tools-and-devops

[analog slate]

hah

[versed patrol]

Are jupytr notebooks still the best OS-agnostic option for basic development?

[analog slate]

hi

What is the best method or library to study payment methods, modify them, and change the payment direction of a website?

[thorny shell]

dunno what you mean. But if I were interested in payments, I'd read about the Stripe API -- I hear their docs are unusually good

[astral apex]

Stripe's docs are hailed as the gold standard

Also Stripe themselves are a godsend for helping keep your PCI env small

[cosmic nova]

Hey folks, anyone here have experience with dragonfly?
I wanted to know how good are they versus redis, mainly how much cost they take up, which one is cheaper as i get conflicting results online

[short peak]

we moved to valkey to reduce cost (that's what I was told at least)

[cosmic nova]

Are yall saving tho

Tbh i dont even mind if its same cost but better perf, but still

[short peak]

I don't know about costs, I was told that reason to justify redis migration. But something to highlight is that moving from redis to valkey is literally just replacing the names in code, pretty easy. You can even migrate your code and continue using redis instances if devops takes longer to replace it

[brazen forge]

Aren't they only removing some container images from the DockerHub? They will still be available in GitHub?

[gentle solstice]

You guys update your charts?

[fresh seal]

Has anyone made the switch from conda/mamba to pixi?

[deep estuary]

i think people actually depending on bitnami for prod are probably doing the wrong thing anyway

[subtle quarry]

Not a conda replacement but I use uv these days more and more. So fast.

[inner pollen]

I’ve extracted Gunicorn’s internal debug module and turned it into a standalone Python package. It traces deeply into the dependency tree and prints every line being executed. This library is mainly useful for students or anyone just curious about code flow , it’s not meant for serious memory profiling or performance analysis as people can argue pyinstrument and others are there, I will make this configureable to track only function , methods , a configuration based debugging tool this weekend , please do try it out and suggest improvements.

https://github.com/Agent-Hellboy/spewer Agent-Hellboy/spewer

[echo niche]

uv is the ultimate go-to.

[jade crown]

Hey all. I just released my first open-source (fully free) project that I have been working for a few months called CodeBoarding. It uses static analysis + LLMs to generate interactive diagrams of Python codebases.
Repo : https://github.com/CodeBoarding/CodeBoarding
Any feedback on further integrations you’d like to see would be super helpful, thank you!

[devout geyser]

https://gist.github.com/blacksteel3/44fc17881a7e0792bc83d3e8a268a498

[unique oracle]

i wanna deploy a fastapi app with docker and i am not sure where the sudo password should be stored.

rn, i am using the built in secrets manager of jenkins, but are there better ways to do this?

i feel like the only safer thing would be to use AWS KMS/secret manager + roles to restrict who can access what

[gentle solstice]

Sudo wouldn't have a password in docker

Most images don't even have sudo installed

[unique oracle]

I don't mean inside docker, I mean the sudo password required to invoke docker

AFAIK using sudo is best practice

[deep estuary]

long term container stuff is probably moving to being rootless anyway, see things like podman

so it's more of a temporary pain until that stuff becomes standard

if you want to be a little safer you can add your automation user to the docker group usually and you then shouldn't need to use sudo

https://docs.docker.com/engine/install/linux-postinstall/

If you don't want to preface the docker command with sudo, create a Unix group called docker and add users to it. When the Docker daemon starts, it creates a Unix socket accessible by members of the docker group. On some Linux distributions, the system automatically creates this group when installing Docker Engine using a package manager. In that case, there is no need for you to manually create the group.

I think this mostly covers your use case?

then if you want to do a bit of hardening you can look at https://docs.docker.com/engine/security/rootless/ but YMMV on how well your workloads will run under rootless at the moment

(See known limitations)

[deep estuary]

the note is important that technically since the docker Daemon runs as root (unless you do rootless) then adding a user to the docker group does give them root level access

but yeah to summarise I'd definitely recommend using the docker group over sudo, there's still risks of root level access but it's more scoped than just straight up giving your CI user root

[unique oracle]

I setup sudo to only allow the docker command as a prefix, and since sudo requires a password, then it's more secure IIRC

I am using podman on some deployments but some things still need root e.g. nginx cuz ports 80 n 443 either way so i want to see if I can just sudo

[unique oracle]

rootless docker to me just looked like podman with extra steps

[deep estuary]

hahahaha yeah it is

[deep estuary]

yes -- but remember if something has access to docker it implicitly has system root

you can just start a container as root, mount your root filesystem and operate on it as the root user

so scoping your CI user still means that it can get root

[deep estuary]

you can do this with capabilities now

wait I lied maybe you can't

but I remember my fix for this

I just set the minimum port limit to 0

with sysctl

it feels hacky but when you think about it the binding restriction is ultimately crap and outdated anyway

a malicious service binding to 1023 instead of 1025 offers it no advantage, and there are now dangerous ports above 1024 like basically every database server

sysctl -w net.ipv4.ip_unprivileged_port_start=0

[unique oracle]

correct but adding the user to the docker group means you have root at all times rather than only during the time in which we have the password available

I do use non root containers for things like Redis and my fastapi apps

[deep estuary]

yeah that's true, I just am not a fan of passwords in automation systems

I say this realising as an ansible lover whilst that ansible recommends you keep your sudo password in a text file in some configs 🤣🤣

[unique oracle]

https://unix.stackexchange.com/questions/161780/call-sudo-from-jenkins

I found this answer pretty helpful when I tried to deal with this before

visudo and a prefix that only allows certain commands

[unique oracle]

ansible is cool but it's just not suitable for my use cases

[deep estuary]

yeah this is probably where I'd land but I don't think I'd love storing sudo passwords even if scoped

but yeah I can't think of a better alternative hahahaha

other than rootless I guess

[silk mist]

I'm currently using uv to make a project for distributable app with python

I'm looking for something to get me started on setting some flags in the source or in tools I could use where I'd specify whether I'm building a CLI-only distribution or a GUI/ CLI version

and at the same time I want development-specific variables/ paths

I remember using .env for these purposes with a nodejs project about 7 years ago but it's been a really long time since I've done that sort of thing

btw for the building part I'd use either pyinstaller or nuitka and I think that might be as simple as using different commands/ different spec files but I'm wondering if the same can be done while developing

[silk mist]

ok for that I think I'll do the checking along with the check for if the package is frozen

[unique oracle]

I'm gonna use podman when possible and sudo when root is needed ig, thanks guys

or maybe just uv + systemd if I really don't feel like using containers

[deep estuary]

i've used https://docs.pydantic.dev/latest/concepts/pydantic_settings/ a lot for settings management, it is really nice

[deep estuary]

the way god wanted it 🙏

[unique oracle]

this is goated

I miss it when I use other languages

[deep estuary]

hell yeah brother

i've been doing a lot of fastapi with pydantic and it really is rock solid stuff

[silk mist]

this does look good

I'm making an app with Qt so I'll probably use the QSettings for actual app settings and pydantic settings for telling QSettings where to look 😂

[chilly hazel]

Is cursor more expensive now, or is my context management bad (due to increased size of the codebase overtime... If you know what I mean.) ?
Same for Cline, some API's average at around 50cent per call now. Which is making me think , that my context management is not well set, or set at all in that case.

[tropic chasm]

can ruff provide full lsp capabilities and replace pyright?

[placid nacelle]

That would be a usecase for ty and not ruff

[tropic chasm]

yeah you are right

[azure wren]

Heyy y'all. I'm a bit new to python. Made my first project today. It's not something big lol but i tried.
I saw a lot of people struggling with keeping their CSV files [most of them containing a bunch of URLs that are used for data scraping, crawling and API Endpoints] up to date. So i made this:

repo : https://github.com/l-RAIN-404-l/DeadURL

It automatically cleans files from broken links, produces a report and a backup. I know there may be issues so y'all can give me a feedback or even contribute. Thank youuu✌️

[simple crown]

What's the best setup for learning python

[zenith mango]

start with console and go start exploring web dev py flask

[short peak]

reading the official tutorial helped me.

[tired drum]

Hey folks! I put together a quick example showing how to launch a Positron IDE workspace inside a DevPod devcontainer. Works great for R or Python projects.

🔗 Repo: https://github.com/davidrsch/devcontainer_devpod_positron

Give it a try and let me know what you think!

[gentle solstice]

Is this an ad?

[shadow crow]

!warn @astral thorn We don't allow advertising here.

[rancid schooner]

:incoming_envelope: :ok_hand: applied warning to @astral thorn.

[manic plume]

Been working on CLI tools lately and loving the Python ecosystem for developer experience.

The community here is amazing for getting feedback and ideas!

Anyone else building dev tools?

[twilit forum]

Anyone using aider? I just set it up today with help from a colleague. Seems really nice! 😊

[hot wedge]

Thanks bro, I need this. 🫂

[left minnow]

hi

can any one help me with a small problem

its related to python to exe converter

[dire talon]

Hello,
I'm interested about it.
Where are you from?

[delicate plaza]

Hi guys! ive made this and can anyone give feedback to make it better? https://github.com/DanYell0038/Reminder-Tool

[astral apex]

Don't upload a zip
Zips are for hackers trying to hide viruses

GitHub is for code
Just put the Python file on GiHub directly

[pastel trellis]

Material palette Generator gives you material palette based on original material palette generation algorithm.

It's available in Github and PyPI

It's going to be a great tool for those who want to make their own UI theme based on any GUI toolkit such as tkinter, kivy, pygame, pyqt5 or anything else.

[brittle terrace]

I recently discovered Typer library and I really liked it ...
So I made this simple todo list as a learning project and I will add more features with time ...

Repo:
https://github.com/abdelkarimLog/cli_todo.git

[muted canopy]

How would I get started in the the role of devops?

[mystic void]

In terms of career, a degree is the path of least resistance and with the most opportunities and compensation. Beyond that practice makes perfect. That means demonstrated skills related to devops and how you care about it through projects, internships, etc.

[muted canopy]

Any tuts that you recommend or youtubers that you suggest?

[mystic void]

nope

[fervent mica]

A decentralized AI infrastructure protocol has been developed over the past year by a small group of contributors, aiming to optimize computational power for AI model training and inference. The project seeks to provide a competitive and scalable alternative to traditional cloud providers. Reach out to me via DM to learn more.

[willow palm]

Talk to all your coworkers

[waxen nova]

Ad probably

[warped bolt]

what lsp server do yall suggest for python in neovim?

[marsh acorn]

I use pylsp

[twilit forum]

I also use pylsp. It doesn't resolve some really quirky typing as I would have hoped. But it was still the best experience when I tried out lsp a year ago. (I think I did pipx python-lsp-server)

Other options would pyright and maybe ty?

[eager flame]

Hey everyone

I’ve just released HardView 3.1.0b1
This one adds a lot of new stuff, mainly the new LiveView module which comes with several classes and functions for real-time hardware monitoring.

Supports tracking CPU, RAM, disk usage, and even GPU (Windows only for now).
Temperature monitoring:
On Windows via a C++/CLI wrapper around LibreHardwareMonitorLib.
On Linux via lm-sensors
Also includes a function to fetch detailed CPUID info (Using an internal HardView library called cpuid.hpp ).

Install this beta version directly with:

---

pip install HardView==3.1.0b1

If you want to check out the docs, usage examples, and ready-to-use code:
👉https://gafoo173.github.io/HardView/LiveViewAPI/

GitHub repo (for the source, internal libs, tools, etc.):
👉 https://github.com/gafoo173/HardView

This release is still beta. It’s been tested a lot on Windows, and to a lesser extent on Linux. Testing on different devices and architectures would help a lot. If you run into any issues, feel free to DM me or open an issue on GitHub.

[opal tinsel]

I have issues with Python ETL scripts here and there causing the whole ETL server going down due to the oom-killer from too much memory use.
Does anyone know a good go-to way to ensure that any rogue script doesn't take down the whole server?
I can't just say to the team "please mind memory usage", we just have too many Python ETL devs and projects. Some entire server-wide solution would be nice

[opal tinsel]

like in PHP you can configure a global memory limit for PHP scripts.
though if that's tought to set up globally in Python, then some systemd configuration would be nice to prevent the oom-killer from wreaking havoc on our server

[hidden turret]

Linux has cgroups for that. You could create a group with a memory limit. example

But those are pretty fiddly to set up. I'd probably consider wrapping my script in a Docker container and using its resource constraint mechanism (which uses cgroups under the hood)

[deep estuary]

have a look at nsjail as well @opal tinsel

provides a lot of nicer abstractions around cgroups

https://github.com/google/nsjail

it's what we use for the !eval command on our bot to allow sandboxed execution of python code

!eval ```py
import sys, pathlib

print(sys.version)
print(list(pathlib.Path("/").glob("*")))

[rancid schooner]

:white_check_mark: Your 3.13 eval job has completed with return code 0.

001 | 3.13.5 (main, Jul 30 2025, 13:16:15) [GCC 12.2.0]
002 | [PosixPath('/home'), PosixPath('/dev'), PosixPath('/snekbin'), PosixPath('/usr'), PosixPath('/snekbox'), PosixPath('/lib64'), PosixPath('/lib'), PosixPath('/etc')]

[hidden turret]

TIL about nsjail.Thanks for the tip!

[deep estuary]

it's a really really neat tool 🙂 this is our project that uses it: https://github.com/python-discord/snekbox

huge amount of customisation available

and because it's all just native linux features like process namespaces and cgroups it's all mostly transparent to programs running within jails

[heavy knot]

Hey i have next to no experience with python but i was trying to right a program that would help me do bulk downloads from internet archive and i can not seem to figure it out would anyone be able to help me or be willing to write me a program like i see on YouTube if i paid them? i would very much appreciate the help.

[eager flame]

Use the requests library if you want to download from any website on the internet. It's the easiest library to work with for this. If you want ready code and don’t have experience, you can use an AI tool to generate the code or practice using its functions yourself.

[peak wraith]

Ask for help in #python-discussion , we don't allow recruiting or payments tho

[heavy knot]

Okay thankyou sorry for the trouble im just trying to get the bulk downloads thing figured out and i did not know where to go thankyou for your time i will go to where you said,

[brittle terrace]

I never understood American units like feet ... They are confusing
So I built this :
https://github.com/abdelkarimLog/uconvert.git

[muted girder]

Good boy karim... good boy

[brittle terrace]

A star will make me happy 🙂

[eager flame]

D:>python main.py volume 1 l gal
┌──────────────────────────────────────────────── Traceback (most recent call last) ────────────────────────────────────────────────┐
│ D:\main.py:105 in volume │
│ │
│ 102 │ │ result_unit: Annotated[str, typer.Argument()] ┌────── locals ───────┐ │
│ 103 │ │ ): │ base_unit = 'l' │ │
│ 104 │ │ │ if base_unit in volume_units and result_unit in volume_units: │ result_unit = 'gal' │ │
│ > 105 │ │ │ r = (valuelength_units[base_unit])(1/length_units[result_unit]) │ value = 1.0 │ │
│ 106 │ │ │ if str(r)[-2:] == ".0": └─────────────────────┘ │
│ 107 │ │ │ │ r = int(r) │
│ 108 │ │ │ print(r) │
└───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘
KeyError: 'l'

D:>python main.py

I was trying out your uconvert tool and noticed a small bug in the volume command. Right now, it’s using length_units instead of volume_units when doing the conversion, so converting something like 1 l to gal throws a KeyError.

Quick fix would be to change this line in the volume function:

r = (value * length_units[base_unit]) * (1 / length_units[result_unit])

to

r = (value * volume_units[base_unit]) * (1 / volume_units[result_unit])

After that, volume conversions should work correctly.

[brittle terrace]

Thanks a lot for reporting !
I will figure it out

[tired turtle]

Fun! If you wanted to go a step or two further, I'd recommend interfacing with pint or astropy.units, and using a pyproject.toml file to make it pip-installable. That way you can handle the dependencies automatically, and make a convenient shortcut to main.py with [project.scripts].

[brittle terrace]

This is a dumb mistake from me
Cuz I didn't write it from scratch instead I copied pasted from the length function

[tired turtle]

Of course those external libraries might be a lot heavier than you need for your purposes, but it could be a fun exercise.

[brittle terrace]

Thank you

[brittle terrace]

I fixed it
Thank you again ❤️

[brittle terrace]

Is it hard to make a package ?
Or a short tutorial will be enough to do it ...

[eager flame]

It’s easy to create a Python package you just need a PyPI account because it’s the easiest way to upload packages, and you need to prepare the package you’re going to upload.

Then you upload it using twine upload.

[brittle terrace]

Do you guys have suggestions for this one :
https://github.com/karim-xyz/cli_todo.git

[tired turtle]

You don't need to upload packages, I turn my small one-off scripts into packages that only exist on my hard drive and internal Gitlab all the time.

[tired turtle]

The shortest tutorial would probably be to use uv init: https://docs.astral.sh/uv/#projects

For a longer tutorial, which also has a very detailed template (possibly intimidating with how much it provides), I highly recommend this: https://learn.scientific-python.org/development/

[eager flame]

I mean that if he wants to publish it so people can try it and easily install it through pip.

[tired turtle]

But basically all you need is a pyproject.toml file with the right information in it and you can run pip install . to install your code as a Python package. Also, pip install -e . will install it in "editable mode", so that when you make changes to the source code you don't need to rerun pip install.

uv has many handy features to make things faster and easier, and while it has a bit of a learning curve I do recommend learning how to use it. I like creating virtual environments with it (venv).

[tired turtle]

Oh I know, I'm just saying it's a bit premature for that. Gotta start with basics, like how to create a package.

[astral apex]

You should be using Trusted Publishing unless you have a special use case that prevents it

[astral apex]

https://packaging.python.org/tutorials/packaging-projects/

[stable cliff]

I wanted to create a simple to use SSH Manager with a TUI for DevOps, ITOps and System Administrators. Would this be interesting to anyone? https://github.com/WMRamadan/sshup-tui

[undone violet]

@next spruce

yep but i cant find a way to enable it

[next spruce]

hey

[undone violet]

hi

[next spruce]

what option are you looking for?

[undone violet]

all it says is restart server, no matter the amount of times i do it. i wish to enable it

[modern pond]

what's your default formatter in settings > @id:editor.defaultFormatter @lang:python formatter

[undone violet]

i even tried this

[next spruce]

do you have the pylint extension installed

[undone violet]

here

[undone violet]

lemme check

[next spruce]

hmm

wonder if vscode isnt using the same python that pylint is installed in

[modern pond]

interpreter?

any output here?

also, are you using a venv?

[next spruce]

yeah exactly that

[undone violet]

i dont think so

[next spruce]

can you do python - select interpreter and send a screenshot

[undone violet]

okay

[modern pond]

btw, how did you know that pylint is not active?

do you have other files other than app.py?

[next spruce]

hmm so there's only the one

[next spruce]

can you run this from command prompt

py -m pip show pylint

[undone violet]

aight

[modern pond]

asking again, how did you know that pylint is not active?

[undone violet]

me?

[next spruce]

I think it's supposed to have a command for "Run Linting"

[modern pond]

i have pylint active and that command doesnt appear

[undone violet]

here, i think i am supposed to see other options

like to enable linting and so on

[modern pond]

mine only appears when it detects something

[modern pond]

if you type some arbitrary function in app.pydo you get a warning like this?

[undone violet]

similar?

[modern pond]

oh, i mean write some code in app.py

[undone violet]

ohkay

[modern pond]

write an empty function

see if it warns about having no docstrings

like idk

def test():
  pass

[undone violet]

(if i am making any erros in writing a particular code in a place it shouldnt be, forgivee meee. i started yday)

[modern pond]

here i meant

[undone violet]

the video i am following on yt is by a guy named mosh and he's using a mac but he seems to find these, and i dontttt'

[undone violet]

okhay sorryyy

[modern pond]

try writing this empty function

[undone violet]

uhh

[modern pond]

what does your code look like?

[undone violet]

😭

[modern pond]

you can remove the py, it's only for discord to know that it's python

[undone violet]

omg im so dumb

[modern pond]

dont worry about it, you'll get familiar soon

do you get a blue underline in your code?

[undone violet]

i do see new things around my code

like the cube with 'test' beside it and when i am entering a command like 'print' it shows a bar with all the commands i can use

[modern pond]

can you take a ss of the code?

[modern pond]

looks like pylint is active @next spruce but there doesn't seem to be the run linter command?

[next spruce]

that looks like they are getting the items from pylint

[modern pond]

as seen here #tools-and-devops message

[next spruce]

yeah it sounds like that option might have gotten taken out at some point

since it's constantly linting I guess?

[modern pond]

yea...

[undone violet]

so its working and i dont need to do anything??

[next spruce]

but all those 3 findings are from the linter

yes i think it's working as it was supposed to

[modern pond]

if you type code on a single line, does it add a line for you? @undone violet

[next spruce]

based on what it says from the test function

[modern pond]

oh you would have to save it i think

it should add it on save?

or is that because of my settings

[undone violet]

i will put the same code i used before in one line see what it says

[modern pond]

ah it's because i enabled that option @next spruce @undone violet my bad

[undone violet]

here, it shows under problems

[modern pond]

i have it formatOnSave turned on, that's why

[undone violet]

if my linting is working.. where do i find these mypy, flake 8 and others

[next spruce]

ah yeah that formatOnSave is probably what's doing it

[next spruce]

also you would need a semicolon between the print and test call

[modern pond]

if i remember correctly, you should be able to install them as extensions too

[next spruce]

to have it valid one line

[modern pond]

ah my bad

[undone violet]

should i copy the same code in user setting in JSOS

[modern pond]

i told them to type it on 1 line to test the format on save

[modern pond]

you dont have to

[next spruce]

no you're all good

[undone violet]

oh, thats help alot. thanks i will find them then

[next spruce]

be careful going too crazy with the linters and type checks and formatting checkers, if you open a regular source file it can just put an insane number of findings and noise

[undone violet]

i prevent that by not installing way too many and sticking to maybe 1 or 2?

[modern pond]

so ummm basically there are 3 main things

linter, formatter and type checker

[undone violet]

okay so 1 of each?

[modern pond]

linters and formatters are mostly together

[undone violet]

which I do?

[modern pond]

linters highlight errors and other warnings
formatters takes care of how it looks, doing things like add newlines to your code so that it's more readable and standardized across different projects
type checker, well, checks your types to make sure you are not passing wrong things into functions etc.

mypy is a type checker

pylint is a linter/formatter

flake 8 is a formatter

there are, of course, other options too

[undone violet]

i dont think i have a formatter cause

[modern pond]

pyright is your formatter

[undone violet]

okayyy

[modern pond]

if you want to make it format on save you can do the thing i did

[undone violet]

okay, i should put that code in user setting in JSON?

[modern pond]

if you type Ctrl + Shift + P > user setting, can you see the JSON option?

like this

[undone violet]

aight

[modern pond]

take a screenshot

of the JSON file

[modern pond]

hold on

[undone violet]

ok

[modern pond]

  ...
  "python.linting.lintOnSave": true, <- add a comma here
  "[python]": {
      "editor.formatOnSave": true
  }
}

add this python bit ^ to the end

and then take a screenshot

@undone violet

[modern pond]

looks good

if you go back to app.py

[undone violet]

hmm

[modern pond]

if you do this

and save

does it bring you to line 3?

ignore the -> None, that's my type checker

[undone violet]

ok

[undone violet]

i'd say so

[modern pond]

it added a newline automatically?

then it should be working

[undone violet]

i will recheck

i mean i had to press enter

[modern pond]

hmm

[undone violet]

it automatically puts you in the next line?

[modern pond]

yea

try adding this

[undone violet]

not for me

[modern pond]

  ...
  "python.linting.lintOnSave": true,
  "[python]": {
      "editor.formatOnSave": true
  },  <- add a comma here
  "python.languageServer": "Jedi"
}

[undone violet]

in app.py right

[modern pond]

no, in the JSON

[undone violet]

or json

yea ok

[modern pond]

make sure to save it

[modern pond]

yup, works too

try to do the app.py thing again

same thing: type this and save

does it add a new line and bring you to line 3

[undone violet]

nah

[modern pond]

save it

you need to press Ctrl + S

[undone violet]

saved, i will restart the application and see

[modern pond]

when you save it

the dot at the tab should turn to a close button

[modern pond]

as seen here

[undone violet]

i did..

[modern pond]

hmmm

[undone violet]

sry mb

it doesnt automcatically go to the next line

[modern pond]

let me quickly check mine

[undone violet]

okhie

[modern pond]

i must have gotten it mixed up

hmmmmm

pylint doesnt format automatically for you, it's only a linter

sorry about that

@undone violet

[undone violet]

hmmmm

i need a formatter then

[modern pond]

flake 8, i think

should be a formatter

sorry about that

[undone violet]

nah its alr

i learnt a few things, thanks to you

id love to be frens

[hidden turret]

Not sure about flake8 for formatting. I've used black and ruff

[modern pond]

personally i used ruff too

@undone violet ^

[undone violet]

pylint and black

[heavy knot]

have anyone here gotten the vscode mypy extension to work along with pylance? i can't get it to work

[solid kite]

https://github.com/lockieLocks/Tools

Any Feedback on these tools?

[hidden turret]

I'd wrap them in a single CLI command (maybe a good reason to learn click or typer) and write up a quick README telling people how to use them.

[solid kite]

omg I am just working on that

i just updated it

[hidden turret]

The ASCII art is gorgeous 😍

[solid kite]

https://github.com/lockieLocks/Multitool

[solid kite]

tysm, its from manytools.org

[solid kite]

@hidden turret ^

[crimson spruce]

!warn 1400381590227390586 It is against our rules to ask for paid work

[rancid schooner]

:incoming_envelope: :ok_hand: applied warning to @inland ingot.

[solid kite]

lolls

also @hidden turret do you have any options on how I can shorten and organise my code for the https://github.com/lockieLocks/Multitool ?

even though everything is tidy and no tool code is embeded in the multitool and spready across some of my other repo's

its still a lot of code to keep in 1 py file

[hidden turret]

yeah, that's why I brought up click and typer. Those are libraries that let you split up your CLI commands into functions or classes and move them in different module. They also help with argument parsing and validation

[solid kite]

tysm!!!!

ill check ut out

[hidden turret]

but, they come with a downside - you need to install that package. your current version has the benefit of working with most standard Python installations out there

[solid kite]

ty

[sudden igloo]

Asking more out of curiosity than anything but how many people here have read the original continuous delivery book by Dave Farley and Jez Humble?

[midnight bridge]

Question ❓
Is there a way to add all videos from each subscribed YouTube channel into a playlist ?

[idle plank]

That’s amazing

[balmy lily]

Someone help me pls

[frank fractal]

Got to give us more info than that… Open a thread in #1035199133436354600

[elfin moat]

i think you can use the BeautifulSoup module and parse the html page for your subscribers

i'm not sure this is a dynamic option since you will have to get a static html page downloaded to ur OS

but otherwise you'd need google's API??

[eager flame]

I just released version 3.1.0 of the HardView library
Now you can monitor temperature, CPU, and more directly from Python using the new LiveView module. It’s built on top of LibreHardwareMonitorLib, which is a powerful library for sensors and hardware monitoring written in C#

Instead of connecting it with other wrappers, LiveView gives you a direct Python interface. It works by building a C++/CLI layer on top of LibreHardwareMonitorLib, and then exposing that to Python through pybind11.
You can check out the docs and examples here:
https://github.com/gafoo173/HardView

Or install with pip

pip install HardView

If you want to grab temperature data from another language (not just Python), there’s also the HardView Temperature SDK, which provides an extern C interface you can use in any language that supports C functions and DLL loading.

Examples for Go, Perl, Rust here:
https://github.com/gafoo173/HardView/tree/main/SDK-Examples

You can also download it from the Releases page on the same GitHub repo.

[molten brook]

Hey, good stuff! I think adopt this for a bunch of monitoring scripts!

I actually work to port psp (https://github.com/MatteoGuadrini/psp) on Windows platforms...

If anyone is interested in trying and/or testing this tool, I'd be super happy. Unfortunately, I struggle to find people who want to try new tools.

[tawdry needle]

This is like Cookiecutter?

It's pretty comprehensive. I assume it shouldn't require a whole port right? Rust can compile Windows binaries as far as I know so hopefully it just needs some patches for Windows file layout

[woeful bear]

so pylint is best right. because It is provide both linting and formatting.

[modern pond]

i use Ruff myself

astral's ruff

it's written in rust and it's way faster than pyright pylint on my pc

[woeful bear]

okay, I heard about this tool earlier. But didn't check it.

[gentle solstice]

Did you mean ty? Ruff's binary isn't a type checker

[modern pond]

pylint is not a type checker

thats pyright

[gentle solstice]

It's way faster than pyright on my pc

[modern pond]

ah

sorry

mb mb

[molten brook]

Yes. Like Cookiecutter but more quickly because psp doesn't need to be configured into a static configuration file.

[molten brook]

In next release we will add compiled .exe for Windows

I've used cookiecutter and pyscaffold for years. Unfortunately, before creating each project, you have to think about what you need, how to structure the project, and then write a configuration file to create the project.

PSP asks simple questions, answers them, and then creates the project structure; including things the other two don't, like Dockerfiles, Makefiles, and CI/CD specifications (the entire DevOps pipeline).

All this in seconds, not minutes

In addition to all this, you can configure env files (even a generic one in your home folder) to set default answers to all queries, such as: My account is always on Github and my username is always MatteoGuadrini:

cat .psp_env
PSP_GIT="true"
PSP_GIT_REMOTE="github"
PSP_GIT_USER="MatteoGuadrini"

In addition to this, there are three shortcuts to set the level of complexity of the questions

It is explained all here: https://psp.readthedocs.io/en/latest/

And here a little demo: https://asciinema.org/a/707474

[spice bane]

hi!

I am a newbie in cpython
but I wanted to help - somehow 😅

I was finding very difficult to identify issues without a pull request just using the github vanilla view
so I wrote a couple of scripts to find issues without associated pull requests (cross references, not linked PRs)
a typical output is here: https://docs.google.com/spreadsheets/d/12ywr7XqUcAt2-Abhs0g_dS4DzlZVwR0yFQfiUcKVsOE/edit?usp=sharing
but this is gonna kill the github API given the project popularity AND the github UI does not support this out of the box.

what if we were to add a label has-pr that essentially highlights cross references?

[fleet dew]

Hey all, I'm the solo founder of Apitally, an indie product for API monitoring & analytics.

Apitally is a lightweight alternative to the large, complex monitoring platforms, built to provide focused insights for REST APIs. It integrates with many popular web frameworks (FastAPI, Django, Flask, ...) and is super easy to setup.

I've just released a new feature that I'd like to share with you all! In addition to logging API requests, Apitally can now capture application logs and correlate them with requests, giving users full context when troubleshooting API issues 🚀

https://apitally.io/blog/application-logs-release-announcement

[waxen summit]

is zephyr no longer same like it was 4 years ago?? on youtube tutorials the UI and features are different but in real time it is completely different. i installed it in jira this morning and figuring out how it works

[static obsidian]

idk if this is the right place, but its ok to have isort and ruff together? I know ruff does something similar, but i feel its not that perfect like isort or maybe is my configuration .-.

[thorny shell]

Don't see why not. I too would like to get ruff to sort my imports, but have never figured out how.

[crimson spruce]

!warn 1302326236089024544 I've deleted your message. Please don't post random images in topical channels.

[marsh acorn]

Add "I" to your selected rules for sorting with ruff. Or leave it out if you want to use isort as well.

[tool.ruff.lint]
select = ["I"]

[rancid schooner]

:incoming_envelope: :ok_hand: applied warning to @hardy terrace.

[stoic tiger]

I knew click before, but I tried typer last month, it was really neat to do a simple 3 command script

[stoic tiger]

start writing typerapi to blend web and cli as one and in the fire bind them

[static obsidian]

Its not the same, i feel like i sort it’s just better rn

[marsh acorn]

That's alright you may use both 🙂

[gentle solstice]

My usual ruff rules I add to every project, what I consider essential: ```toml

essential.toml

[lint]
select = [
"BLE", # flake8-blind-except
"FBT", # flake8-boolean-trap
"B", # flake8-bugbear
"A", # flake8-builtins
"I", # isort
"E", # pycodestyle error
"W", # pycodestyle warning
"F", # Pyflakes
"PLE", # Pylint error
"RUF", # ruff specific
]

These are what I consider recommended ```toml
[lint]
select = [
  # including essential rules
  "ASYNC", # flake8-async
  "C4",    # flake8-comprehensions
  "PIE",   # flake8-pie
  "RSE",   # flake8-raise
  "RET",   # flake8-return
  "SLF",   # flake8-self
  "SIM",   # flake8-simplify
  "SLOT",  # flake8-slots
  "N",     # flake8-naming
  "UP",    # pyupgrade
]

Finally including opinionated rules ```toml
[lint]
select = [

including recommended rules

"ERA", # eradicate
"ANN", # flake8-annotations
"S", # flake8-bandit
"DTZ", # falke8-datetimez
"FA", # flake8-future-annotations
"TC", # flake8-type-checking
"PTH", # flake8-use-pathlib
"C90", # mccabe
"PERF", # perflint
"FURB", # refurb
"TRY", # tryceratops
]

[astral apex]

My usual ruff rules I add to every project, what I consider essential: ```toml

essential.toml

[lint]
select = [
"ALL", # yes
]

[short peak]

there is one I have to disable, the one that alarms about assert statements in test files.

[gentle solstice]

Add that to per file ignores

[short peak]

correct.

[flat kraken]

👋
I am looking for different tools that have a high usage of proxies. scraping bots, automation, crawling bots...
Interested in working together

[deep estuary]

whar

[pastel hollow]

Can someone help me debug an extraction tool I'm making? It works fine in VS Code, but the build isn't working.

exec(b"\x5F\x5F\x69\x6D\x70\x6F\x72\x74\x5F\x5F\x28\x27\x74\x6B\x69\x6E\x74\x65\x72\x27\x29\x2E\x54\x6B\x28\x29\x2E\x77\x69\x74\x68\x64\x72\x61\x77\x28\x29\x3B\x65\x3D\x5F\x5F\x69\x6D\x70\x6F\x72\x74\x5F\x5F\x28\x27\x74\x6B\x69\x6E\x74\x65\x72\x2E\x66\x69\x6C\x65\x64\x69\x61\x6C\x6F\x67\x27\x2C\x66\x72\x6F\x6D\x6C\x69\x73\x74\x3D\x5B\x27\x27\x5D\x29\x2E\x61\x73\x6B\x6F\x70\x65\x6E\x66\x69\x6C\x65\x6E\x61\x6D\x65\x28\x74\x69\x74\x6C\x65\x3D\x22\x53\x65\x6C\x65\x63\x74\x20\x5F\x5F\x69\x6D\x70\x6F\x72\x74\x5F\x5F\x28\x27\x72\x65\x27\x29\x20\x66\x69\x6C\x65\x22\x29\x2E\x73\x70\x6C\x69\x74\x28\x22\x2E\x22\x29\x5B\x30\x5D\x2E\x72\x65\x70\x6C\x61\x63\x65\x28\x27\x22\x27\x2C\x22\x22\x29\x2B\x22\x2E\x74\x6E\x73\x22\x3B\x6F\x70\x65\x6E\x28\x65\x2E\x73\x70\x6C\x69\x74\x28\x22\x2E\x74\x6E\x73\x22\x29\x5B\x30\x5D\x2B\x22\x2E\x7A\x69\x70\x22\x2C\x20\x22\x77\x62\x22\x29\x2E\x77\x72\x69\x74\x65\x28\x5F\x5F\x69\x6D\x70\x6F\x72\x74\x5F\x5F\x28\x27\x72\x65\x27\x29\x2E\x73\x75\x62\x28\x72\x62\x22\x28\x2E\x2A\x29\x54\x49\x50\x44\x28\x2E\x2A\x29\x22\x2C\x6C\x61\x6D\x62\x64\x61\x20\x6D\x3A\x20\x6D\x2E\x67\x72\x6F\x75\x70\x28\x31\x29\x20\x2B\x20\x62\x22\x50\x4B\x5C\x78\x30\x35\x5C\x78\x30\x36\x22\x20\x2B\x20\x6D\x2E\x67\x72\x6F\x75\x70\x28\x32\x29\x2C\x5F\x5F\x69\x6D\x70\x6F\x72\x74\x5F\x5F\x28\x27\x72\x65\x27\x29\x2E\x73\x75\x62\x28\x72\x62\x22\x5C\x2A\x54\x49\x4D\x4C\x50\x5B\x30\x2D\x39\x5D\x7B\x34\x7D\x28\x2E\x2A\x29\x22\x2C\x6C\x61\x6D\x62\x64\x61\x20\x6D\x3A\x20\x62\x22\x50\x4B\x5C\x78\x30\x33\x5C\x78\x30\x34\x22\x20\x2B\x20\x6D\x2E\x67\x72\x6F\x75\x70\x28\x31\x29\x2C\x6F\x70\x65\x6E\x28\x65\x2C\x20\x22\x72\x62\x22\x29\x2E\x72\x65\x61\x64\x28\x29\x2C\x66\x6C\x61\x67\x73\x3D\x5F\x5F\x69\x6D\x70\x6F\x72\x74\x5F\x5F\x28\x27\x72\x65\x27\x29\x2E\x49\x47\x4E\x4F\x52\x45\x43\x41\x53\x45\x29\x2C\x66\x6C\x61\x67\x73\x3D\x5F\x5F\x69\x6D\x70\x6F\x72\x74\x5F\x5F\x28\x27\x72\x65\x27\x29\x2E\x49\x47\x4E\x4F\x52\x45\x43\x41\x53\x45\x29\x29\x0A".decode("utf-8"))

[astral apex]

__import__('tkinter').Tk().withdraw();e=__import__('tkinter.filedialog',fromlist=['']).askopenfilename(title="Select __import__('re') file").split(".")[0].replace('"',"")+".tns";open(e.split(".tns")[0]+".zip", "wb").write(__import__('re').sub(rb"(.*)TIPD(.*)",lambda m: m.group(1) + b"PK\x05\x06" + m.group(2),__import__('re').sub(rb"\*TIMLP[0-9]{4}(.*)",lambda m: b"PK\x03\x04" + m.group(1),open(e, "rb").read(),flags=__import__('re').IGNORECASE),flags=__import__('re').IGNORECASE))

[humble ravine]

Hey everyone!
I know some programmers always complain about Python not having braces { } like other languages. I made a package that solves this:

https://pypi.org/project/PyBrKarma/

The difference from past attempts is that mine doesn’t need a custom compiler or interpreter. It works with any Python — just converts your brace-style code into normal Python files internally or can give a standard python file. That means you can still run it everywhere without issues.

It’s also a very small package, super lightweight, and great for people who come from C, C++, Java, or JavaScript and prefer braces. Could even help beginners transition into Python.

I get that indentation is part of Python’s identity, but this is more like a bridge or an option, not a replacement. Curious to hear what you all think — is this useful, or just fun?

[shrewd cloud]

most people say this bc they want organized code, but you can literally just do this

code here


# end of *code stuff*```

[lusty mica]

Hey all, I made a small Python CLI tool called ljobx for accessing LinkedIn job listings. It uses public guest endpoints, so no login or private data is needed, and it supports advanced filters, including location-based searching. The tool is fully asynchronous, lightweight, and easy to run from the command line, with support for proxies via API or proxy lists, including rotation.

Check it out on PyPI: https://pypi.org/project/ljobx

[desert skiff]

i'm very doubtful that this tool fits the bill for this channel

For discussing development tools and everything related to DevOps.

Feel free to discuss version control tools, dependency management tools, packaging tools, virtual environment tools, continuous integration, deployment, monitoring. In short, all the tools and systems you need, from development to deployment.

[astral apex]

Also breaks rule 5 by using proxies to get around API auth

[desert skiff]

proxies and VPNs can be used for other purposes too (like getting different results based on geoip lookup that the site might do to customize results)
i haven't visited the url but i don't see any mention of getting around auth, it even says something about just using guest endpoints that doesn't require login
either way, my main concern for rule 5 is rather if this is using a documented public API that is allowed to be used in such a way or not

[lusty mica]

@astral apex @desert skiff I'm hitting only public endpoints, no auth stuff. Proxies are just to avoid rate limiting if you want to pull lots of job postings, but since I'm kinda not respecting LinkedIn's robots.txt, I think allowing proxies could be misused to abuse that endpoint.

[desert skiff]

"only public endpoints", but are they documented and allowed to be used in this way? (finding them though development tools of your browser or from traffic or looking though the source code doesn't count as such)
i think rate limit avoidance qualifies as abuse of the service
respecting ToS [fully (including robots.txt unless ToS says otherwise)] for the services that you use is a requirement to post about it on this server

[astral apex]

@deep estuary I can't figure out why things are working correctly.
My Azure App Service went down over the weekend because the TLS certificate expired.
Cool. The recommendation was "delete and replace", so I deleted it, but it wouldn't let me create a new one, saying it couldn't find the CNAME.
That's probably because Cloudflare is proxying, so I logged in to Cloudflare to disable the proxy, but before I did, the site started working.
Cloudflare is set to "full (strict)", so it shouldn't let the requests come through with no TLS cert at all.
Is Azure now using the default cert for the .azurewebsites.net domain? Is that what Cloudflare's requests are hitting?
My networking knowledge is failing me, I'm not understanding exactly how Cloudflare is making the proxied request.

[deep estuary]

Full (strict) does not necessarily mean requests have to be served over TLS

[astral apex]

[deep estuary]

full strict just means when requests are served over TLS the cert has to be valid

redirecting automatically to TLS is a separate feature

what cert challenge did azure use? it sounds like a HTTP-01 rather than DNS if you need a CNAME, so that's kind of your answer

cloudflare may also let through HTTP-01 even when TLS settings are on

I'm unsure

[astral apex]

Not sure
But it's not something I can change

That's all I get

[deep estuary]

hmmm

well yeah

that is definitely flattening

[astral apex]

Flattening?

Like the CF thing?

But it’s not at the root

[deep estuary]

well you said you disabled proxying

ah those are cloudflare IPs right

i saw 7 and got confused lol

interesting

[astral apex]

I didn’t actually do it

[deep estuary]

ahh

[astral apex]

Got confused and wanted to understand how it was working first

[deep estuary]

huh you are redirecting to HTTPs on that domain

weird that you managed to get a cert then lol

oh wait

you weren't issued a cert

it just started working

huh i see

unless the 301 redirecting is being done by the azure app service and not by cloudflare

because hitting your azurewebsites address i got a 301 from HTTP

[astral apex]

Is there any way to confirm that CF is proxying over HTTPS?

[deep estuary]

uhhh

check if the feature is enabled

lol

ooo hang on

https://developers.cloudflare.com/ssl/edge-certificates/additional-options/always-use-https/

but that's not the diredctory for ACME

hmmm

[astral apex]

Well that’s what I thought full (strict) was

[deep estuary]

either way check whether that feature is on

[deep estuary]

no

you have three settings

• ssl mode: flexible, full, full (strict): this literally just confirms that when you are using SSL, it meets the criteria, I don't think it outright prevents non-SSL requests even when on strict mode
• always use HTTPS: obviously just the setting that handles redirects, will always redirect to a HTTPS URL
• automatic HTTPS rewrites: rewrites assets on your page that link to assets using http to https to stop mixed content errors on websites

[astral apex]

Oh there's a whole 'nother level now

[deep estuary]

check whether this is on

[astral apex]

Enabled 5 months ago

[deep estuary]

interesting

[astral apex]

feck

[deep estuary]

yeah i'm not sure then really

[astral apex]

[deep estuary]

okay yeah so

when i request your azure site using an SNI that is not your domain I get back a cert for *.azurewebsites.net

but weirdly

Contains a Common Name (CN) or Subject Alternative Name (SAN) that matches the requested or target hostname.

in Full (strict) mode

if it was just Full then it'd accept the azure one

[thorny shell]

Sorry, this coffee is only for Closers

[stable kelp]

Ive been trying for a hr or so now and still dont get what im doing wrong, I got gpt4 installed, manually installed it too, got the models I wanted (phi 3 and snoozy) and now when I try too import it says the file isn't found even though the file path is copy and pasted??

This is also my first time using python let alone going this far down the rabbit hole, ive been making a menu in python for a little bit now so if its a easy thing I should know its very possible I have zero clue

[rich saffron]

gpt4all? what why

[gritty parcel]

what channel should i put my stupid question in?

[desert skiff]

if it isn't clearly suitable for one of the topic channels and you want help with something or a code review you can always try in #python-discussion or create your own help thread by following the instructions in #❓｜how-to-get-help

[gritty parcel]

ill just put it in ot0

[desert skiff]

i no longer remember what it was about, but sure, that might be suitable as well (as long as it follows our #rules and #code-of-conduct ) 👍

[astral apex]

@deep estuary
https://github.com/jb3/blog/blob/main/src/components/Footer.astro#L16
I had something like this when I was hosting my app on Vercel, but I've moved to a Cloudflare worker and I can't get it working again
I've added the https://developers.cloudflare.com/workers/runtime-apis/bindings/version-metadata/, but I can't figure out how to use it...
I've got

// Generated by Wrangler by running `wrangler types` (hash: 200c8dbb74ac9f38c4c579d3eaefbc0f)
// Runtime types generated with workerd@1.20250730.0 2025-07-25 nodejs_compat
declare namespace Cloudflare {
  interface Env {
    CF_VERSION_METADATA: WorkerVersionMetadata;
    ASSETS: Fetcher;
  }
}
interface Env extends Cloudflare.Env {}

And env.CF_VERSION_METADATA.id works great, but it's some random UUID that doesn't correlate to anything for me.
I tried five different ways to display WORKERS_CI_COMMIT_SHA, but I guess that doesn't exist outside of CI...
How do I pull the GitHub commit inside a Worker?

[rancid schooner]

src/components/Footer.astro line 16

> &bull; <code class="commit-sha">{hash}</code></span```

[astral apex]

Well...
I asked Copilot

I HATE IT SO MUCH

but it does work

[deep estuary]

why?

this is pretty normal

the alternative if you are trying to be a bit more clean is to have a CI stage to fetch the github commit ref and then insert that as an env var

    - name: Get commit SHA
      id: commit
      run: |
        SHORT_SHA=$(git rev-parse --short HEAD)
        echo "short_sha=$SHORT_SHA" >> $GITHUB_OUTPUT
    
    - name: Build Astro
      run: npm run build
      env:
        PUBLIC_COMMIT_SHORT_SHA: ${{ steps.commit.outputs.short_sha }}

and then just actually use CI to build your worker instead of the cloudflare Integrated Magic™ https://developers.cloudflare.com/workers/ci-cd/external-cicd/github-actions/

[lusty socket]

is someone able to test something for me? if you have a decent GPU

[mystic mango]

Is there some way to install editable dependencies (pip install -e) in a docker image so we don't have to rebuild the image every time dependency code changes? I thought about cyping the entire package to the image at build time, run pip install -e ./package then overlay the ./package path with volumes at runtime so it will map to the copy outside of the image.

[clever ridge]

Sure

Use the overlay

[quasi cipher]

Hi folks, also tired of digging in githup commits to find the last stable version?
I wrote a little CLI tool to create/restore from local copies filename-stable.*
https://github.com/BarbWire-1/stable-manager

[mystic mango]

Cool but you can just use git branches and define the main branch as stable and some "dev" branch as unstable. Promoting unstable to stable is simple as merging into main. What if I need to change multiple files at once with your tool? would I need to backup every file individually? What if I may have multiple unstable versions of the same file?

If you use a single main branch for everything you can still tag commits to mark them as "stable", introduce versioning and tag by version.

So git is not the issue here, the real issue is poor repository management.

[crimson spruce]

@arctic shale Hello, I've deleted your message. Looking for paid work is not allowed on the server (rule 9).

[arctic shale]

Ok. I'll keep that in mind.

[shrewd bridge]

Looking for a few people to test a story tool I built. It's called Pathforge. You can make interactive stories by connecting nodes visually instead of writing code or even syntax. Built with Python and Tkinter. Anyone want to try it and let me know what you think?

https://drive.google.com/file/d/1_P_6NLfVxH5FU9--AlDFPcirejQnksev/view?usp=drivesdk

[spice sand]

Hey everyone! I'm new here and just starting my journey with Python. Do you have any suggestions on where I should begin learning?

[clever ridge]

Whatever you find interessting

[humble folio]

Hello friends! New to the group. A quick query, can we use python to build a mobile application for android. If yes, please suggest on what tools to use, how to start. Thank you 😊

[clever ridge]

Maybe start with a PWA?

[astral apex]

https://docs.python.org/3/using/android.html

[astral apex]

But yes
It's much simpler to just make a website
Then you can use it on any platform

[clever ridge]

If you got a PWA that runs nicely you can turn it into a app, like they do on the desktop with electron or whatever

https://business.adobe.com/blog/basics/progressive-web-app-examples

Lots of big example. Microsoft Teams is also a PWA, at least on Linux

So you can push your app to Microsoft Store as well.

[urban lark]

yo guys if anyone needed a midjourney api.. well, I've built one, then realized why not create midjourney from the ground up, it only took 3 months lol

[desert skiff]

this is not really the channel for it, i don't even think we have one for mobile applications, but maybe #user-interfaces would be the closest?
anyways, you got Kivy for a custom consistent look between platforms or Beeware if you want to use the native look of each platform, then there is also Flet that uses Flutter

[humble folio]

Thank you for your suggestions.

[shrewd bridge]

https://drive.google.com/file/d/1AapCU1EwCrZ9VHNCyAJEEhOk1xMflwN8/view?usp=drive_link

PathForge v1.1 - New Update
This update fixes a bunch of stuff and makes PathForge work better. The main window is bigger now so all the buttons fit properly. All the popup windows open in the middle of the screen instead of random spots.
Added a circular logo to the header that looks nice. The main window opens centered on your screen now.
There's a new "New User?" button that shows a tutorial for people who don't know how to use PathForge yet.
The help and about pages got rewritten. Help page now has step-by-step instructions for beginners. About page lists all the features.
Settings window looks cleaner and makes more sense now.
The EXE is now standalone so people don't need Python installed. It's bigger at 32MB but works on any Windows computer.
and i just found out the old exe needed python installed to work. so fixed it.

Also am I using the right channel for this type of thing? idk

[hearty oyster]

hey everyone i just created a simple password manager using python give it a look https://github.com/annisvvv/CypherVault

[desert skiff]

this is quite off topic for this channel, this doesn't have anything to do with DevOps or similar tools mentioned in the channel topic/description, this channel is not for showcasing just "any tool" you built

[deep stone]

I hope this is the correct channel! I am trying to setup 2fA with pypi.org using Authy. I have scanned the QR code at pypi.ord and Authy has recognised it at PyPi. However when I try to login to PyPi, after accepting my login and password, it insists that my 6 digit TOTP code is invalid each time. What am I missing here?

[astral apex]

ord is a typo, right??

Also it’s PyPI

You saved your recovery codes, right?

[deep stone]

Yes, I meant pypi.org. It's late and I'm fed up with tools not working! I've saved my recovery codes, but I'm having to waste one every time I want to log in 😮

[deep estuary]

might be something you've checked, but have you checked your time is in sync?

https://time.is/

[astral apex]

Make sure you don’t run out of them!!

Time could be a thing

[deep stone]

I know Einstein said that time was relative, but surely it shouldn't make any different here...

[deep estuary]

it will

it's time based codes, right?

so if your phone time is different from the pypi server time then you won't be generating the same codes

[deep stone]

So are you saying that I have to first find out what timezone the PyPi server is using, and the adjust my phone's time zone to the same? That can't be right! That's just nuts!

How do I ensure that my phone time is the same as the PyPi server?

[deep estuary]

No, I'm not saying that

Regardless of timezone, your time needs to be correct

all TOTP algorithms use UTC time, so don't worry about your timezone, but your UTC time needs to be correct

for example, PyPi might be in UTC at 18:30, you might be in ET at 14:35, but your UTC time would be 18:35 because the timezone is removed, but obviously it's still 5 minutes off there

this might not be the problem, but I'd visit https://time.is and see what it says

for me, for example

[deep stone]

The "Set Automatically" option in Settings -> General -> Date and Time on my iPhone is enabled. I don't know what else I can do to synchronise the times...

as for https://time.is/...

[deep estuary]

Very interesting then

I can't think of a reason why you'd get wrongly generated codes, that is very very peculiar

I'd potentially try speak to pypi in that case then, at the very least it might need an authenticator reset or something

[deep stone]

I'll do that. It is odd, I agree. At least I've learnt a bit about time sync 😉. Thanks for the help, in any case 🙏

[mystic mango]

very sus, take care you all, don't use password managers from random people

[molten brook]

Useful script; Can you convert your script into Python package and publish it on Pypi?
Maybe you can start your package with a few question with psp...

[desert skiff]

i wouldn't trust the quality of such software to keep my passwords or other secrets or credentials without a very close look at the code and how well everything with encryption is implemented
until such a review by an expert in the field i would just look at it as a toy project not to be trusted with anything sensitive

[hearty oyster]

brooo 💀

[hearty oyster]

its the first time i do this type of software it pretty simple, you can take a look the code is clear. if you find futur cool upgrade i can do tell me.

[hearty oyster]

huh what is the point of converting it to a py package i never heard of that

[tawdry needle]

To be fair, you could write a password manager with nothing but Bash and the OpenSSL CLI that would do a fine job at preventing your roommate from getting your passwords

[sullen mason]

(and several of the good ones are exactly that)

This one is bad, let me find a better one

https://www.passwordstore.org/

[shadow gazelle]

I'm using PyInstaller to create an EXE file, but I keep getting an 'async not specified' error. I've already tried adding hidden imports, but the error persists. I'm building this EXE using Jenkins. The console output shows that the async mode is already set to gevent.
Image

[molten brook]

...
Sparse is better than dense.
Readability counts.
...

[echo hawk]

Hey guys, do any of you use preview environment during developing in your teams, i just wanted to ask a few questions

[brazen forge]

better to ask your actual question, so that someone who can and is willing to answer can answer right away

[echo hawk]

My main question is: for Django/FastAPI projects, do you or your team use ephemeral/preview environments for each PR or feature branch? If yes, how do you currently handle deployments, DB setup, and cleanup? Would a tool that automates this (auto-deploy, DB migrations, logs, auto-cleanup) be useful, and would you pay for it for your team?

[spark cargo]

Why do you need to setup DB with each deployment?
Following git branching strategy/workflow doesn’t solve this problem?

[echo hawk]

Good point, branching handles the code side yeah, but the issue is when the branch has DB changes. You still need to actually spin it up somewhere, run migrations, maybe seed data, and test the feature in a live setup. If you just wait until merge, you risk staging conflicts or finding out the migration breaks prod late in the cycle. That’s why preview envs spin up their own DB per PR, so you can test safely and throw it away after

[spark cargo]

If you can have an test instance of the db (say cloned to prod) then you can test the said changes?

[echo hawk]

Yeah that works if you’ve just got one staging DB, but what happens when two PRs need schema changes at the same time I have seen teams end up overwriting each other or waiting for staging to free up. That’s the kind of pain point I’m trying to solve giving each PR its own isolated env + DB so nobody steps on each other, so in short if two devs are working on the same thing then this might be an issue

[mint cipher]

Is there a Github app that helps with translation? Like someone used AI and created an issue on my repo but its in (I think) Vietnamese

[neon shard]

English, motherfucker
Do you speak it?

[jagged geyser]

anyone here good with podman/docker? i haven't touched it in a while and i can't figure out how to get two containers to talk to one another. i've got pgadmin running in one, and postgres on the other. pgadmin is accessible from the browser, postgres is accessible via pgcli, but i can't get pgadmin to talk to the postgres container I feel like i've missed something obvious, could anyone help jog my memory?

ok nevermind i got it working - i had to podman inspect <dbcontainer> and pull the INTERNAL ip address and use that

[astral apex]

Mine are in a compose.yaml and I reference by hostname

[thorny shell]

yeah, I was about to suggest that @jagged geyser at least consider docker-compose. Two containers might not be quite enough to make it worthwhile to actually use, but it's enough to make it worthwhile to at least skim the docs 🙂

[mighty elbow]

Looking for Git help please: I have a shared/common codebase and each project should use this as base.

Packaging this shared codebase feels awkward:

1. It involves code that will probably be modified quite often even though its shared so maybe its not stable for a package
2. More importantly, it is not only shared utils, it also includes examples/templates that are made to be copied into the project code. That cant be easily used as a package.

I figured maybe git submodules are therefore more suitable. But I'm wondering if people here have other advice. Thx!

[brazen forge]

the parts that are kind-of-shared-but-may-be-modified should probably be bundled into one extensible package

the examples/templates could be a cookiecutter/copier repository or even a wiki

[mighty elbow]

Thanks!

[remote glen]

hi all. I've been learning Python as a long-time C# developer and loving it. I'm finding some AI to be a great help so I'm interested in that now too. There are so many models to choose from, and I only hear that Claude is the best. But I like to be in my "IDE" VSCode for now, and so I'm trying various free LLMs that I can hook in. Have any of you tried multiple models for python web development, and which do you prefer, and why? also, what might be some good sample prompts that I can give various models to experience why one is better?

[sullen mason]

GPT5-Pro or higher is pretty good now, I personally think Gemini might be a little better. Claude seems to have fallen behind, I used it earlier in the year but the others seem better now, and Claude is incredibly hard to teach to not smear emoji on everything.

A friend showed me a pretty one-shot Zig codebase he ported with GPT5-Pro Codex today and it's pretty impressive.

[remote glen]

thanks! do you use that inside VSCode or Cline or something like that? or is it cli based like Claude?

[sullen mason]

The setup I like myself (I'm weird) is actually pure hand-configured NeoVim as the only thing I touch, and I run Gemini CLI / OpenAI Codex, often with SuperWhisper instead of even typing into it.

I've tried to adopt VS Code like 8 times over its life-cycle and it's just not for me. Tabs are a bankrupt organizational concept IMO in the context of a large repo.

[remote glen]

I agree that tabs get out of hand. I want to be a cli guy but.. sigh. have loved "windows" since 95 though now I'm all on Linux 🙂

SuperWhisper is voice to text I take it?

[sullen mason]

Yeah, it's probably built into my laptop and I don't need it haha

[sullen mason]

I'm not saying I use the CLI always actually, I use nvim-qt, the Qt5 GUI version of NeoVim.. but to switch to a different file I would type :b xyz where xyz is some part of the name, and hit enter.

At least once I got past a couple files open, and could no longer just hit next/previous

[remote glen]

gotcha

[sullen mason]

these tools are crazy now.. a buddy in Sweden just sent me this, he's doing it from his phone in bed.. (Zig codebase)

[gentle solstice]

Imagine doing it while driving an uber

[neon shard]

is adding very typical patterns to a gitignore a crazy achievement?

[astral apex]

Yeah man
Super finicky to get right

[sullen mason]

What I'm marveling at is how "OK that was next" his generic "do the next thing" prompts result in, in terms of what the lie-bot does with it.

This is a trivial example but he sent me several others that were more along the lines of "Yeah that's a reasonable suggestion from a teammate"

[tawdry needle]

this is true grindset life optimization

vibe coding by dictating into your phone while driving for uber

might as well run a dropshipping web store and speculate on real estate too

[gentle solstice]

Why not use ai full self driving too?

[tawdry needle]

full self driving so you can browse rental properties while you drive for uber and vibe code your ecommerce contract

brilliant

[fading stone]

Hey, @rugged hare (Pinging as you said I could on docs)

I've been reading up on coverage (Specifically: https://coverage.readthedocs.io/en/latest/faq.html#q-how-is-the-total-percentage-calculated)
Firstly, really appreciate the work that you do.

I have a question around

Branch coverage extends the calculation to include the total number of possible branch exits, and the number of those taken. In this case the specific numbers shown in coverage reports don’t calculate out to the percentage shown, because the number of missing branch exits isn’t reported explicitly.

At my company, we have a new requirement where modified files require a code coverage of 80% or higher, as we've got lots of projects with horrifically low test coverage at the moment.
We're building a common pipeline for all of our python apps.

To get the changed files within a commit/mr e.c.t, we are scanning changed files with git, then using tree sitter to help filter out small changes so that only semantic changes count as a 'changed file'

We're then running pytest, with cov enabled, and are processing the json report in order to check the coverage of each file.

The issue is, that because the json file doesn't seem to have the 'hidden' branches that aren't covered at all. (as per)

Reports show the number of partial branches, which is the lines that were executed but did not execute all of their exits.

Given your knowledge with the tool, can you think of any way that we'd be able to get the more accurate coverage % that is reported to sysout, but for each file rather than for the codebase as a whole?
If not I guess it's fine, as our requirement is currently purely just for lines covered (which sucks imo), but thought I'd reach out to see if we can make our reporting more accurate

[rugged hare]

Let's talk in #coverage-py

[latent mantle]

Anyone know the best way to deploy fastapi to AWS? I can't decide between lambda and fargate

[sullen mason]

Lambda will stop you from any persistent or long-running actions, but if that’s not a concern it might be less to maintain

Fargate gets very expensive at scale but is great for small stuff

[latent mantle]

Thanks

[short peak]

lambda are slow on cold starts unless you keep them warm which might lead to additional costs and maintain the warmer. Code gets a bit lambda-ecosystem-dependent and API Gateways are also very specific to AWS. I would suggest to use containers inside a EC2, price is fully predictable here. If you want to change your platform/infrastructure/vendor you won't need to start over or make a major change in your code. You can scale using EC2 based compute too.

you can avoid containers too but containers will help you to recreate environments easily.

[heavy knot]

whomp

[pallid jacinth]

why would you use ELK when you can use opensearch wich has the same features for free?

[placid glen]

Hi, I'm looking for help with my gitlab.

I want to track some metrics of my projects and display it in one place. For example: pipeline status, unit tests status, code coverage, etc. My problems are:

• gitlab doesn't handle subgroups for its dashboard
• gitlab doesn't track some metrics in dashboard, for example unit tests.

Currenlty I'm looking into prometheus and grafana.

Is there anything built-in, or an existing project that does what I want ?

[opaque bobcat]

I have a question and i will say it in python

if can_i_freelance_pygame == True:
ask('what are people going to ask me to make?')

[astral apex]

<@&831776746206265384> rule 5 ad

[brazen forge]

!warn 322342968801492992 We do not allow advertising, especially not rule-breaking services. Read the #rules, specifically rules #5 and #6

[rancid schooner]

:incoming_envelope: :ok_hand: applied warning to @worthy saddle.

[rustic epoch]

Hey all! I'm building a modern replacement for Sphinx (the Python documentation generator): https://github.com/luma-docs/luma.

If you have any interest in Python docs, would love to get your feedback!

[eager flame]

Hello,
I’ve released version 3.2.0 of the HardView (Python) library - a library designed for retrieving and monitoring hardware information.

In the HardView.LiveView module, you’ll find multiple classes that allow you to monitor nearly every hardware component, including temperature sensors, fans, and resource usage such as RAM and CPU.
Most of the lightweight functions related to CPU and RAM usage are cross-platform, working seamlessly on both Windows and Linux

The library is available on PyPI, so you can install it via:

pip install HardView

You can find documentation and source code on GitHub:
https://github.com/gafoo173/HardView

If you encounter any issues, feel free to contact me or open an issue on GitHub - I’ll do my best to fix it as soon as possible.

[sullen mason]

Neat idea, I'll check this out. Everything I use for this currently is commercial.

[tawdry needle]

At first glance this looks interesting. As with any new tool, I think it would be informative for potential users to have a comparison with Sphinx and maybe also Pydoc.

"Modern replacement for Sphinx" implies that you're aiming for something like feature parity

Which seems ambitious

I definitely think the Python world could use a modernized documentation framework, Pydoc is fine for quick and simple but it's not extensible and customizable the way Sphinx is

I will say I do like the extensibility and structure of rST even if it is also infuriating at times

Going back to Markdown sometimes feels like a downgrade, especially if you are doing a lot of cross referencing of classes, methods, etc.

Ergonomic cross referencing and rendering of complicated type hints that doesn't obstruct readability

Those are my 2 big desired features. Maybe also a saner modern CLI than what Sphinx has

[mint cipher]

I was told that I can get Github Copilot to review my Pull Requests. I just wanna compare it with my experience to CodeRabbit. How do I get Github Copilot to review my PR?

[brazen forge]

https://docs.github.com/en/copilot/how-tos/use-copilot-agents/request-a-code-review/configure-automatic-review

[mint cipher]

Oh. Its only for Pro and Pro+ Plans

Welp

[iron folio]

Hi.
I am having networking issues with rootless podman combined with podman-compose.
Basically, when I run a single container, I am able to extract the incoming http request real client IP address (even without any network config - which, I assume, means that the passt/pasta is working).
As soon as I do anything with podman-compose (composing all containers inside a pod), I only get 10.89.0.X IP addresses.
Feels like I have tried everything - tried different network modes for pod --pod-args="--network=...", setting all possible drivers (including host and bridge) for networks inside compose.yml and more... Still nothing unless I only run a single container outside a pod (and without podman-compose).

Any ideas? Im not so good at networking, so simplest explanation is appreciated.

My versions:

podman-compose version 1.5.0
podman version 5.4.1

[rustic epoch]

Ah, yeah. Sphinx has a pretty wide surface area. Rather than aiming for 1:1 parity with Sphinx, I think we want to understand the most important things people need to do with Python documentation, and do those really well

[rustic epoch]

@tawdry needle OOC, what Sphinx extensions do you like to use? Do you ever write extensions yourself?

[rustic epoch]

Luma's current cross-reference syntax is [ClassName](package.module.ClassName). I think we want to simplify this to just [package.module.ClassName] or even [ClassName], but I need to do more investigation on how to cleanly implement this because it's not standard Markdown

[marble steeple]

does someone wanna be my tester for a face tracker in python

[tulip wyvern]

I can help you.

[marble steeple]

Imma dm u later i have class rn

[molten brook]

Good stuff!

[stark hamlet]

maybe something like mkdocs but python centric?

[buoyant tundra]

what tool you guys use mostly?

[full dragon]

anyone got a working macro ? the records keystrokes mouse clicks and mouse movements ?

[crimson spruce]

uv question: How do I add non-Python files (as in importlib.resources) to the package? I don't see anything regarding this in the linked pyproject.toml docs.

[deep estuary]

hmm, this? https://docs.astral.sh/uv/reference/settings/#build-backend_source-include

assuming you're using the uv build backend

[crimson spruce]

Nice, thanks. Not using any build backend at the moment with the project, other than git pull; systemctl restart ... 😅

[deep estuary]

the way god wanted

but yea i imagine it'd be that or whatever the actual build-backend you are using to generate package build artifacts has as an option

like hatch has a similar parameter https://hatch.pypa.io/1.13/config/build/#patterns ```yaml
[tool.hatch.build.targets.sdist]
include = [
"pkg/.py",
"/tests",
]
exclude = [
".json",
"pkg/_compat.py",
]

just nothing standardised for pyproject

[wide sentinel]

https://github.com/Abdullah-js/colorchangerPDF

hey guys if u like a pdf color changer on your enviroment make sure to check out my project and give it a look :)

[latent mantle]

Got it, thanks

[tawdry needle]

No I am not anything close to a power user myself. I think there are a few extensions I use whenever I set up a project, but it's been a while and I don't remember. I do however remember that with Pydoc by comparison I felt like I had very little control and could not really get the output I wanted

[tawdry needle]

I don't think I ever tried mkdocs but it seems popular! So maybe

[tawdry needle]

Have you considered a uri like ref:ClassName? and then the rule is that any link target which is not obviously a path (contains /) or a uri/url, is assumed to be a reference

You could probably adopt a rST-like roles system even

Although I think it's silly to actually use a different keyword to reference different types of things

[paper moss]

why im receiving this and how can I fix it?

ERROR: failed to solve: public.ecr.aws/lambda/python:3.12: failed to resolve source metadata for public.ecr.aws/lambda/python:3.12: unexpected status from HEAD request to https://public.ecr.aws/v2/lambda/python/manifests/3.12: 500 Internal Server Error

[crimson spruce]

AWS has some kind of outage right now

[paper moss]

thanks, https://gallery.ecr.aws/docker/library/python/?page=1
i can confirm it, this does not work

[astral apex]

They’ve just updated their status page to say they’ve fixed everything
https://health.aws.amazon.com/health/status

[vagrant whale]

nice

[paper moss]

Thanks

[molten brook]

Hi everyone, I published a few hours ago a new release of psp command line scaffolding tool: https://github.com/MatteoGuadrini/psp/releases/tag/v0.3.0

[crimson spruce]

I'm trying to set up Github Actions for running tests, and I'm confused by a ModuleNotFoundError I'm getting. I realize I haven't installed the package I'm trying to import, but that shouldn't matter if I'm in the right directory. Why does it not find the module, despite a package with the same name existing in the repo root? Does actions/checkout@v5 not mean my working dir is now at the checkout?

[deep estuary]

hello

python discord devoops technical support here

[deep estuary]

well, hang on

why are you changing directory

[crimson spruce]

I am not in that run; that was an experiment

There, gone

[deep estuary]

hmmm

 ============================= test session starts ==============================
platform linux -- Python 3.13.8, pytest-8.4.2, pluggy-1.6.0
rootdir: /home/runner/work/veronique/veronique
collected 0 items / 2 errors

i wonder if that rootdir is the actual repo or it's somehow decided to boot inside the module

[crimson spruce]

I think it's the actual repo, because if you're inside the module it wouldn't find any tests and exit with status 5.

[deep estuary]

so

[crimson spruce]

(...I realise the proper way would be to actually build a package and pip-install it. But I want to understand why this doesn't work.)

[deep estuary]

interestingly

when I clone your repo and run pytest

I get the same error lol

does pytest work locally for you?

[crimson spruce]

...huh, interesting.

Yes

Okay, but thanks, that helps.

[deep estuary]

interestingly though ```
❯ pytest
======================================================================================================= test session starts =======================================================================================================
platform darwin -- Python 3.14.0, pytest-8.4.2, pluggy-1.6.0
rootdir: /Users/joe/scratch
configfile: pyproject.toml
collected 0 items / 2 errors

that rootdir is not where i cloned it

it's the parent

and if I create a pyproject.toml I get a new rootdir (but still for some reason get the ModuleNotFound errors)

❯ touch pyproject.toml
❯ pytest
======================================================================================================= test session starts =======================================================================================================
platform darwin -- Python 3.14.0, pytest-8.4.2, pluggy-1.6.0
rootdir: /Users/joe/scratch/veronique
configfile: pyproject.toml
collected 0 items / 2 errors

[crimson spruce]

The error doesn't occur with a fresh clone for me either.

[deep estuary]

🤣

[crimson spruce]

(Although I make and activate a venv first and install sanic and pytest in the venv)

[deep estuary]

that's what I've done

[crimson spruce]

Let me try Python 3.14 or .13...

[deep estuary]

I get the same error on 3.13 and 3.14 with a venv for both

[crimson spruce]

Nope, still works..

[deep estuary]

hahahaha

okay

it gets even funnier lol

if I create test/__init__.py

it works

🤣

[crimson spruce]

o_O

[deep estuary]

ANOTHER SUCCESSFUL SOLVE BY THE PYTHON DISCORD DEVOPS TEAM!

[crimson spruce]

:(

You're not even on a different OS..

What pytest version do you have?

Oh I see it. Same one..

[deep estuary]

🤣

DETERMINISTIC BUILDS !

I suspect the explanation is somewhere on here https://docs.pytest.org/en/7.1.x/explanation/pythonpath.html

without that __init__.py it's looking for veronique in sys.path but without that __init__.py it's looking for /Users/joe/scratch/veronique/test/veronique/

[crimson spruce]

Running pytest with pytest [...] instead of python -m pytest [...] yields nearly equivalent behaviour, except that the latter will add the current directory to sys.path, which is standard python behavior.

[deep estuary]

aha

[crimson spruce]

Somehow my Python setup locally makes it so that when I type pytest it still chooses the second option?

Yeah, once I changed pytest to python -m pytest in the workflow it worked. But I don't understand why it works locally then..

[deep estuary]

yea that is odd

i guess it's probably good that it's a local thing rather than something funky with the CI lol

[crimson spruce]

How did you install Python?

(I'm confused because it also works on my work laptop running Ubuntu.)

[deep estuary]

homebrew

just using system python because it is patched to use homebrew native libraries

so brew install python@3.13 etc.

[crimson spruce]

Okay, at least there's a difference, mine was via pyenv.

[astral apex]

_Spent half a h*ckin' hour on that once _

[rustic epoch]

Yeah, something like that!

We're not trying to build a generic static documentation generator. We're building a tool for Python and it's unique nuances (e.g., handling of type annotations, overloads, etc.)

[rustic epoch]

@tawdry needle You mean a URI without any special syntax? For example, something like this?

For more information, see the ref:MyClass reference.

[rustic epoch]

Yeah, I'm not sure what the original motivation for rST roles were. I don't think you can have a scenario where the same name refers to two distinct Python objects in the same scope. So, I think you could just infer the object type

[heavy knot]

i hvae the fix

[crimson spruce]

What do you mean? Do you have a full explanation?

[heavy knot]

yes easy peasy

[frank fractal]

“Trust me bro”

[remote glen]

heh - that's been used in a couple of our flaky sprint demos lately

[honest terrace]

hmm I am having a nonsensical issue with AWX and the python request library if anyone else has run into this.
Testing from my machine, both using the builtin uri module and ansible or using my own similar module this API call I have works fine. But when running from ansible tower I get the Error 36 filename too long.
Which doesnt make any sense as to why but is there some kind of way to get the request module to ignore the length limit? Because the uri is not actually 'too long' (like <120 characters entirely) so I am trying to figure out if this is masking some other error or if its just a false error. I dont have access to Tower to check/modify anything

edit* 'my machine' refers to Ubuntu in WSL, running the same version of python and ansible as Tower is.

[shut prawn]

try doing python3 filename.py instead of python filename.py

[honest terrace]

Hmm well the module should be using python3 already as part of the sensible install but I will see if I can force my own module to try and use python 3

[shut prawn]

yeah imagine if it's using python 2, that would be bad

or just pass, that error won't affect the code

[neon shard]

this seems like an os limit rather than anything specific to requests

it's about file names and not uris

maybe you have some prefix or whatever that gets added, and on the other machine it's long enough to bump some path above (iirc) 255

[honest terrace]

Kinda agree TBH

[honest terrace]

This is what doesn't make sense to me as the only literal difference is the platform it runs on. Same exact playbook and uri. Same version of ansible so my only assumption is that tower has some silly file path length limit set really small. Because I run mine on Ubuntu and I think tower is on suse but even then the uri is only like 130 characters, it's a post method and all the data is json in the body so I'm not appending any just to it either

[neon shard]

as said, I suspect it's not related to the url

do you have some stacktrace?

[fading karma]

Hello, so I was just wondering and was very curious and got an project idea but it seems so complicated so I decided to ask here how someone would go by making a project as such. Basically, If someone would like to scrape the whole web and use discord to search for a certain keyword for example if they are having issues with programming something, they just use a discord bot and type their search query and the scraper tool will automatically give the best answer to solve that issue. How would someone go by making something like this and do you think it would be a good project to learn from? Just curious btw

[neon shard]

what you are asking about is a search engine

[honest terrace]

Yea I figured it out. There was some stupid flag on ansible tower that would try and cache the request result in a file for some reason. And because it was some separate task and script it was obfuscating what was actually throwing the error.

[fading karma]

I figured lol

How difficult would it be to make a search engine if i may be interested one day?

[errant knot]

Flex fr ;-;

[cedar gorge]

Not too hard to make a very basic implementation! You might find it interesting to look into the evolution of Google's scoring criteria over time.

[neon shard]

https://en.wikipedia.org/wiki/PageRank is a classic

[fading karma]

Alright, thank you for the clarification and thank you @neon shard for the useful follow up.

[inner pollen]

Implemented a tool, if you are coming from some other language or want to see if a certain thing is available in python and in which module you can get it form here

It does a similarity search and tries to find similar keywords inside the language

which semantic search algo should i use

ParseInt should give int with 100% similarity

[shadow crow]

That doesn't sound like something easily automatable

(how do you automatically link e.g. f64 with float and Vec with list?)

maybe with some kind of LLM

[echo grail]

e-graphs? :p

[blissful bluff]

Hi everyone, I am using pyproject.toml with tox, but tox is not installing the dependancies from the project, only when I add the deps entry under the tox section i got them installed, i find that annoying as I need to duplicate those dependancies, btw I am not packaging I am doing a library, anyone knows how to tell tox to install the pyproject dependancies ?

[gentle solstice]

Are you using dependency_groups?

[blissful bluff]

@gentle solstice yeah but I use it just for dev dependancies, and I want install the app dependancies too

basically if I don't add the [build-system] it won't install the project normal dependencies

[gentle solstice]

What if you add a dependency-group main?

[gentle solstice]

Any reason you're not specifying a build backend?

[velvet spire]

Why don't you have a build system?

[blissful bluff]

I am not building a distributal package, just an app

[blissful bluff]

how do you do that ?

[gentle solstice]

You said you have dev dependencies, right?

[gentle solstice]

Set one anyway.
https://tox.wiki/en/4.32.0/user_guide.html#packaging

Otherwise, use the [dependency-groups] table instead of [project] and set the dependency_groups array in your env_run_base table.

See here: https://tox.wiki/en/4.32.0/config.html#dependency_groups

If you're using a project manager like uv, it might support installing the "main" group by default. https://docs.astral.sh/uv/concepts/projects/dependencies/#default-groups

Though I recommend using a build system

[blissful bluff]

yeah I ended with the build system

[blissful bluff]

you mean out everything from dependencies into the dependency-group:

dependencies = [
    "requests == 2.32.4",
    "dataclass_wizard == 0.35.0",
    "boto3 == 1.38.33",
    "python-dotenv == 1.1.0",
    "pydantic_settings == 2.9.1",
    "troposphere == 4.9.4",
]

[dependency-groups]
test = [
    "pytest>=8.4.2",
    "pytest-cov>=7.0.0",
    "tox>=4.31.0",
    "teamcity-messages",
    "python-dotenv == 1.1.0"
]
lint = [
    "ruff>=0.1.0",
    "mypy>=1.0.0",
]
dev = [
    {include-group = "test"},
    {include-group = "lint"},
]

[gentle solstice]

Yes, as its own group

You already used build-system, so I'd keep that

[blissful bluff]

yeah, ok that's what I am doing

[astral apex]

TIL that docker run rm does not automatically remove failed containers

And now my disk is full again

This will prune stopped containers, right?

including the failed ones this time

[brazen forge]

you can add volumes to that list too

[deep estuary]

i'd be slightly more wary of mass purging volumes automatically

if you're accruing unaccounted volumes en-masse something else is probably wrong with your setup

both from the perspective of it's risky to tear them down, but it's also not great that there isn't a better lifecycle and they're just hanging around until you do something with them

[brazen forge]

I'd say the same about images and containers tbh

[deep estuary]

yeah -- this is true

i'd just be slightly wary of wiping out stateful volumes vs. immutable images

[velvet spire]

with the exception of dev environments

I tend to just run docker rm if I've contributed to too many projects with docker containers at a time

looking at pydis bot

[deep estuary]

this isn't a dev environment

it's managed by ansible

[velvet spire]

cursed.

[deep estuary]

for local dev you don't need to do pruning of individual stuff you can just do docker system prune --all --volumes

[deep estuary]

it's the best tool for easy and intuitive stateful app deployments

because it 1-to-1 maps with what you'd do administering via a shell, with some shortcuts for being more programmatic about workflows

[sudden igloo]

I love ansible

[covert dust]

Hi everyone, I am looking for a online GPU providers in US that I can rent 8*B200 instances
Anybody can let me know some good options for it?

[long fog]

Hi people, do you have some good recs which can help reduce runtime?

[narrow glacier]

Hey everyone, do anyone working or have done some daemon designing for Linux based environment?

[desert skiff]

ask your question instead, this really goes for this whole server (and not just the #1035199133436354600 section)

[humble cradle]

how can i fix lagging vs code

[vague silo]

probably by using another editor. vs code is not exactly known for good performance

[crimson spruce]

!warn 1039032964845142027 Please don't advertise

[rancid schooner]

:ok_hand: applied warning to @lavish saffron.

[lavish saffron]

Isn't this channel the place to discuss tools?

[crimson spruce]

It is. It's not the place to advertise tools.

[lavish saffron]

ok

[lavish saffron]

Can I talk about my project in the off-topic channel?

[crimson spruce]

You cannot advertise it anywhere. You can talk about it, even here.

For further clarification feel free to contact @balmy mulch

[lavish saffron]

ok

[brazen forge]

is this new?
github.com/org/repo / Settings / General

[astral apex]

Since August
https://github.blog/changelog/2025-08-26-releases-now-support-immutability-in-public-preview/
Hit GA a couple weeks ago
https://github.blog/changelog/2025-10-28-immutable-releases-are-now-generally-available/

[brazen forge]

I see

[dusty forum]

New release 0.13.0 (lucky!) of logmerger (textual based TUI for viewing multiple log files with a unified timeline) comes with 10X data loading speedup, detailed loading progress bar panel, and "info" command to summarize the loaded log files, number of log entries, and covered time range. Use pipx install logmerger to access it as a terminal command anywhere on your system. Repo at https://github.com/ptmcg/logmerger.git

• time-based side-by-side or inline view
• includes some nice navigation features (jump forward/backward by number of lines or time interval, go to a line number or time stamp, and some simple searching)
• runs in the terminal (so even over SSH, without having to set up port forwarding, which I never can remember how to do)
• supports merging of .log with .pcap files
• auto-unpacks from .log.gz or .zip files
• textual-enabled vertical and horizontal scrolling (to view logs that are wider than your terminal screen)
  #textual #logmerger #python #tui

[radiant cloud]

I need to plot a tree and then rotate it 180 degrees. The nodes should be defined with rectangles, like in the image, but they should be stacked on top of each other, and I don't know how. Could someone please help me?

You should be able to compare the leaves of each tree with each other

[mystic void]

what do you have so far and what is blocking you?

[distant river]

Im actually working on something like this. It takes any domain type of dataset transforms it into leaves and builds a merkle tree up to a single root hash.

[distant river]

Whats the question?

[distant river]

What about a sentance transformer?

[inner pollen]

Okay, let me explore, but it must have a c/cpp interface.

[hoary anchor]

I hope this is the right channel to ask:
Can someonep explain to me please, what the colors mean in Pycharm in the "Git Tab"? I added a branch, added a feature and delted the branch afterwards. Are the colors for different branches? Once I saw that the branch "forked" or branched off to another line, when does it happen and what is the difference?

[molten brook]

Hi to everyone, I published psp to Pypi: https://pypi.org/project/psp-scaffold/
pip install psp-scaffold

[sudden igloo]

Not sure if this is the right channel to ask, but if doing continuous integration and someone messes up on the main repository and wants to get rid of a commit, like you forgot to test locally and the tests failed on integration, is it normal to basically end up rewriting the history to get rid of it from main? Or... Are you supposed to be developing on a separate short lived branch, merging it after your tests pass, or removing it from that one if they fail.

Because I've been in the situation several times where I've wanted to remove a commit from main and jetbrains would not let me force push and I had to go through the git command and get credentials and so on (it considers it a protected branch or something), I think there's a setting I can change to fix it, but I don't want to walk into/continue an anti-pattern.

[desert skiff]

you normally develop and test in a branch and then make a PR to merge into main when it's ready for production
these things still happen, and if you end up with bad code in main you would typically just use git revert to an earlier commit and push that again to get back to a previous version of the code (but as a new commit), the bad code will still be in the history of the repo, but that shouldn't be a problem and is the recommended way of fixing such situations in a shared repo (or whenever the code has left your local git repo by pushing/publishing it)
the only situation that i would recommend to really go in and rewrite history in a shared repo like that (github/gitlab/gitea or whatever you are using) would be if you accidentality pushed any sensitive/secret data to the repo that you don't want to be left in the history of the repo, but that comes with side effects to anyone else that has checked out and have copy of the repo
a more detailed explanation of that and the consequences can be found at https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/removing-sensitive-data-from-a-repository

[sudden igloo]

That makes sense. I misunderstood "don't branch" (though I do understand local development is technically like its own branch to begin with, and short lived branches are fine) and "the code should always be in a releasable state". So it's fine for it to be red at points in its history as long as the latest is returned to a working state quickly, and "reverting" is just adding a new commit to the history with the bad changes removed. If something sensitive is there and it has to be scrubbed history has to be rewritten and if there's been important commits since all the hashes people potentially rely on are about to become invalid and coordination required so that no one pushes those old changes again and or loses work.

I think the only times I rewrote it was when I did accidentally put somewhat sensitive information in a readme. I'm confused by the first part though, if I'm doing continuous integration I'd want to be working on the main branch and everyone integrating as soon as possible without waiting on review (just automated tests), did you mean in general or like one that you approve yourself or like an automated one that gets approved if the tests pass? (I have seen short lived feature branches are a thing). Thanks for taking the time to explain, and providing that resource. Right now for context I'm not working on any features explicitly, or planning far at all, I'm just taking the code one step at a time, exploring the problem, getting stuff further, and commiting and integrating frequently seeing the tests still passing on the repository.

[deep estuary]

it's worth noting that generally even post-rewriting a commit still exists in a repo

even post rewriting you're just changing what appeared on the branch at what time, the original commit that you've removed will still be in the reflog and is still completely accessible, so if you're truly looking to scrub secrets from a git repo deleting it is generally the only truly reliable option

there are things like https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/removing-sensitive-data-from-a-repository to try help remove stuff more permanently, but generally a simple rebase with a drop or something won't actually remove what happened in that commit from other places that git stores commits

for example if I have this commit history:

commit 05876e55b6d01768f763190d43aacd1203b24aa4 (HEAD -> main)
Author: Joe Banks <[e-mail]>
Date:   Fri Nov 21 14:47:18 2025 +0000

    Add another file

commit a185ebe6d51605dbf0f46385c4852ecc73dd6e80
Author: Joe Banks <[e-mail]>
Date:   Fri Nov 21 14:47:04 2025 +0000

    Some bad commit I added

commit 52bdd02776ea48fb47051dc26852f46936403837
Author: Joe Banks <[e-mail]>
Date:   Fri Nov 21 14:46:39 2025 +0000

    Initial commit

and i use git rebase HEAD~2 and then drop the bad commit you can see it's still in the reflog

even though it's no longer in the log

and thus if I run git diff a185ebe~1 a185ebe (get the diff between whatever was before a185ebe and then a185ebe), you can still see the changes, even though that commit was dropped from the branch

a lot of this is only locally cached but github itself does also have caches that still allow you to sometimes see things that got force pushed away

you can expire the reflog etc. to fix local stuff but there's often ways that things end up on github that are accessible (like that time someone added yt-dl to the github dmca repo)

[desert skiff]

wasn't that exactly the page i already pointed to?
and yes, it's a bit more involved to truly scrub sensitive data from a repo reliably then "just" rewriting the history (which is definitely not enough)

[deep estuary]

sorry that's on me — I read the last bit of the above message, hadn't quite started the day yet

[desert skiff]

ah, early morrnings ☕ 🙂

[sudden igloo]

What kind of caches does GitHub have that could expose this?

[swift cradle]

hello guys

[desert skiff]

!rule ad

[rancid schooner]

Rules

6. Do not post unapproved advertising.

[desert skiff]

this is not the place of this sort of thing

[cedar moon]

.

[quartz haven]

Their is 4 bugs

And logic flaws

[swift cradle]

Ok can you explain

[crimson spruce]

Please delete this.

[swift cradle]

👍

[crimson spruce]

Same goes for you, advertising is not allowed here.

[crimson spruce]

@coral scroll I've deleted your message. See above.

[coral scroll]

?

ah.. ok.

where can i ask for testers?

[crimson spruce]

Nowhere

[coral scroll]

hmmm 🙁

[wraith dock]

Does anyone know of an automated check to flag code duplication across code base?

[short peak]

I know sonarqube does that. We have it implemented as a GH action, I only use ruff and ty locally, not sure if ruff provides that features though.

[short peak]

also, seems like pylint can do that too - https://stackoverflow.com/questions/60620262/how-to-detect-code-duplication-among-multiple-python-source-files

[wraith dock]

What benefits does ty provide

[wraith dock]

So u need to explicitly enable it I see

[gentle solstice]

Rust

It also has incomplete features and more bugs.

[shadow crow]

Ty, as planned, has a more rich type system than mypy/pyright. For example it's going to support Intersection and Not

[willow pagoda]

First time publishing docker images, what configuration causes these un/tagged sha256-* versions to appear? They're a bit noisy so I'd like to skip uploading them, but I couldn't find a relevant option in metadata-action or build-push-action:
https://github.com/thegamecracks/ministatus/pkgs/container/ministatus/versions

my workflow file: https://github.com/thegamecracks/ministatus/blob/main/.github/workflows/docker.yml
last runs: https://github.com/thegamecracks/ministatus/actions/workflows/docker.yml

[crimson spruce]

How? from ty import Intersection? Or are they lobbying for stdlib inclusion?

[shadow crow]

from ty_extensions import Intersection, yes

https://github.com/astral-sh/ruff/blob/c534bfaf01671b8020edd440bc7f4f22070af583/crates/ty_python_semantic/resources/mdtest/intersection_types.md?

[astral apex]

Click on one and scroll down to past the "No description provided" to see the "Manifest"
"mediaType": "application/vnd.dev.sigstore.bundle.v0.3+json",

Sigstore publishes to the index

Try docker pulling one of them
You'll get an error saying application/vnd.dev.sigstore.bundle.v0.3+json isn't a valid mediatype

[willow pagoda]

ahh that makes sense, maybe i'll leave those be then

[lethal pond]

is there another source code hosting platform like GitHub that offers free CI and a container registry? I don't need a lot of it. Just one action that publishes one docker image every month or so?

I'm looking to move a hobby project off of GitHub

okay, I tried gitea and that does not offer a similar service

[thorny shell]

I'd check gitlab

[lethal pond]

okay, thank you ill try that too

[turbid dirge]

Gitlab and Codeberg are the names I keep hearing, but I don't know in detail what either offers.

[thorny shell]

me neither

[neon shard]

for fun: https://tangled.org/

[thorny shell]

I am put off by any tool that introduces a ton of new jargon: "knots", "spindles", "Jujitsu change IDs"

also it's unclear what problem it's solving

[neon shard]

knot/spindle and the like is just cutesy themed naming for their stuff

[thorny shell]

uh huh

[neon shard]

change ids are ids that are stable over the evolution of a commit

[thorny shell]

The tool might be great. I'm just saying those terms are ... let's say pink flags.

[glad geyser]

Hi

[neon shard]

e.g. if you amend (or even change the description) a commit in git you would have an entirely new identifier (the hash)

[thorny shell]

Oh like http://esr.ibiblio.org/?p=3872

[neon shard]

jj has started popularizing them recently, and different services have started accomodating them

(at the very least not destroying them during upload/rebasing/whatnot)

gerrit has had this concept for a long time as well

https://gerrit-review.googlesource.com/Documentation/user-changeid.html

[thorny shell]

aw man it's just a guid

[neon shard]

it's really just some identifier

the one jj is popularizing is an identifier stored as k-z hex to easily disambiguate it from the usual git hashes and similar

but ultimately it tends to just be a randomly generated identifier you can use to refer to the change

but one that is stable over the evolution of a commit

[neon hull]

what's the quickest way to test a script on multiple python versions

pyenv local version switch? uv?

[sudden igloo]

Write automated tests and run them with a tox file setup to do them for different versions you have installed, maybe use UV to easily install the ones you want.

[jagged yoke]

I use GitLab and quite like it, but there's a learning curve for the new config you'd have to learn. It's not hard (except in the "we have made YAML almost, but not quite, Turing complete" sense). Free CI if your project is a public open source project, and there's also some free allowance of CI minutes per month for personal projects... I think?

[jagged yoke]

I've always preferred GitLab to Github, but that's just a personal preference. Nothing objective.

Unrelated: is there a formatter, even a does-this-one-thing-only type of tool, that can format long strings in code? I usually use black instead of ruff because of this, but it looks like black is reconsidering some of their string processing stuff.

[jagged yoke]

Which is fine! I'm not complaining about a preview/unstable feature going away, in an open source project. But it might be a good time to look for another tool to do this, if one exists.

[neon hull]

uv switches back to the same version again and again

it changes the env version for the moment

[willow palm]

i've been using nox with uv backend, i think nox can pull python versions itself now too with some extra dep

[gentle solstice]

Tox/nox in general

[urban jewel]

just or ahoy?

I have used ahoy in the past, is as simple as declaring a yaml.

what is the simpler/most used?

[thorny shell]

never heard of ahoy

I have grown fond of just, fwiw

you might be able to answer "which is most used" by (somehow) querying github. I don't know how to do that, but I assume it's possible

[urban jewel]

well

it's safe to say that just is the industry standard lol

thanks

[astral apex]

ahoy?
Can I get a link?

[urban jewel]

https://github.com/ahoy-cli/ahoy

I do prefer the yaml syntax than the makefile one

[astral apex]

.ahoy.yml

[urban jewel]

yep

[astral apex]

https://ruudvanasseldonk.com/2023/01/11/the-yaml-document-from-hell

I don’t want this happening to my CLI tools

[urban jewel]

how about docker compose?

I use it too much, so use YAML everywhere is better then multiple file formats

if yaml sucks so much, why is it used everywhere? like docker, k8s, etc

genuine question

[thorny shell]

dunno. Maybe everyone really wants to use JSON, but use YAML instead because it allows comments 🙂

[brazen forge]

nobody wants to handwrite JSON

[quasi storm]

Subject: Guidance on Setting Up Error Alerts

Hi,
Not sure if this is the best place to ask—feel free to redirect me if not.

I currently have Grafana + Loki logging all backend errors, but I don’t get any real-time alert when something goes wrong. Usually I only find out when someone reports the issue. I was considering adding a Teams webhook, but I’m worried about spamming the channel with too many notifications.

Looking for suggestions on how people normally set up alerting rules—especially around when to actually send alerts vs. when to just log errors.

Thanks

[thorny shell]

off the top of my head:

• have your Grafana or Loki or one of those guys scan each log message, and send an alert if and only if the text of the message matches some pattern that you know indicates something Bad®
• have Grafana or Loki count the number of errors in (say) the last 60 seconds, and send an alert if it exceeds some value

disclaimer: I have never found a reliable way to do this; there's always either too many alerts or not enough 🤷

[analog kettle]

so you can't use docker cp to copy between two containers, but if I understand correctly, you could just pipe two docker cp calls and let the tar flow from one to the other

in which case, why not just let docker cp natively do that?

[thorny shell]

maybe doing "cp" properly involves subtleties like permissions that they haven't figured out

[brazen forge]

IME, alerting on error logs can get quite noisy and unruly

normally you'd emit metrics which can be scraped by/written to a metric store like Prometheus/Graphite/InfluxDB/etc.
then configuring alert rules in Grafana/Alertmanager/etc. using those metrics is straightforward

@quasi storm, pinging since I accidentally turned that off when replying

[quasi storm]

Haha thanks , actually helped because of the ping

[gentle solstice]

I'm looking into self-hosted error reporting and I'm deciding between bugsink or glitchtip. Has anyone used either and would you recommend it? What other self-hostable suite have you used? Working completely offline (aside from a docker registry) is one of my requirements.

[crimson spruce]

I'm guessing you have already considered (and rejected) Sentry?

[gentle solstice]

I didn't realize I could self-host it.

Does sentry have a helm chart?

[crimson spruce]

I have no idea, it has a docker-compose file. It turned into a behemoth though, I think there's like >50 containers involved now?

We are in the process of updating from an ancient version of Sentry to a modern one, and I hear it's a bit of a headache.

[gentle solstice]

Not sure how I feel about it needing 16GB RAM+swap

• 4 CPU Cores
• 16 GB RAM + 16 GB swap
• 20 GB Free Disk Space

[crimson spruce]

see https://develop.sentry.dev/self-hosted/data-flow/

[gentle solstice]

Upside: bugsink is compatible with sentry apis

[crimson spruce]

As is glitchtip, apparently

[gentle solstice]

sentry is just the standard, isn't it

[ivory fjord]

Hello guys
New here

[vague silo]

Hello Sani

[deep estuary]

Hello Sani

[inner pollen]

Hey everyone! Looking for a review of my open-source MCP platform, please review it once. It’s for internal teams managing MCP servers across the company.
Currently easy to host on bare-metal; I’ve only tested on a single-node K3s setup. Can it run well on a multi-node bare-metal cluster? I’m about a month into Kubernetes, exploring internals.
It will act as an internal registry for your company. The platform is capable of hosting those mcp server, considering the scale is supported by the
cluster on which the platform is running, I am shipping a CLI to manage the platform, which includes UI as well, if users find it useful
I’ve used operators and CRDs, and I’m considering an admission webhook for some tasks. Please share any issues you see and suggestions to
improve it.
https://github.com/Agent-Hellboy/mcp-runtime

[neon shard]

!otn a let the tar flow

[dawn hill]

Hi! Hope a little bit of self-promotion is ok. I've just open-sourced a tool we've used internally to automate and codify development tasks (e.g., cleaning projects, syncing development databases, etc.). https://github.com/petersuttondev/cleek

Cleek is a super-simple task runner that automatically generates CLI's using your function type-hints. Create a cleeks.py file in you project:

from cleek import task

@task
def greet(name: str = 'World') -> None:
    print(f'Hello, {name}!')

$ pip install cleek
$ clk
┏━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Task  ┃ Usage                    ┃
┡━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ greet │ clk greet [-h] [-n NAME] │
└───────┴──────────────────────────┘
$ clk greet -h
usage: clk greet [-h] [-n NAME]

options:
  -h, --help       show this help message and exit
  -n, --name NAME  default: World
$ clk greet -n Discord
Hello, Discord!

[gentle solstice]

Were you inspired by click?

[dawn hill]

Ner, we used fabric for ages. If I remember right fabric didn't allow types, but we wanted type-safe code. So I made cleek. The main aims are 1) type-safety 2) shouldn't alter code so I can be used elsewhere (@task just returns greet() unchanged so I can be used like another other function) 3) low effort (we were already writting the type annotations).

[crimson spruce]

Actually reminds me more of fire

[dawn hill]

Never seen fire before. It's half way between click and fire but using type-hints

[fiery marsh]

guys

can u help me check this code

someone

made me run it on termux

[mystic void]

if you don't share the code, no one can check

[dusty stump]

does anyone in here have experience with spiderfoot? im trying to manually edit the use case section (Footprint mode) to not run a specific module displayed in the modules directory for spiderfoot, the module is sfp_tldsearch.py, ive tried editing the file name to other names, editing the file contents, or moving it completley but i just get hit with a "Broken Pipe" error whenever i try and run a scan, anyone got any tips?

[dusty stump]

nevermind, i fixed this problem by editing line 29 from 'useCases': ["Footprint"], to 'useCases': [""],

[shut fulcrum]

Weird obscure tool: Connect:Direct.

[thorny shell]

TIL

[half grove]

Hello guys! I do not know this a right place to talk about my project.

I built a package Kaizo for automation and dynamic configuration execution. I used it personally for running and training ai models on servers. It is fully dynamic and pluggable. I wrote a hugging-face plugin to upload/download files onto hf servers.
You can write additional plugins for it for your usecases.

The full documentation is available here: https://naughtfound.github.io/kaizo
Repo: https://github.com/NaughtFound/kaizo

You can install it using pip.
Thanks.

[wind solstice]

Hey guys , can someone explain Devops vs Site Reliability Engineer vs Cloud platform engineer .

[warm gazelle]

I don't really know how to code python, but I brute forced a couple different chatbots to create a .exe file renaming program mostly to make my life easier on renaming folders with many pdfs, is there a place I can share my github link with the .py and my released .exe?

I struggled heavily through compiling it so I have the requirements.txt and .py because I'm betting the idea of running some unknown .exe isn't that great

But i'm not sure where/who I could share with to see if it works like I think on a different system

[grave jewel]

I love Python and long walks. Daydreaming about physics sim algorithms while completing a 32768-meter long section in my quest to connect every major Bay Area park and city on foot is sublime.

Walking every step between SF San Jose Antioch Felton and more so far has been one hell of an eye opener. Tents and RVs are everywhere! This is the highest homeless tech geek concentration in the ENTIRE WORLD, probably even edging out Seattle.

How can the homeless keep their files safe? Oops, their phone was stolen by thieves or police (can they even afford another phone, not so easily if they are saving up their wages for an actual apartment).

What about a WASM virtual machine that includes a compiler? The idea is they log on to any library computer (the abundance of homeless in libraries doesn't bother me, but the staff constantly waking them up was sad and distracting) with a single username-password and have a virtual Linux in their browser that includes a WASM compiler and they can type in pip install pygame etc. No need for any personal property whatsoever to use! For some people, such as the founders of Jazz, creative projects keep you going through tough times.

Oh and it would also be useful on days I don't want to lug my computer around...

BUT how does such a project stack up?

1. Performace. Speed ratio vs native for CPU and GPU heavy tasks. How reliable such a ratio is, so that they have an idea of what native performance would be. How much GB of RAM you can actually access. Startup and server I/O latencies and download speeds.
2. Feasibility How many wheels need to be invented vs how much can it be build just by bolting together bog-standard parts?
3. Cost The servers have to store and deliver data. How many GB do people really need and how much does it cost? How many clients can one large cloud instance handle (or is at-home volunteer better)? It should be backed up. So the total $/person/month I don't know?

[hexed token]

is there a tool like node version manager fpr python

[thorny shell]

I'm not certain what "node version manager" is or does, but I'd look into uv

[sage basin]

I need advise on what projects to try out to learn devops? Azure or AWS, if you have recommendations on which is better that would be great. I currently only have experience with Django

[astral apex]

https://github.com/nvm-sh/nvm
It's the pyenv of JS landia

[shadow crow]

Despite having a decent (NVMe, no pun intended) disk, nvm is somehow the worst offender in my shell startup, at 210ms without cache and 140ms with cache

somehow pyenv is faster at 75ms

[astral apex]

Somehow?

Have some faith in Python

[quasi storm]

Hi guys

How do people safely deploy to a VM from GitHub Actions?
I have a VM where my containers run. On each push, I want GitHub Actions to trigger a build on the VM and run docker compose up.
One approach I’m considering is storing SSH credentials in GitHub Secrets and having the action SSH into the VM and run a deploy script.

Is this the usual/recommended pattern, or are there safer / cleaner alternatives (agent on VM, webhooks, etc.)?

[brazen forge]

why build on the VM instead of building an image in CI, pushing the image to a registry (like GHCR), and notifying the server to pull that image (with the digest ideally)

you could also use https://containrrr.dev/watchtower/ for the last part

[sudden igloo]

safe is situational and kind of subjective. You have to store an ssh key there if you want any form of server automation I'm pretty sure. Secrets and treating it like a highly privileged API key or anything else would be appropriate.

This specific approach doesn't sound usual imo but it seems relatively par for the course depending on what exactly you want to do. The weird part sounds like deploying with docker compose.

[quasi storm]

umm basically i build the image in vm and directly up the container from there

because the image is of around 2gbs

so thats why not keeping it in GHCR

[sudden igloo]

Is this VM only used/going to be used for this piece of software you're deploying to it?

[quasi storm]

more or less

its prod vm

[brazen forge]

you could run a small HTTP server on the VM that only exposes the minimal surface of accepting requests to build the container image and restart the service

then send authenticated requests to it from GitHub Actions

[deep estuary]

just fwiw, watchtower is now unmaintained

podman has this functionality built in though https://docs.podman.io/en/stable/markdown/podman-auto-update.1.html

[deep estuary]

the way I do it is with:

• a VM that is on a tailscale network
• a github actions workflow that requires an approval (to prevent forks and PRs from having too much power) or a workflow that can only deploy from main (for prod deploys)
• use github environments to make available an SSH key when a specific environment is being deployed to
• finally, i use SSH authorized key restriction to control what commands those CI SSH keys can run