro(*args, **kwargs)
File "/bot/bot/exts/evergreen/latex.py", line 105, in latex
raise Exception(process.stdout.read().decode())
Exception: libtk8.6.so: cannot open shared object file: No such file or directory
it's not writing the file 🤷♂️
or not writing it correctly
green oriole
Yeah
Although
It is interesting
Because it is failing before the backend is initialized
gritty wind
alright, tested on a clean linux install
still same prob with 100mb
bumping it higher is fine
(note: I've jumped back to 600dpi)
placid ermine
with just the command and without the dpy boilerplate the mem usage rarely jumps over 80mb for me /shrug
with the gc.collects that is
green oriole
So the library is doing something before initializing the backend that requires that much memory
green oriole
Is it because it is a forked process
gritty wind
possibly
green oriole
Let's see
gritty wind
that doesn't explain the remaining 300mb but even then
placid ermine
half of that is from the memory profiler
green oriole
How do you create the new process again?
placid ermine
its 45MB for me then, discounting for the memory profiler
i have absolutely no idea why you're needing something like 400mb
green oriole
Hmmmm it seems to indeed be sharing the same resources as the main process
gritty wind
one sec, going to push the current changes too
green oriole
Can you try to run the generation script directly, not through the bot?
green oriole
¯\_(ツ)_/¯
placid ermine
I'm supposed to profile only render right
gritty wind
Well, try profiling it from the subprocess, instead of on it's own
that way we can see where the extra size is coming from
and to confirm: running the render on its own does in fact use more than 100mb
I called it like python utils/latex.py test ./something.png
and it errored out
vale ibex
out of interest, is the huge memory requirement just on the first plot?
IE if you try to plot a second directly after, does it use a normal amount of memory?
placid ermine
module 'signal' has no attribute 'SIGXCPU'
🤔
gritty wind
what OS are you on
gritty wind
These changes are for unix based system exclusively
gritty wind
what are you using to profile?
placid ermine
!pypi memory_profiler
stable mountainBOT
gritty wind
ok, going to test the second drawing idea
then I'll do profiling
ok, the second one does run, I'll try to do it from the bot itself now
hmm, no, I lie
it failed on the second one
also failed from the bot obv
thorny obsidian
So question: is this still dealing with mpl's latex rendering? Could we just sidestep this and see if a slim but robust latex install might be the way to go?
gritty wind
yes
placid ermine
interesting
what about the infinite loops and file read write and all that that would come with it
wonder how much plain latex allows, hmm
gritty wind
Profiler refused to profile, so I'll just try the function itself now
Line # Mem usage Increment Occurences Line Contents
============================================================
57 84.7 MiB 84.7 MiB 1 @profile
58 async def _latex(self, ctx: commands.Context, *, text: str) -> None:
59 """Renders the text in latex and sends the image."""
60 84.7 MiB 0.0 MiB 1 text = self._prepare_input(text)
61 84.7 MiB 0.0 MiB 1 query_hash = hashlib.md5(text.encode()).hexdigest()
62 84.7 MiB 0.0 MiB 1 image_path = CACHE_DIRECTORY.joinpath(f"{query_hash}.png")
63 84.8 MiB 0.1 MiB 2 async with ctx.typing():
64 84.8 MiB 0.0 MiB 1 if image_path.exists():
65 84.8 MiB 0.0 MiB 2 await ctx.send(file=discord.File(image_path))
66 84.8 MiB 0.0 MiB 1 return
everything after this line is unprofiled
that's ugly
there lol
green oriole
It seems to have 80Mb of globals doesn't it?
placid ermine
profiler is approx 50mb
green oriole
What am I supposed to look at then haha
gritty wind
yup, the profiler was kind of useless
I'm currently trying to profile the function itself
placid ermine
looks like even minimal latex installations still take at least 50MB+
green oriole
I think 50MB of library is honestly pretty okay
We aren't short in memory by any mean right now
placid ermine
hmm
fallen patrol
looks like even minimal latex installations still take at least 50MB+
Lance has ffmpeg when it barely uses it
If at all
That takes up 200mb
Including its 100 dependencies
placid ermine
lol
vale ibex
I don't think it's used at all, need to do more testing though. sir-lancebot#686 for reference
dusky shoreBOT
fallen patrol
it would be nice if tests were complete
Lol
In that case you could just delete the dependency and run all of the tests
vale ibex
There's no tests on lancebot
patent pivot
Yeah heavy testing raises the entry bar to contribution, especially for discord.py bots
fallen patrol
Yeah
fervent sage
testing in bots is miserable
vale ibex
sir-lancebot#686 is ready for review, couldn't find anything that actually uses discord voice or ffmpeg
dusky shoreBOT
placid ermine
noob question
how would latex be installed locally
not like you can just download a file and run the installer or something
right
?
gritty wind
between linux internals, and a random pypi library, not sure which to believe
oh, ignore the logging lol
fallen patrol
!pypi latex
stable mountainBOT
placid ermine
makes pdf
not jpeg
or png
we could still use matplotlib
and just make it use system's latex instead of its own engine
placid ermine
and, well, if that has memory issues
fallen patrol
Not sure if that thing would work
placid ermine
we could switch
placid ermine
Not sure if that thing would work
which thing
fallen patrol
http://stackoverflow.com/questions/1381741/ddg#36249956
.
fallen patrol
I forgot
@placid ermine Recap: whats the current issue and also main issue with the latex command?
placid ermine
memory usage goes brrrr
fallen patrol
Iirc what's happening is its not closing right?
gritty wind
Well, I'm not really sure there is anything left for me to try with the process stuff, so I'll just deploy it on my cluster, and we'll see if it hits limits
placid ermine
'ight
fallen patrol
Well, I'm not really sure there is anything left for me to try with the process ...
your cluster?
gritty wind
kubernetes cluster
placid ermine
any updates
gritty wind
I have access to the manifest used to deploy sir-lance, so I can get something pretty similiar
Well, it's a 200mb image
taking a minute to push to the hub lol
~100mb done
placid ermine
o
fallen patrol
kubernetes cluster
where do you run it?
patent pivot
iirc scaleios self-hosts k8s
gritty wind
managed DO cluster
patent pivot
who was it that self-hosts
gritty wind
I would self-host if I wasn't a moron
gritty wind
yeah lak haha
stable mountainBOT
gritty wind
Ok, we really need to fix the dockerignore
it loaded my .env
so I accidentally just published a bunch of tokens
vocal prairie
Oof
green oriole
Uh oh
vocal prairie
Time to regen ig
vale ibex
dockerignore doesn't work with docker compose afaik
tawdry vapor
And that is exactly why the approach of a whitelist is better than a blacklist.
# Exclude everything
*
# Make exceptions for what's needed
!bot
for example
gritty wind
That is the current config
# Exclude everything
*
# Make exceptions for what's needed
!bot
!Pipfile
!Pipfile.lock
!LICENSE
But I just sshed into the container, and there it was
my .env
tawdry vapor
Was it inside the bot directory?
tawdry vapor
Then as far as I can tell that is an impossible situation
fallen patrol
so I accidentally just published a bunch of tokens
if its in the docker folder how is that published?
fallen patrol
yeah
gritty wind
which I published before noticing
tawdry vapor
If you run a container locally do you also see the file in there?
gritty wind
but ideally, the .env file would be loaded as env variables, not an actual file
gritty wind
If you run a container locally do you also see the file in there?
yes
vale ibex
So do I
tawdry vapor
COPY respects dockerignore
gritty wind
Then theoretically, it shouldn't have copied my env
tawdry vapor
What actually happens is that the build context is modified by the dockerignore
gritty wind
Well, shouldn't it still ignore everything
vale ibex
I think it's this in the docker-compose ```
volumes:
- .:/bot
Right now, it is not ignoring anything
vale ibex
since that'll share the whole folder right?
gritty wind
I think that's the interior bot folder
tawdry vapor
It may need to be ** instead of *
tawdry vapor
I see some saying ** and others *
gritty wind
the star didn't seem to matter
tawdry vapor
I don't think ** is necessary
vale ibex
👍
tawdry vapor
I didn't expect a volume to matter for building purposes
tawdry vapor
If you look at docker build there isn't even a way to specify a volume during building
tawdry vapor
What was happening then? The volume being mounted was giving a misleading result when running the container?
gritty wind
I tested locally, and saw the .env
I didn't actually check the remote image
Didn't consider how the compose would affect it
vale ibex
Yea, so the volume was being mounted and all the files from the project root were put into the container
tawdry vapor
I thought you said you ssh'd. I interpreted that as ssh'ing into the prod or something
vale ibex
k8s ig?
gritty wind
I don't think it does, that was not the right terminology
I opened a terminal in it locally
vale ibex
ahh just spawning a shell process
gritty wind
Someone snuck in an nbsp into the dockerfile haha
fallen patrol
Someone snuck in an nbsp into the dockerfile haha
&
vocal wolf
@sleek steppe Greetings. You around?
sleek steppe
Yep
vocal wolf
Alright so about this PR: https://github.com/python-discord/sir-lancebot/pull/684
Although I realize I've already approved, we cannot accept the pull request since we can't exactly filter out all emojis that are inappropriate.
That's why we only allow the ones that we have on this server to be shown.
No relevant issues as this was discussed in .
This also cannot be accepted, as this PR should've been discussed within an issue before making a PR.
green oriole
May I ask who you talked to about this issue?
sleek steppe
So after I opened the PR I had already asked some people, and then you and ks discussed it here <#dev-contrib message>
I mean we could do !e print("bad word") really
sleek steppe
Other commands can do that too
sleek steppe
Someone with an nsfw pfp can use an avatar command
green oriole
Yeah but I doubt they'll stick around for any longer
Any community they are in can moderate them and report to discord
sleek steppe
Don't think there will ever be a scenario where someone with an nsfw emoji will use an emoji command, but I guess you don't want to take that risk ¯\_(ツ)_/¯
limber jolt
So, hi guys. Im from Rus, and i downloaded Ubunt (Linux). And i have many problems with python on terminal in this system. Who can help me to download and use python on Linux?
fallen patrol
Although I realize I've already approved, we cannot accept the pull request sinc...
That was discussed
This thing first mention: #dev-contrib message
gritty wind
How do people feel about adding aliases to the doc command. From a cursory look, it seems possible to add a list of aliases that the bot translates before searching, but it might require a schema change on site, and quite a bit of added logic to the bot.
The use case for something like this would be aliasing certain modules that are better known by other names. For example: np or pd (maybe dpy to a lesser extent). Other challenges I foresee, besides the complexity issues, are issues with namespace pollution. Adding np as an alias is fine, but you run into problems when you want to add a package called np. I don't think that's true for any of our current docs, but it's a consideration.
brazen charm
It would be fairly simple bot side if it just tried the alias replacements in case of a failed lookup like it's done here https://github.com/python-discord/bot/blob/main/bot/exts/info/doc/_cog.py#L222-L234. For the use I personally don't see much benefit to it as you're probably already typing out things so the difference isn't huge, but then again I also don't use and fetch info from for example numpy often
cold island
How do people feel about adding aliases to the doc command. From a cursory look,...
iirc we had some complaints about not being able to search just by using the name discord (something about the inventory path being different)
stable mountainBOT
cold island
nice
fallen patrol
How do people feel about adding aliases to the doc command. From a cursory look,...
i think the bigger issue with this is !d discord.ext.commands.Bot shows the exact object path in the name. !discord ext.commands.Bot just doesn't make as much since, as it is always part of the discord object.
Does this make sense?
patent pivot
hey folks! I've just deployed some changes to pythondiscord.com and a set of our other services to introduce a set of new security headers, stuff like disabling permissions we don't need, tightening our referrer policy, rejecting iframes and so on (and of course disabling the horror that is FLoC). If you notice anything weird about resource loading on site please let me know.
fallen patrol
(and of course disabling the horror that is FLoC)
every header. its stupid way to opt out
fallen patrol
i mean that's how you opt out, no? every request has the specific opt out header
fallen patrol
ah
i think i need to learn some front end some time
puts it on a list for months from now
patent pivot
I am also deploying an exiting change shortly which is uhhhh
actually I'll just deploy it and then talk about it
fallen patrol
for those who wanna know if they're in the "beta" https://amifloced.org/
fallen patrol
huh?
i just realized as of now it seems mostly im-- wait nvm i... yeah.
was gonna say it isn't possible for FLoC to be on iPhones since every browser has to use webkit but that's wrong
patent pivot
allllllrigh
t
there we go
should be live
CSP is enabled on all python discord web properties, for now though we're only reporting instead of enforcing
we're also passing it through a worker to strip off IP data
vocal prairie
hey folks! I've just deployed some changes to pythondiscord.com and a set of our...
What's FLoC?
patent pivot
What's FLoC?
vocal prairie
Is bot#1498 still possibly useful, or not?
dusky shoreBOT
thorny obsidian
lol that's a great title
patent pivot
basically instead of identifying people individually through cookies identifying them through groups of cohorts with similar interests
vocal prairie
https://www.theverge.com/2021/3/30/22358287/privacy-ads-google-chrome-floc-cooki...
👍
vocal prairie
basically instead of identifying people individually through cookies identifying...
Fancy. But seems like a weird, not very good idea
thorny obsidian
FLoC is FLoCing terrible
patent pivot
yeah, it's a very bad idea
cold island
Is bot#1498 still possibly useful, or not?
yeah that seems fine
@sleek steppe assigned you
sleek steppe
Oh thanks
I think I'll need to rework redirect_output or make a new utility function
Maybe an allowed_channels parameter
fallen patrol
yeah that seems fine
i mean it only works for #python-discussion now so
fallen patrol
CSP is enabled on all python discord web properties, for now though we're only r...
what is scp?
patent pivot
content security policy
cold island
i mean it only works for <#267624335836053506> now so
that's still useful
patent pivot
a header which basically says which origins/locations any scripts/stylesheets/images can be loaded from
patent pivot
right now it's set to anything https://, pythondiscord.com or inline, I'll make it more granular in coming weeks
fallen patrol
i thought it was
related to tracking since you said it right after floc and cookies
patent pivot
nah, CSP is a security move for clients
so if we have a failure in that policy I see something like uhhh
fallen patrol
ah
sleek steppe
I think I'm ready to make a PR now, I also made not_in_blacklist useless
short snow
https://github.com/python-discord/bot/pull/1446#pullrequestreview-637609616 any confirmation on this?
clever wraith
If I wanted to add a feature to the @stable mountain bot do I just go for it with a pull request? Or discuss it here first or..?
I'm totally new to open source stuff 
And I wanted to see if I (or someone) could make it so editing !e into a code post will execute it
thorny obsidian
So the first step would be to open an issue on the @stable mountain repo. There should be a link in the description or in the pins
From there discussion will happen and a core dev will approve the issue and assign you to it (if you want to work on the issue). From there you're welcome to PR!
We do have contributing guidelines and all that good stuff to walk you through the process
We're also super friendly here, so you're always welcome to ask questions and provide suggestions
clever wraith
All makes sense 
thanks
thorny obsidian
I'm a bit tired, so hellooo to a bit of a philosophical rant.
One of our core ethos, especially for development, is that we do encourage people to learn new things and contribute to open source. @dusky shore is our fun and community bot that's encouraged for first time contributors to get started with.
That being said, our core server functionality (like the !e) does stay in Python. We have some stricter standards in Python because that bot is responsible for administrating the community, but our core devs and contributors are wonderful folks who can help you out
Also, remember to lint before you push 
(really you should lint before you commit)
vocal prairie
Also, remember to lint before you push <:hyperlemon:435707922563989524>
Didn't lemon make a song with that?
thorny obsidian
yes!
short snow
Didn't lemon make a song with that?
clever wraith
Do you that that feature would be likely to be approved? I help in the help channels a lot and it's just annoying when you type out a bunch of code but forget the !e at the start. I've seen other people try to edit the !e in to to retroactively execute.
vocal prairie
Noice
clever wraith
vocal prairie
Thanks Jason!
thorny obsidian
I do know our bot has the functionality of if we run !e, and then edit it within a time frame, it'll provide the option to re-run the eval with the edits
clever wraith
yeah, that's real handy
my idea is the same but for if you edit the !e in at the start of the message
short snow
Do you that that feature would be likely to be approved? I help in the help chan...
yeah, that would be a really handy feature, not just !e but it coould be done with all of the commands
thorny obsidian
I'm not sure how the bot will do with editting in the command since it'll be a different process. Someone a bit more knowledgeable will have to correct me here, but I believe when we do !e the process of listening for corrections only happens once the command is initially invoked. But maybe Python could still listen for the command on message edit
vocal prairie
I'm pretty sure you only check when the message is sent
clever wraith
yeah, that would be a really handy feature, not just `!e` but it coould be done ...
ooh, true, though that sounds trickier, heh.
short snow
I'm not sure how the bot will do with editting in the command since it'll be a d...
we can just use on_message_edit, and then call the process commands, like it happens on on_message
vocal prairie
So you'd have to listen for commands on message edit
short snow
that's my guess, not sure
thorny obsidian
Well, definitely write up an issue! This is interesting and I'm all for helping people help people...
If it's your first time writing an issue, you can check out the issue template on Sir Lancebot's repo. It'll give you some guidance on what to include
or just blindly steal the format from an existing issue
clever wraith
Alrighty 
I will look into writing an issue
vocal prairie
sir-lancebot#685
dusky shoreBOT
vocal prairie
It's in there
short snow
clever wraith
thanks
sleek steppe
For bot#1498, should I only include the eval command?
dusky shoreBOT
green oriole
For bot#1498, should I only include the eval command?
What do you mean? If we want to redirect the output of other commands too?
cold island
@short snow it would have been better to break up this commit into several. The title doesn't convey what it changes properly either
short snow
<@!736234578745884682> it would have been better to break up this commit into se...
firstly how should i do that? delete the commit and make new commits?
secondly is that really necessary, i have explained the changes in commit description
cold island
Yes, the commit title is important. You go to the body to see more details, but the title needs to convey what was changed
Adding an edit listener is not a refactor
brazen charm
having separate commits for separate changes with a fitting subject is important both when going over the history and with the blame
short snow
agree with that, any idea how i would be able to do it
brazen charm
If it's already pushed and reviewed then redoing it would probably lead to a mess in the PR but it is something to keep in mind for the future
short snow
maybe i can edit the commit title to make it clear? if that's possible
brazen charm
That will have to redo the commit as any other change, I'm not sure if review comments would be lost or kept with a detached commit on a force push
short snow
hmmm, what do you think @cold island ?
cold island
You can leave it for now, I'll see what others think
short snow
ok
thorny obsidian
@green oriole Could I trouble you for a re-review for lance's int e?
gritty wind
https://gist.githubusercontent.com/HassanAbouelela/0ca74e258caf57c07dd6f5e545e33...
That gist was specifically for PRs, not issues
When opening a new issue https://github.com/python-discord/sir-lancebot/issues/new/choose, and selecting feature, it will use the current recommended template
Anyways, I don't know how feasible it would be to listen to commands being added in after the fact. The eval command runs through the following checks and conversions in the regular on-message flow:
@command(name="eval", aliases=("e",))
@guild_only()
@not_in_blacklist(channels=NO_EVAL_CHANNELS, categories=NO_EVAL_CATEGORIES, override_roles=EVAL_ROLES)
async def eval_command(self, ctx: Context, *, code: str = None) -> None:
Now, these checks are not run for message edits, but they will have to be run to add your feature
That's a lot of overhead, because a ton of edits happen (trust me, we have logs lol)
I can't say for certain how high that overhead will be
But I can definitely tell you copying your message and sending it again is not too difficult
This also raises questions about how you make the distinction between an eval that had already been run, and one that has just been edited in
The event does pass in the old and new message
but that means you may end up running the checks twice
so that's doubling the overhead
</rant>
green oriole
<@407110650217627658> Could I trouble you for a re-review for lance's int e?
Sure thing!
!remind 30M review
stable mountainBOT
cold island
Yes
It should be implemented with that idea in mind (meaning, it should be a simple decorator), but the issue deals with eval specifically
sleek steppe
I added a categories + channels parameter to redirect_output
stable mountainBOT
@green oriole
It has arrived!
green oriole
Hey @thorny obsidian re: https://github.com/python-discord/sir-lancebot/pull/673#discussion_r611223893 sorry I didn't see your answer.
You can run exit() in the container to make it exit which will make k8s restart it, that's quite handy. But due to the fact that str.startswith was used, doing so will just trigger the context reset while it wasn't what you necessarily wanted to do. You can do !int e 1;exit() to bypass this restriction but that's a bit silly.
clever wraith
hello, would you guys agree to add a similar feature to lancebot as 8bitify?
my idea is to take an image, split it in x pieces, randomize those pieces and stitch it back up
green oriole
Hey <@!212644551926611969> re: https://github.com/python-discord/sir-lancebot/pu...
Also merged the PR
green oriole
hello, would you guys agree to add a similar feature to lancebot as 8bitify?
my ...
That sounds interesting, is it something you would be interested in contributing?
clever wraith
yes ill do the whole thing
i have it done already
but i would just have to make it work as a discord command
thorny obsidian
Hey <@!212644551926611969> re: https://github.com/python-discord/sir-lancebot/pu...
oh yeah! I removed the starts with check so now the exit() will restart
green oriole
but i would just have to make it work as a discord command
Cool, can you open an issue on https://git.pydis.com/sir-lancebot first please?
thorny obsidian
.int e ctx.send("hi")
green oriole
lol wait a minute for it to deploy
clever wraith
ill try, gotta do smth real quick
thorny obsidian
.int e ctx.send("hi")
dusky shoreBOT
hi
[Captured] <Message id=833009569848492092 channel=<TextChannel id=635950537262759947 name='dev-contrib' position=59 nsfw=False news=False category_id=411199786025484308> type=<MessageType.default: 0> author=<Member id=528937022996611082 name='Sir Lancebot' discriminator='9543' bot=True nick=None guild=<Guild id=267624335836053506 name='Python' shard_id=None chunked=False member_count=178917>> flags=<MessageFlags value=0>>
eeeeeey
vocal prairie
Would it be possible/useful to do permission/role validation before channel validation?
balmy sparrow
int e doesn't need you to await things? 
patent pivot
how FAST do you WANT deployments to be SMH
green oriole
nope
green oriole
how FAST do you WANT deployments to be SMH
faster
balmy sparrow
interesting
green oriole
just faster
patent pivot
the deployment is the fastest bit it's the docker part that is slow because docker is slow 
vocal prairie
Would it be possible/useful to do permission/role validation before channel vali...
Because I can type .int e here and it says I can do it in #sir-lancebot-playground, when I obviously can't, because I'm not an Admin.
patent pivot
green oriole
Would it be possible/useful to do permission/role validation before channel vali...
I don't mind that
clever wraith
@green oriole hi i added the issue with an example
clever wraith
ok
sleek steppe
`int e` doesn't need you to await things? <:pithink:652247559909277706>
if '_value_last_expression' in locals():
if inspect.isawaitable(_value_last_expression):
_value_last_expression = await _value_last_expression
_eval_context._value_last_expression = _value_last_expression
sleek steppe
It should be implemented with that idea in mind (meaning, it should be a simple ...
Alright I guess I'm just going to do it for !eval
upper lion
random question - is that beautiful open graph protocol github repo preview something that github has done lately? : o
sleek steppe
I assume so
sleek steppe
Bot#1531 needs some reviews.
dusky shoreBOT
green oriole
Don't worry, we will review it in due time :D
sleek steppe
Great
green oriole
Otherwise Xith doesn't get his weekly lemon delivery
green oriole
Hey @clever wraith, I forgot to ask, where did you get the code from?
clever wraith
myself
green oriole
Is it just something you kept locally, it isn't licensed anywhere?
clever wraith
i wrote it from yesterday to today, its nothing special
its not some api or anything like that
just some PIL magic
green oriole
Cool, all good then
green oriole
.source 8bitify
dusky shoreBOT
green oriole
Riiight here
clever wraith
thanks
fervent sage
i have a connection to the 8bitify command by now lol
clever wraith
aight, so it works as a discord command @green oriole
fervent sage
basically all of my lancebot stuff has been invoking commands from other commands, so 8bitify is my normal test subject
clever wraith
my command that is
im not really sure what should i do from now
i dont use github so i dont know how you guys add the code there
gritty wind
Have you ever used git or some other version control software before?
clever wraith
ive used git for heroku ig
but i dont really know about that
i can just give you guys the code maybe
gritty wind
If you really wanted to, it might be possible, but I'd also be down to walk you through some git basics
sleek steppe
Sir-lancebot has a contributing guide here too: https://pythondiscord.com/pages/contributing/sir-lancebot/
clever wraith
If you really wanted to, it might be possible, but I'd also be down to walk you ...
as you wish
also i have another question
sleek steppe
Sir-lancebot has a contributing guide here too: <https://pythondiscord.com/pages...
And a specific guide for working with git: https://pythondiscord.com/pages/contributing/working-with-git/
clever wraith
im suppose to do all the error handling for my command, right?
sleek steppe
Nope, sir-lancebot has its own global error handler for commands
gritty wind
Some of it, yes. For the most, you'd handle errors specific to your code, for example, errors that PIL might throw, but the error handler can format it for discord for example
clever wraith
i have some local stuff here that is relevant only to my own command so ill do it rn
sleek steppe
For example, type hinting to int for the amount of slices will raise a BadArgument if someone were to pass foo for it, and sir-lancebot will send an error message saying that converting to in failed
clever wraith
so for the wrong argument types that bot will handle it?
i just have to type hint it?
sleek steppe
Yep.
clever wraith
ok ty
sleek steppe
You'd need to type hint anyways because of flake8
clever wraith
and my command only takes numbers that you can get the square root of, so would this be allowed to add:
if not math.sqrt(splits).is_integer():
await ctx.send(f'Could not get the square root of `{splits}` :confused:')```I mean (0.0).is_integer() is True
I would raise commands.BadArgument so that the error handler could catch that and note in the docstring that the number must be a perfect square
clever wraith
I mean `(0.0).is_integer() is True`
what
but it doesnt make sense
anything that has .something is a float, no?
sleek steppe
Yes. (0.0).is_integer() will return True. Only floats have that method
!d float.is_integer
stable mountainBOT
clever wraith
@sleek steppe ok thanks
clever wraith
hello, so i finished my command fully and now would like someone to help me add it
sleek steppe
What do you mean add it?
clever wraith
What do you mean add it?
to the lance bot
fallen patrol
hello, so i finished my command fully and now would like someone to help me add ...
i assume you already made an issue on the repo?
sleek steppe
Yeah they did
clever wraith
yes
sleek steppe
It's like a 50% chance it actually shows
fallen patrol
kind of annoying
fallen patrol
to the lance bot
basically make a pr on the repo
how you do that is with git
short statement: git is complicated so we can walk through every step 
@clever wraith what do you have so far?
sleek steppe
Sir-lancebot has a contributing guide here too: <https://pythondiscord.com/pages...
@clever wraith follow this guide
clever wraith
ok
clever wraith
<@456226577798135808> what do you have so far?
the code for what the command does and the actual discord command for it
i can also show how it all works in my testing server
sleek steppe
You'll probably need to run the PIL and BytesIO stuff in an executor
fallen patrol
i presume its running on a lancebot clone?
clever wraith
i presume its running on a lancebot clone?
on my own bot
fallen patrol
the code for what the command does and the actual discord command for it
i mean the whole path to get it from start to finish on lance
clever wraith
i made a testing one
clever wraith
so am i suppose to fork lancebot?
fallen patrol
yeah
clever wraith
i think i did it
sleek steppe
Follow the guide and test the command to see if it works
clever wraith
i got it in my pycharm i think
the whole lancebot
btw what are environment variables?
not sure what to do here
oh i found them
clever wraith
yeah and it says error during connect: This error may indicate that the docker daemon is not running.
vale ibex
What OS?
clever wraith
windows
vale ibex
Do you have docker desktop running?
clever wraith
prob not
clever wraith
clever wraith
is the ones u get to access when ur pc starts up?
sleek steppe
Ah right had to do that the first time I tried using a vm
clever wraith
like while its starting up i mean
vale ibex
Yea, it is
vale ibex
You'll need to enable it for the same reason as a vm, docker engine uses virtualization for the container
And you need to enable that option in the bios
clever wraith
one of the circled settings is that vm thing and i dont even have it
clever wraith
You'll need to enable it for the same reason as a vm, docker engine uses virtual...
how do i enable it?
vale ibex
In that link, does it explain anything about bios options?
I'm on mobile so can't find you resources myself
clever wraith
has this
so i gotta restart my pc, get in those settings and enable that Virtualization Technology (VTx)?
vale ibex
Yea, exactly
clever wraith
ok ill go for it
vale ibex
If you can't find the option, google the name along with your motherboard name
sleek steppe
See you on the other side 👋
clever wraith
soo smth weird keep happening
whenever id go into those settings, my mouse kept getting repositioned in the top right corner and it was all glitchy
i tried restarting multiple times and also shutting down the pc and then booting it back up but same thing
clever wraith
i installed pipenv with pip but when i try to use it, it says its unrecognized
fallen patrol
Path, more like pita
fallen patrol
i installed pipenv with pip but when i try to use it, it says its unrecognized
In the repo directory, run python -m pipenv
clever wraith
In the repo directory, run `python -m pipenv`
it printed some inctructions here
im doing py -m pipenv run now but it says im missing a command and i bit confused here
vocal wolf
im doing `py -m pipenv run` now but it says im missing a command and i bit confu...
That's because you're not supplying a script. If you want to start the bot, you'll have to do pipenv run start.
In @dusky shore, you can find all the scripts in the Pipfile file. Located here: https://github.com/python-discord/sir-lancebot/blob/main/Pipfile#L35-L38
short snow
akarys told to move the webhook as cog level attribute that means i would have to add:
self.incidents_webhook = await self.bot.fetch_webhook(Webhooks.incidents)
to the __init__, so i would need to make the function async as it uses await
so the question is, is it ok to make init async?
gritty wind
I don't even think that's possible
The way to go about that here would be to call using an asynchronous caller
something like an asyncio task, or if this is @stable mountain, the scheduler
short snow
that fetches the webhook every x minutes?
gritty wind
is that what you want to do, or are you saying that's what it does
short snow
no i mean, should i make it fetch every x minutes or just fetch once at runtime
gritty wind
I don't have full context, but I'm assuming if it was updated, you'd need to update the config too
No real reason to fetch it multiple times
short snow
the config has been updated, joe had made a webhook in incidents channel, bot#1446
gritty wind
I think you missed my point
There is no need for it to be fetched repeatedly during runtime, because there are no runtime changes that it can adapt to
patent pivot
I'm meh on attachments being included
I think that has potential to be very noisy
for example, some reports may be to do with a question in a help channel and people might attach a code screenshot
I'll check internally though
brazen charm
Looks like it's missing a slash in the base url
gritty wind
yup
I'm trying to pull it but
get just caused the last error in #dev-log
What happens if I just try to override it
brazen charm
Should be able to change it with the web admin panel
gritty wind
it drops me in a user-password login page, but maybe I'm on the wrong site
!d assert
stable mountainBOT
gritty wind
git fetch upstream && git merge upstream/main
while you have your main branch checked out
fallen patrol
well, the cache is corrupt
nope the links are invalid
gritty wind
because it was not updated
fallen patrol
not https://docs.pytest.org/en/stable/index.html
gritty wind
I'm talking about the cache that holds the actual content of the symbol
the one that formats the embed
gritty wind
Missing settings
It tries to read them from the redis cache
if they aren't there, it writes defaults, and sends a warning
(It is no longer a thing, but long ago whenever someone would start up their bot in the staff testing server, we'd all get a nice ping)
brazen charm
The cache shouldn't be set on failed items so it's probably just parsing the html weirdly. But I wonder why the python symbol is the moved one
fallen patrol
git fetch upstream && git merge upstream/main
any way to do that on the github browser?
gritty wind
I think you may be able to PR or something?
Doesn't seem possible
Ok found a way
vocal prairie
I actually thought "your_username" was an account.
gritty wind
haha it probably is somewhere
patent pivot
it drops me in a user-password login page, but maybe I'm on the wrong site
if you login on the main page under the navbar and then navigate it should let you in
brazen charm
https://github.com/python-discord/site/pull/373 could use a look at if it's still up to date to also prevent the error in the site side
stable mountainBOT
short snow
https://github.com/python-discord/bot/runs/2372148113#step:10:41 how should i fix this? even if i set is as None in the __init__ it would fail
fallen patrol
very cool github
gritty wind
there's a problem with your approach
fallen patrol
gritty wind
you assume the task will be done before the webhook is ever used
which won't always be true
if d.py has any sort of async call on load, that would be the way to do it
otherwise, you're better off checking if the hook exists whenever you go to use it
short snow
if d.py has any sort of async call on load, that would be the way to do it
lemme see that
i went through the duck pond and python news, and they are done in the same way, dunno why it is failing on mine
gritty wind
You aren't calling the init in your tests
neither of those modules seem to have tests lol
short snow
https://github.com/python-discord/bot/pull/1446/files#diff-f74532881e346e2d6fb8a5423b30c4f50511188f90af67cf716a0c4f1d065c23R310 isn't it automatically called here
short snow
neither of those modules seem to have tests lol
lol, uhh oh i hate pycharm, everytime i reformat the file, this happens 🤦♂️
gritty wind
Oh, I didn't see the parent class
Note that this will not schedule `crawl_incidents` in the background, as everything
is being mocked. The `crawl_task` attribute will end up being None.
The create task isn't called by the mock
short snow
should i just add a line self.cog_instance.get_webhook() in setUp then?
gritty wind
Maaaaybe? Technically that is the behavior your init is doing, but if someone modifies the init without modifying that setUp, the tests will be exhibiting weird side behaviors
Ultimately this will fall down to the reviewer, but it makes more sense to me to possibly monkey patch that create_task
Basically:
Monkeypatch self.bot.loop.create_task to map it to a create_task from your async setup
short snow
Maaaaybe? Technically that is the behavior your init is doing, but if someone mo...
yeah, we could possibly add a comment to that tho
Basically:
Monkeypatch self.bot.loop.create_task to map it to a create_task from your async setup
I am not exactly sure what you mean by "create_task from your async setup"
gritty wind
The action that gets skipped, but we want called is:
self.bot.loop.create_task(self.get_webhook())
that's because self.bot.loop.create_task doesn't actually do anything
when defining your bot, you can point self.bot.loop.create_task to the create task of a real loop, or optionally asyncio.create_task
You will have to see how that affects the other create_task in that init though
short snow
rather than monkey patching self.bot.loop.create_task we could just make a async function to this, which gets called in init, iirc it is the asyncio equivalent of starting a thread.
gritty wind
asyncio.create_task exists
short snow
yeah, either could be done i guess?
gritty wind
If I understand you correctly, you want something to call create_task without actually having the bot.loop
in which case, you don't need to roll your own helper for that
short snow
right, i will try doing that
fallen patrol
Guys @patent pivot is being mean on metricity#3
dusky shoreBOT
short snow
lol
green oriole
Lmao joe
brazen charm
Can someone refresh the doc inventories?
brazen charm
refreshdoc
green oriole
!refreshdoc
brazen charm
(under doc)
green oriole
!doc refreshdoc
stable mountainBOT
green oriole
That’s a weird command lol
brazen charm
Thanks
green oriole
Np
vale ibex
That’s a weird command lol
Maybe worth a root alias
tawdry vapor
Yeah it's weird to avoid name conflicts with symbols
brazen charm
I don't think it's used nearly enough to have a root alias
green oriole
I wonder if it could be !doc _refresh to avoid conflicts
tawdry vapor
I agree but it can if someone wants I don't really mind
tawdry vapor
That works I would imagine but it's ugly if you ask me
fervent sage
why not just a separate command?
tawdry vapor
Don't remember
That was brought up though
Probably something to do with being unintuitive to separate especially considering it used to all be under one group
fervent sage
tbh i'm not sure being unintuitive is a concern seeing as it's used approximately never
brazen charm
Probably not worth losing much time on it for the same reason
tawdry vapor
Yeah :🙍
Iirc I laid out both options and said you decide which to pick
Well, maybe it was even your idea not mine
brazen charm
For the few uses it'll be easier to check the help on doc than going over the general help
stable mountainBOT
@patent pivot
It has arrived!
green oriole
It is tiiiiiime
placid ermine
@fallen patrol :^)
stable mountainBOT
Hey @sand oracle!
vocal prairie
It's time
patent pivot
merged
clever wraith
That's because you're not supplying a script. If you want to start the bot, you'...
it says pipenv is unrecognized
and i installed it with pip
i tried py -m bot right now (py is instead of python for me) and im getting this:
for res in _socket.getaddrinfo(host, port, family, type, proto, flags):
socket.gaierror: [Errno 11001] getaddrinfo failed```@clever wraith Do you have in your .env file FAKE_REDIS = true ?
clever wraith
BOT_TOKEN='ODMzMDE0NTQNlcKnKGdYQFrw'
BOT_GUILD=833014638723072020
BOT_ADMIN_ROLE_ID=833090174942052383
ROLE_HELPERS=833090255573876776
CHANNEL_ANNOUNCEMENTS=833090325605515274
CHANNEL_DEVLOG=833090377916874772
CHANNEL_COMMUNITY_BOT_COMMANDS=833090470052626482
USE_FAKEREDIS=true```this is it
i altered the token so dw btw
clever wraith
<@456226577798135808> Do you have in your .env file FAKE_REDIS = true ?
.
and this is where its loacated
thorny obsidian
What does py -m pipenv run start
do?
clever wraith
OOH
its creating smth
some vm env
now its saying im missing module arrow
even tho i installed it
thorny obsidian
Let's do py -m pipenv install
thorny obsidian
You did install it, but not in the venv that was just created
clever wraith
ah
clever wraith
You did install it, but not in the venv that was just created
got this now
thorny obsidian
Alright, now let's do the py -m pipenv run start
clever wraith
ok
oh it sent some messages in my server
i think its running
is it ok if a lot of channels are missing?
cuz i just made the required ones
that were in the guide
thorny obsidian
Oh yeah, that's fine
vale ibex
Nice! I'd suggest you also do py -m pipenv sync --dev and py -m pipenv run precommit
This will get you setup with the precommit hook
clever wraith
its already running rn so should i turn it off and run those commands?
vale ibex
so how do i now add my command to this bot?
You can do it now, or after you play with the bot a bit to make sure its working
clever wraith
ill just do it now
vale ibex
Is your command setup as a cog?
vale ibex
I mean you said you have code right?
clever wraith
i have the whole code but it was only made to work with my test bot
vale ibex
Is it written as a cog
clever wraith
its just written for a bot
vale ibex
Or a plain command
vale ibex
Alright if you take a look at bot/exts/evergreen there will be a bunch of files in there
clever wraith
yeah already saw
vale ibex
Those are a few commands we have there
clever wraith
should i use 8bitify for reference?
vale ibex
I suggest you copy one of the smaller files and use that as a base
clever wraith
cuz its kind of similar prob
vale ibex
Wait, what issue is this for?
vale ibex
Hmm, I've got a PR open at the minute that changes all of the profile pic modifying code
clever wraith
btw the way i coded it was that i had 1 module for the actual pfp mod and other module for the bot so should i merge them together in one module for this?
vale ibex
Hmm, let's get it into a PR, so we can see what your code looks like, and go from there
clever wraith
im not too familiar with all this github stuff, so to get this clear, i should code it into a cog for lancebot and then do the pr?
vale ibex
Did you fork the repo?
clever wraith
yes i have everything already here
vale ibex
Cool, I'd suggest making a branch on your repo, it'll make it easier later
1 sec, let me get a guide
clever wraith
ok
vale ibex
OK, from the root dir run git checkout -b name
Where name is what you want to call your branch
clever wraith
ok, also is the name for me to decide or does it have to be approved
vale ibex
You can call it what you like
clever wraith
ok
vale ibex
Just try to summarise what yours doing in 3-4 words
Doesn't need to be too descriptive
clever wraith
oh
i thought it was suppose to be the name of my command?
or can that be multiple words
cuz i just named it splitify
similar to 8bitify
clever wraith
ok ok
vale ibex
So you've got a branch now Yea?
clever wraith
ye
vale ibex
Alright, make a new file in the evergreen folder
Call splitify or whatever you want it to be called
vale ibex
Yea, put all the cog code in there
vale ibex
Sure
stable mountainBOT
@gritty wind
It has arrived!
sleek steppe
pipenv run precommit
git add .
git commit -m "Your commit message"
i just have 1 question
my command requires a perfect square num and i made a little error handler to check that and if the num isnt perfect square it raises commands.BadArgument().
but the thing is that i dont think that error that the user would get from the lancebot would clarify why their provided num was not accepted
this is what i mean
placid ermine
you can specify an error message like raise commands.BadArgument("not a square")
clever wraith
```
pipenv run precommit
git add .
git commit -m "Your commit message"
```
should i do this now?
it says that pipenv was not recognized
sleek steppe
python -m pip install pipenv
Try running the commands again then?
clever wraith
which?
clever wraith
so i should just skip the first one?
and when i try to add it says: /usr/bin/env: ‘python’: Permission denied
clever wraith
if anyone can help, ping me