tools-and-devops

[astral apex]

I don't know what that means

But.... I don't think so?

[south marsh]

I'm not sure what this section is trying to explain then

[astral apex]

That you can use a context manager to automatically remove the temp file when you exit it

[south marsh]

what temp file

I'm just trying to read the file as-is

[vapid hedge]

hi everyone is there anyone who have experience in maya Python Scripting or mel? i need some small help.

[obsidian moat]

whats up

[obsidian moat]

very nice

keep up

[misty elk]

I'm new to git and I'm trying to reset back to an old commti (just learned that revert doesnt work like that). When I reset things work but it's only locally- how would I push those to the main branch? I can't stage commit push as I have some conflicts and I don't think it works like that in general.

[rapid sparrow]

i hope you are single contributor to this branch

in this case you can always git push --force-with-lease

in order to overwrite history

if u aren't single contributor, don't do it. main/master branch is supposed to be not overwritten in its commits

git revert commit is only allowed in them with working in collaboration with others

u are allowed overwriting history in feature branches u own only (if working in collaboration)

[misty elk]

im the only contributor i only use this repo to backup my notes

"fatal: The current branch main has no upstream branch."

[rapid sparrow]

okay, then it is legit

[rapid sparrow]

write answer of git remote -v

[misty elk]

origin https://github.com/mygithub/repo.git (fetch)
origin https://github.com/mygithub/repo.git (push)

[rapid sparrow]

write answer of git status ? 🤔

[misty elk]

on branch main nothing to commit working tree clean

hmm

[rapid sparrow]

write answer of git status when u encounter your problem
or this problem "fatal: The current branch main has no upstream branch."

copy answers ^_^ git already highly likely told you how to fix this error

[misty elk]

fatal: The current branch main has no upstream branch.
To push the current branch and set the remote as upstream, use

    git push --set-upstream origin main

To have this happen automatically for branches without a tracking
upstream, see 'push.autoSetupRemote' in 'git help config'.

[rapid sparrow]

🤔 your git repository manages to tell that u don't have main branch yet

do you not have main branch remotely?

did u push repository to remote before?

what is default branch? is it master or main

also, how can u created this git repository 😄 was it created manually via git init, or you git cloned it from remote

[misty elk]

it's main

i made a repo in a folder then moved the .git files into my note taking app folder (obsidian)

cause i followed a tutorial

[rapid sparrow]

how did you made a repo in folder

[misty elk]

github desktop client

so same thing as if i did git init

[rapid sparrow]

So, what can u see if you photo your remote GUI?

https://github.com/mygithub/repo.git what this address shows in GUI? by default?

[misty elk]

it shows my repo

the main branch

with the bad commit i made

[rapid sparrow]

can u show more precisely 😄

[rapid sparrow]

do you wish to overwrite/delete it completely with new git repo state?

[misty elk]

nah i have the id of an old commit from yesterday. i want to just take that commit and make that my new main branch

so yeah i think thats what you mean

[rapid sparrow]

nah i have the id of an old commit from yesterday.
failed to comprehend this sentence

[misty elk]

I made a commit yesterday (which didnt have conflicts) and i want to revert all changes i made since that commit yesterday. so i essentially want to remove the current files and replace them with the ones from the commit yesterday

[rapid sparrow]

anyway, looks like u wish to overwrite remote completely with your new local state created via git init to remote
deleting remote completely and overwriting with your current local state, correct?

[misty elk]

yes

[rapid sparrow]

okay, then just use git push --force --set-upstream origin main

[misty elk]

error: failed to push some refs to 'https://github.com/stillmygithub/myreposname.git'

when i use that command

[rapid sparrow]

🤔 lazy to create repository replicating this situation

git push --delete origin/main ?

[misty elk]

ahh i get what you mean i dont have a repo with the old commit

[rapid sparrow]

yeah, u have different initial commit starting your tree

[misty elk]

thanks for the help man i got to go sleep or i will be dead tomorrow. hope i figure it out. have a good night!

[rapid sparrow]

u can always just delete your repository remotely and create new one with empty state (no license or readme! or anything)

then u will be able to push git push --set-upstream origin main

[zinc flint]

does anybody here know about using printers with python and could maybe help me? I need to be able to print a pdf (or postscript) from a printer and configure the settings for the print without admin permissions. Basically what word does when you wanna print a file. How can I do that? I have the win32print library but I can't figure out how to do it without admin perms since the only way i found it works is with "SetPrinter". I would really appreciate any help!! Sitting on this problem for a while already and can't find anything

[mint cipher]

Having trouble to deploy a very simple flask app to fly.io. Not sure how to fix. Error is gunicorn not found from bash and I don't understand why its missing or what I can do to install it on fly.io. Help!

[thin jasper]

cant give specific advice for fly.io. But looks like there is either no virtual environment with gunicorn installed and/or its not activated in your terminal session

you might run a pip freeze there to check if its there

[mint cipher]

I was able to get it running now. They allow us to define our Docker image and deploy the docker image instead

[wind cosmos]

hi! for create libraries what do you recommend? poetry, hatch, pdm?

[tawdry needle]

i strongly suggest specifying your dependencies in a requirements.in file and using pip-compile to generate a "locked" requirements file from that: https://pypi.org/project/pip-tools/

(note: you should run pip-compile inside the docker image to ensure that the platform and os specifics are reflected correctly)

[worldly hornet]

I've deployed several projects to Fly and any time I have tried to use their magic configurations for common stacks (like Gunicorn + Flask) I regret it. I think using your Dockerfile is better long-term anyways.

[worldly hornet]

If I currently use and understand Poetry, do you think I should check this out?

[tawdry needle]

no need, poetry will do it with poetry export

[tawdry needle]

i personally use hatch because i like it, but a lot of people use poetry and some people use pdm, they are all "good"

[worldly hornet]

Cool, I really need to dig into Poetry and learn more of the depth of the tool. Thank you for the nudge. I agree, pinning and locking dependencies is a good pattern, even for solo projects.

[wind cosmos]

thanks for your answer, i'll continue my research, but hatch has one vote 🙂

[tawdry needle]

i think poetry might have more features included

hatch feels a bit like a "WIP" project still

and some things like exporting a locked requirements.txt requires extra work, whereas it's just included in poetry

[wind cosmos]

ah ok, i've heard about poetry too

poetry is easier?

[tawdry needle]

it has more things built in, which might be equivalent to easier

[worldly hornet]

Poetry is what we use at our company. It works well, but I'm a super shallow user.

[wind cosmos]

i need to create a simple library for internal use so i need something to create that in very little time

i will research peotry

poetry*

thanks!!!

[thorny shell]

I use poetry all the time and recommend it, fwiw

[willow pagoda]

for the various libraries ive written, pip + setuptools is pretty easy to use especially with recent improvements like better dependency resolution and adoption of the pyproject.toml standard, though that's not to say i disapprove of the other options

for example, here's the minimum needed to make a package installable: ```toml

my_package/

init.py

...

pyproject.toml

[build-system]
requires = ["setuptools"]
build-backend = "setuptools.build_meta"

[project]
name = "my-package"
version = "1.0.0"
dependencies = ["click>=8.1.7", ...]

Installing for development: pip install --editable .``` see also setuptools quickstart, package discovery, and MANIFEST.in if you use data files

to be fair you still need a separate tool for managing virtual environments (venv), and if it's going on a package index you also want to build distribution files (build) and upload those distributions (twine or ci/cd like trusted publishing), so the convenience of one tool might be an easier choice for you

[quartz bough]

Is anyone using VSCode Devcontainers with Python?

[rapid sparrow]

i do, dont ask to ask though, just ask

[quartz bough]

I already asked 2 times the Help channel and they both got closed without answers. Just trying to understand where can I get some help about VSCode Devcontainer and Python

[rapid sparrow]

here is a good enough place, this channel is for containers and other stuff

if nobody answers -> then may be your question is not providing all necessary info, or having overly specific problem nobody dealt with, or 100 other reasons why it could be happening

based on you still not asking the question, i predict problem with formulating your question 😄

[rapid sparrow]

anyway, just ask your question already. not really able to help with current amount of provided information (or lack of it)

[quartz bough]

I'm trying to make a simple Devcontainer with only Pylint and Black formatter, but for some reason I can not get the formatter to trigger manually or on save. https://gist.github.com/ivoencarnacao/b748bb96ff75d67eb62240b4ff149a90 ps. Thanks for the patience with me!

[viral marlin]

There's a reason we stop that link from being posted. Do not bypass our filters.

[quartz bough]

What do you mean?

[rapid sparrow]

😊 okay. linking this server analog should work too

[rapid sparrow]

it looks like we are using Dev containers in entirely different ways 🤔

my usage did not require this config

may be we aren't even using same plugin

[rapid sparrow]

could u show your plugin icon and name?

[quartz bough]

I tried with flake8 and autopep8, neither did format manually or on save.

Pylint, works though

[rapid sparrow]

could u show your plugin icon and name?

and may be describing how u launch it

[quartz bough]

https://marketplace.visualstudio.com/items?itemName=ms-python.black-formatter

[rapid sparrow]

i mean plugin of devcontainer

[quartz bough]

Sorry, I don't understand what you mean!

[rapid sparrow]

you say u use dev containers with vscode
I assume it is plugin (and i think it is plugin different from the one i use)
could u show which plugin do u use for this service

[quartz bough]

I think we're talking different things

[rapid sparrow]

well, can't say for sure since u don't link (or screenshot) what u use.

[quartz bough]

This is what I use. https://code.visualstudio.com/docs/devcontainers/tutorial

I already pasted the gist with what I use.

[rapid sparrow]

Yeah, we use same plugin

I just use it differently

never made devcontainer.json

[rapid sparrow]

i used docker-compose to raise my containers each time and connected directly 🤔

[rapid sparrow]

nevertheless fun feature to try i guess 🤔

i have some question if i can define different containers though

hmm...

no, docker-compose is more reliable and IDE cross platform

not going to make dependency on obscure IDE plugin configuration

[ancient grove]

Guys has anyone ever faced this issue when building a ci/cd pipeline in azure app service using GitHub actions. IT was working before but its saying ive hit the artifacts quota. I already implmeneted some solutions in some stack overflow posts like including a step to delete the artifacts and setting the log retension period to 1 day. The delete artifacts step is not giving any errors, but this is still coming up

[sly crypt]

I have a GitHub workflow in which mypy tests fail on the Windows part of the project.

on: [push, pull_request]

jobs:
  build:
    runs-on: ubuntu-latest
    name: Mypy
    steps:
    - uses: actions/checkout@v1
    - name: Set up Python 3.11
      uses: actions/setup-python@v1
      with:
        python-version: 3.11
    - name: Install Dependencies
      run: |
        pip install mypy
    - name: mypy
      run: |
        mypy slodon/

it's because of the line runs-on: ubuntu-latest, what is the best of fixing this error?
Should I just change the run according to the system?

run: | mypy slodon/windows
run: | mypy slodon/linux?

[ancient grove]

maybe you would want to do that

[sly crypt]

maybe?

[ancient grove]

yeah, i am not too experienced with github workflow configs. but this is something i will do . if you are trying to run it for both windows and linux, just define 2 files

[sly crypt]

I'm kind of looking for a better solution, or an acceptance from someone.

[royal cipher]

Hi, I'm trying to merge and I got some conflicts and since they are determined to be complex I cant resolve them thru the web editor. The problem is that I can't re-produce it like I can't get to the point in my local vscode environment where I could see my merge conflicts and I'd be able to resolve them, how can I reproduct the conflicts locally so I can solve them.

[thorny shell]

ensure that your two local branches are at the same commits as the branches you were originally trying to merge.

[royal cipher]

The problem is that I just wanna get to the stage where I can possibly fix the merge conflicts, because I can't do it in the webeditor on github because it is too complicated, so how do I get the conflicts locally so that I can actually see what I have to fix?

[thorny shell]

I just told you how

did you not understand?

[royal cipher]

Solved it now

thanks

Now I merged it btw, but something's wrong with mypy, It just doesnt show up for some reason.

name: Mypy

on: [push, pull_request]

jobs:
  linux:
    runs-on: ubuntu-latest
    name: Mypy-linux
    steps:
    - uses: actions/checkout@v1
    - name: Set up Python 3.11
      uses: actions/setup-python@v1
      with:
        python-version: 3.11
    - name: Install Dependencies
      run: |
        pip install mypy
    - name: mypy
      run: |
        mypy slodon/slodonix/slodonix/exceptions
        mypy slodon/slodonix/systems/wayland
        mypy slodon/slodonix/slodonix/slodonix_wayland.py
        mypy slodon/slodonix/systems/x
        mypy slodon/slodonix/slodonix/slodonix_x.py
        mypy slodon/slodonix/slodonix/tween.py
        
    windows:
      runs-on: windows-latest
      name: Mypy-windows
      steps:
      - uses: actions/checkout@v1
      - name: Set up Python 3.11
        uses: actions/setup-python@v1
        with:
          python-version: 3.11
      - name: Install Dependencies
        run: |
          pip install mypy
      - name: mypy
        run: |
          mypy slodon/slodonix/slodonix/systems/windows
          mypy slodon/slodonix/slodonix_windows.py

@thorny shell

[willow pagoda]

err your windows job is indented inside the linux job

you can also look in the actions tab to see if github recognizes your workflow file but cant parse it

[royal cipher]

Arent they at the same level?

ohh wait

you're right

[willow pagoda]

the joys of 2 space indentation

[royal cipher]

😄

[royal cipher]

Yeah I wouldnt have noticed

[royal cipher]

Now they are both recognised, is there any way to make an if statement of smth?

[willow pagoda]

im not sure what specifically you want a conditional for, but you probably want to look through the github docs
https://docs.github.com/en/actions/using-jobs/using-conditions-to-control-job-execution

or if you want the linux/windows job to run based on which files were changed, i'd probably separate them into two workflows and use the paths: option for the push/pull_request trigger types
https://docs.github.com/en/actions/using-workflows/events-that-trigger-workflows#push ```yaml
on:
push:
paths:
- 'slodon/slodonix/slodonix/systems/windows/**'
- 'slodon/slodonix/slodonix/slodonix_windows.py'

jobs:
windows:
...``` (not entirely sure if the glob pattern in that example is correct)

[royal cipher]

I want to run the windows one whenever the os is windows and the linux one whenever it is linux.

[astral apex]

https://github.com/ClementTsang/bottom/blob/master/.github/workflows/build_releases.yml#L164

[rancid schooner]

.github/workflows/build_releases.yml line 164

if: matrix.info.os == 'windows-2019'```

[astral apex]

The only project I know of that does it pulls the info from a matrix

You might be able to access that information from like the github context without a matrix

[willow pagoda]

i was thinking of a matrix too, but im not sure how you'd use different commands for that mypy step, assuming the goal is to merge the two jobs into one

[sly crypt]

After a brief conversation, it appears that our specific goal is to execute Windows-specific modules on a Windows server, while the Wayland and X components can use the "Ubuntu-latest" environment.

(I'm working with him)

[astral apex]

Yeah
If the workflows are going to diverge significantly, just make two different ones
But it is entirely possible to stick both OSes in a matrix, and then use if(s) to hide the ones you don't need

[royal cipher]

The problem now is that on the mypy-linux workflow for some reason I'm running the files that are specified in the mypy-windows workflow as well. Why?

[willow pagoda]

im not sure why it checked the windows files, but apparently mypy can recognize platform assertions on a module level:
https://mypy.readthedocs.io/en/stable/common_issues.html#python-version-and-system-platform-checks

import sys

assert sys.platform != 'win32'

# The rest of this file doesn't apply to Windows.```

it might be easier to add those assertions and then have the workflow lint the entire package at once, then you don't need to update the workflow for new modules and you can use a matrix instead of OS-specific jobs

[gentle solstice]

because sys.platform is a constant

[smoky vigil]

is python version 3.11.5 is stable , iam developing a webapp in django

[rapid sparrow]

3 months ago libraries in general did not support 3.11 yet. Not recommending. 3.10 is more reliable

[subtle quarry]

Is there any tool to test concurrent POST calls? I split terminal window to like 40 panes and sync typed POST command that way. But there must be a better way?

[rapid sparrow]

<@&831776746206265384> we have continued having FU problem for your awareness

[sage cradle]

the user had been banned some time ago fwiw

[white adder]

hi guys

not sure if this is the right channel to post this in but i'll go ahead

i'm converting a python project into an .exe file

for this purpose i'm using pyinstaller

however i get this error when i launch the exe

Traceback (most recent call last):
  File "kiesel_gui.py", line 3, in <module>
    import kiesel_simulator
  File "PyInstaller\loader\pyimod02_importers.py", line 391, in exec_module
  File "kiesel_simulator.py", line 5, in <module>
    import seasonality_builder as src
  File "PyInstaller\loader\pyimod02_importers.py", line 391, in exec_module
  File "seasonality_builder.py", line 4, in <module>
    import holidays
ModuleNotFoundError: No module named 'holidays'

i had the same problem with openpyxl (i'm using pandas in the py files) and i was able to solve it using a thread from stackoverflow

but for the holidays module there is no previous thread on the internet, can someone help me ?

this is the command that solved it for openpyxl

pyinstaller main.py --hidden-import openpyxl.cell._writer

[subtle quarry]

Does this script work for you without pyinstaller?

[white adder]

yes of course

@subtle quarry

[subtle quarry]

Nice

[white adder]

yeah i checked again right now

the pb comes from pyinstaller and holidays

[subtle quarry]

Then probably pyinstaller doesn't support it. Don't know. Maybe you need to remove this package from your scrifor pyinstaller

[white adder]

scrifor ?

well i was thinking,

since i'm using a conda virtualenv with all my packages

isn't there a way to include all packages from the venv in pyinstaller ?

[subtle quarry]

Don't know don't use pyinstaller that often . Actually used it once

Are there any other packages that you use for that script and they work?

Maybe this https://pyinstaller.org/en/stable/when-things-go-wrong.html?highlight=modules#helping-pyinstaller-find-modules

[willow pagoda]

if hidden import worked for openpyxl, does adding --hidden-import holidays work too?

[white adder]

@willow pagoda first thing i tried

and it failed

this is why i came here

ok we are getting somewhere

i tried another time

i placed myself in the anaconda env i used for this project

and ran my pyinstaller command

pyinstaller kiesel_gui.py --onefile --hidden-import openpyxl.cell._writer --hidden-import holidays -w

when i launch the exe

i get this error

[willow pagoda]

i tried reproducing on linux using a tweaked version of their quickstart example and got the same No module named 'holidays.countries' error, and judging from their source code it seems to mostly be python modules and some locale data files so i went with --collect-all to bundle whatever was missing, now the script appears to run correctly py $ pyinstaller main.py --collect-all holidays ... $ ./dist/main/main date(2015, 1, 1) in us_holidays = True date(2015, 1, 2) in us_holidays = False us_holidays.get('2014-01-01') = "New Year's Day"

[white adder]

oh

you just added --collect-all holidays to my command ?

[willow pagoda]

im not packaging the same code that you're running so it wouldnt make much sense to use your pyinstaller command

[white adder]

to this command

[willow pagoda]

but try replacing the --hidden-import holidays with that

[white adder]

ok

ok i launched the command

brb

ok thank you, the gui launches now

the only pb is that it doesnt do the simulation i want it to do and there should be a matplotlib window popping when i press some button and it doesnt for some reason i dont know

maybe to debug i can take off the -w flag and have aterminal window to see the logs?

bro it's impossible to see the logs in a terminal... this thing driving me crazy

72046 INFO: Looking for dynamic libraries
C:\Users\ianne\anaconda3\envs\kiesel\Lib\site-packages\PyInstaller\building\build_main.py:194: UserWarning: The numpy.array_api submodule is still experimental. See NEP 47.
  __import__(package)
74755 INFO: Extra DLL search directories (AddDllDirectory): []
74755 INFO: Extra DLL search directories (PATH): []
80622 WARNING: Library not found: could not resolve 'impi.dll', dependency of 'C:\\Users\\ianne\\anaconda3\\envs\\kiesel\\Library\\bin\\mkl_blacs_intelmpi_lp64.2.dll'.
80626 WARNING: Library not found: could not resolve 'sycl6.dll', dependency of 'C:\\Users\\ianne\\anaconda3\\envs\\kiesel\\Library\\bin\\omptarget.sycl.wrap.dll'.
80626 WARNING: Library not found: could not resolve 'msmpi.dll', dependency of 'C:\\Users\\ianne\\anaconda3\\envs\\kiesel\\Library\\bin\\mkl_blacs_msmpi_lp64.2.dll'.
80626 WARNING: Library not found: could not resolve 'impi.dll', dependency of 'C:\\Users\\ianne\\anaconda3\\envs\\kiesel\\Library\\bin\\mkl_blacs_intelmpi_ilp64.2.dll'.
80626 WARNING: Library not found: could not resolve 'msmpi.dll', dependency of 'C:\\Users\\ianne\\anaconda3\\envs\\kiesel\\Library\\bin\\mkl_blacs_msmpi_ilp64.2.dll'.
80626 WARNING: Library not found: could not resolve 'pgmath.dll', dependency of 'C:\\Users\\ianne\\anaconda3\\envs\\kiesel\\Library\\bin\\mkl_pgi_thread.2.dll'.
80631 WARNING: Library not found: could not resolve 'pgc.dll', dependency of 'C:\\Users\\ianne\\anaconda3\\envs\\kiesel\\Library\\bin\\mkl_pgi_thread.2.dll'.
80632 WARNING: Library not found: could not resolve 'pgf90.dll', dependency of 'C:\\Users\\ianne\\anaconda3\\envs\\kiesel\\Library\\bin\\mkl_pgi_thread.2.dll'

i noticed that in the logs of my pyinstaller command

maybe that's what makes my matplotlib button not run

[sudden mesa]

Hey, I really want to get started with learning microservices and kubernetes . I've had decent amount of experience building applications using monolith or 3-tier architecture using python (django mostly with some fastapi), golang and a little typescript and have used docker as well. Could someone some good resources to get started?

[rapid sparrow]

Kubernetes recomendation is easy to give, here is a good book to jump start your experience
P.S. if u feel insufficient learning of docker, Docker Deep Dive book can fix it

[rapid sparrow]

as for architecture and microservices.
from your description i got a feeling of insufficiency to build monoliths at the necessary level yet.

Recommendation to read Clean Architecture first, in order to have better hang of structure manipulation at a single application
understanding better code gathering into reusable packages and structuring stuff in general
In order to utilize at full capacity, u need to use Golang.
Or if u want to try in python, Mypy in python in strict configuration can help, and usage of Pydantic + smth like Django Ninja/FastAPI
Best trying it in Golang first though, it is just far easier to do it there

Then u could read materials like From monolitoh to microservices and Building microservices

Microservices create additional problems (wormhole of addtional problems)... decreasing problems with code validation for correct cross service communications is a huge boon.
That is why recommended the stuff regarding tech i recommended

WARNING: Don't build microservices until u really learned to write scalable code of a monolith. Microservices create A LOT of additional problems.

[rapid sparrow]

P.s. I hope u learned unit testing to a good level 🤔 I can drop some books for that too

[latent bison]

RE: Pipenv up-grade issues

Status: (Just a novice pipenv/virtual environment user) I've been using pipenv to manage a few project's dependencies the last year while diving deeper into Python. I was recently using a version of Pipenv for a few months, without upgrading it. All was fine. Then the other day I decided to up-graded Pipenv when my IDE wouldn't recognize new ve's that I created. After up-grading pipenv and dependencies and running pipenv install in a project created by an IDE I was presented with a long list of "issues" being displayed in my cmd window culminating in a permission error - not being allowed to access some postgres data files.

Question: I did not experience this issue weeks ago with the older version of pipenv when creating a new environment, why am I now presented with pipenv wanting access to postgres data files when to my knowledge access to some postgres files were not required previously?

Thanks for the help.

[thorny shell]

can't imagine that's anything to do with pipenv itself

I'd look carefully at the IDE

[latent bison]

ahh.... not sure how an IDE effects installing a pipenv ve via pipenv install in a project other then the initial and un-populated project was created by the IDE, but it does appear that the sub-category of the IDE (Eclipse) hasn't been supported for about a year. You may be onto something though with your suggestion and I may now be faced with a bigger issue in the middle of a class that I'm trying to complete, switching IDEs. (pulls hair out meme)

[thorny shell]

in my experience, IDEs are more trouble than they're worth, unless you're using the same one every day

very very complex

[latent bison]

it does seem that way especially for someone that isn't a systems expert and/or enjoy staying abreast of the whole dev environment.

how I recover/transition to another dev environment mid-educational class experience when I need to focus on some python libraries will be....

[latent bison]

appreciate your attention

[thorny shell]

it doesn't help that VSC and Pycharm change notably as time passes 😐

presumably those changes are meant to be improvements, but they're disorienting

[heavy knot]

Can someone help me put my discord bot code from replit to put it in a vps, i bought a vps but idk how to put my code in

[misty otter]

Which VPS did you buy?

[thorny shell]

hoping it's *nix and not Windows

[misty otter]

Honestly I highly doubt they bought one as they instantly sent the message

Can't purchase a VPS in ~15 seconds lol

[thorny shell]

with lightsail, it's not out of the question

[astral apex]

Can't you?

[misty otter]

For a beginner who had no clue what a VPS was I would say it's unlikely

[thorny shell]

true true

.oO(why had I never come across PEP 9001 before)

[misty otter]

It's the final style guide for python

Objectively the best

[thorny shell]

clearly

I read a paragraph or two before scrolling back up to check the publication date

[cobalt dome]

a topic that i am moving from #ot0-psvm’s-eternal-disapproval

[rapid sparrow]

that depends 😁 both paths are available

1. U can ask for client credit card
2. u can charge client for money u spent on their server resources (putting this money into invoice to them)

That depends on

1. Your hiring contract type, are you employee? Are you b2b enterpreneur? are you freelancer

if you are employee, first path i think is obligation
freelancer... both are usable paths
enterpreneur can utilize both paths too. Depends on his scale of entrereneurism and desired scope of services to provide

So for example... if in the end u are just full time working for someone, first path is prefered
if you are only providing services but not working for them, second path can be prefered (in order to make client more depended on you and providing more obfuscated end result, just giving end result, the working service)

[cobalt dome]

We are just two friends trying to do some biz, and will do some biz in the future too, so yeah freelancer.

[rapid sparrow]

i would say for freelancers and enterpreneur, this is all depending on client requirements. I would not be closed to all options and may be even having for each client different way

[cobalt dome]

Also that scale of entrepreneurism bar is pretty low because I hate tinkering with AWS billing, account managing, organization such so

[cobalt dome]

so this is going to be a literal headache eh

[rapid sparrow]

if you are freelancer / enterpreneur, it is YOUR CHOICE which services to provide 😄

u just limit the scope of clients u agree to work with

and it is completely fine

[cobalt dome]

ooooooooh that is sensible

[rapid sparrow]

i refuse to work for any money with NFT for example (even if offered 200% of current income)

[cobalt dome]

yeah i can limit that too

like i can decide one principle, we are dumping aws bills on you sya

[rapid sparrow]

Freelancers /enterpreneurs can be having exactly presenting web sites that least services they can provide 😁
or they can just tell it during negotiation with clients, your choice

it is your responsiblity to extract requirements out of client

and to make client to agree to certain contract of what will be done and how much will be paid

[cobalt dome]

yeah, unfortunately

i will dump all these talking parts to my friend

i hate any kind of client interaction and I didnt expect to have this many

[rapid sparrow]

for freelancers and enterpreneurs it comes by design (interacting with clients)
employees aren't really escaping this thing too btw, development is a team work.
Any dev from middle rank is expected to interact with other people and extract requirements (mostly from other devs)

[cobalt dome]

i will dodge any role that requires human interaction in the future lmao

there is too much to unpack

thanks for all tips

[strong bison]

PS C:\Users\Isteyak\Downloads> docker -v
Docker version 24.0.6, build ed223bc
PS C:\Users\Isteyak\Downloads> docker ps
error during connect: this error may indicate that the docker daemon is not running: Get "http://%2F%2F.%2Fpipe%2Fdocker_engine/v1.24/containers/json": open //./pipe/docker_engine: The system cannot find the file specified.
PS C:\Users\Isteyak\Downloads> history

Id CommandLine

---

1 docker --start-service
2 cd Downloads
3 docker run hello-world
4 docker info
5 docker info
6 docker logs docker-desktop
7 docker logs docker-desktop
8 "Docker Desktop Installer.exe" install
9 "Docker Desktop Installer.exe" install
10 Start-Process 'Docker Desktop Installer.exe' -Wait install
11 start /w "" "Docker Desktop Installer.exe" install
12 start /w "" "Docker Desktop Installer.exe" install
13 docker run redis
14 docker -v
15 docker ps

no mattr wat i cant set up docker in my p

pc

[mint cipher]

Maybe uninstall it completely, restart PC, and then reinstall?

[rapid sparrow]

i can't recognize all commands, but just to be sure Docker Desktop application is supposed to be running during docker ps command

[indigo zenith]

Is Docker running? In other words, do you see the whale in your system tray?

[strong bison]

That's what I'm thinking of

[strong bison]

Ye

[strong bison]

So weird it works now still curious but don't have time to research on what really went on, Dockr app starts without err and Dockr PS command doesn't give err either

[scarlet coral]

hey guys does can someone help me with tkinter?

I'm done with the introduction of python

and now I'm trying to implant a python project

but since I want to make a graphical interface for it

I need tkinter

but I'm getting overwhelmed with the concepts and subjects of tkinter

so i can use any help

[indigo zenith]

I would try a tutorial or two to start, maybe like this https://realpython.com/python-gui-tkinter/

If you're still stuck share some of your existing code in a help channel #❓｜how-to-get-help

[hoary veldt]

Hi guys! Anyone here developing in VSCode using devcontainers? I'm trying to set up kind of a template project for an backend/API application. It been some time since i developed using Python so I could need some input on how to configure my dev environment correctly.

[alpine horizon]

how do I reset the whole .git folder to its current state in the remote repository?

[rapid sparrow]

sudo rm -R project_folder ; git clone ssh_url 😅

[rapid sparrow]

if u want specific branch only git reset --hard origin/desired_branch to reset
WARNING: it will reset everything, files included

[alpine horizon]

yeah somebody told me to avoid the reset command...

I tried to create a new branch from an old commit git checkout -b new_branch commit_hash but for some reason its content is not matching the content of that commit hash

not sure how I fix that

in fact its content is exactly the same as the latest commit from the branch of commit_hash

[rapid sparrow]

everything is good to use as long as u understand its dangers and when right to use 😉

it is really bad reseting branch used together with others to older commits

u should use only git revert in those situations

reseting is good only.. if u just reset local state to remote
or u reset your personal feature branches to older commits, then it is fine

[alpine horizon]

thats fair, I'll try revert lets see how it goes

[tranquil current]

my github copilot subscription is not working as a student even though I have the student devleoper pack

does anyone have the same issue?

[hoary veldt]

how do I set up venv in a devcontainer? should this be done in the Dockerfile or a post-script from the devcontainer.json?

[hexed osprey]

i wanna make a bot that saves all the unknown number in a group that i am member of. the bot will keep running when mobile start and look for unknown number. If get one then save the number to google account with a string. How can I aproch this.

[rapid sparrow]

why not just to skip setting up venv in devcontainer 😁
install stuff in devcontainer as it is without venv

[hoary veldt]

I've come to understand that it is best practice to use venv even inside a devcontainer. In any case, it cannot hurt ... I just want to do it properly.

[rapid sparrow]

the answer somewhat depends on what u use for dependency management

[hoary veldt]

It's been a while since I developed with Python so I am trying to set up a template environment from scratch that I can use as a basis for new projects. I.e. I'm not really married to any dependency management

[gleaming thunder]

what are afew tools to use python

[sharp flame]

anyone here good with fixing circular improts?

imports*?

[hasty seal]

Hi!
I am trying to move back to a dev role (I was a backend developer ~10y ago) and took the chance to write some tools with python that help me on my daily work. Is there a channel where sharing the repo for these tools is more appropriate?

[rapid sparrow]

GitHub and Gitlab both would work. GitHub is more common despite its issues

[lusty orchid]

how to send messege in messenger using py

[hasty seal]

sorry if my question wasn't clear... I was asking if there is a channel on this server where it is more appropriate to share my repo, so that people can provide feedback and inputs. The code is on github 🙂

Some servers don't allow these type of posts, and have a dedicated #showcase channel for it.

[rapid sparrow]

ah, #software-architecture is dedicated to software... well, design topic. in terms of code architecture.
U could try there as a choice

[rapid sparrow]

advertisements aren't allowed though, stay on topic 🙂

for the goal of..
having questions regarding design of app in software design
or for career channel as presentability for resume
or smth, just not adds

[hasty seal]

thanks!

[indigo zenith]

I'm curious on what basis you came to that understanding

I'm not that expert but I can't think of a reason you would want a venv inside a container and I've never seen it done. If you think you need to isolate dependencies inside a container I would guess separate containers would normally be more appropriate

[indigo zenith]

Is this what you're asking for? https://pypi.org/project/messenger-api-python/

[heavy knot]

can anyone crack a py file that is as app ? if ye dm me

[mystic void]

!rule 5

[rancid schooner]

Rules

5. Do not provide or request help on projects that may violate terms of service, or that may be deemed inappropriate, malicious, or illegal.

[nova wadi]

Hi,
Can I share a Django bnased premium admin template here?

Or just open source projects are allowed?

[thorny shell]

wish I knew

[indigo zenith]

A lot of people seem to like KodeKloud but I don't have a subscription. Most of my AWS learning has involved study books for their certifications.

[hazy echo]

YOu may want to follow #tools-and-devops for a few days to see if it looks like the kind of discussion that would be relevant.

[hazy echo]

I use Kode Kloud for the Kubernetes certification and it seems ok. I don't have a lot to compare it too. They have a pretty active Slack community for each class.

[willow drift]

Guys is learning devops and cloud computing where the money is at? I am noticing at work the guy doing all the cloud stuff seems to be the most important. He built all the infrastructure and pipelines using terraform, boto3 and other stuff I don't understand too well. He designed it in a cost effective way

By using kubernetes and microservices. If I learnt how to build infrastructure in a cost effective way is that the most important thing from a business perspective

[rapid sparrow]

Cost effective
Auto scalable
Fault tolerant
Observable (monitoring, logging, tracing)
Secure
Minimizing manual actions to automated
Keeping stuff as Infrastructure as a code for better self documentation

There are more parameters than just cost effective 😉

[gaunt valve]

.devcontainer multi workspaceMount

Hello, has anyone here worked with devcontainers and knows how or if it is possible to mount two folders into the workspace?

somethign like this, but the problem is, that workspaceMount only takes in a string:

{
    "name": "Python 3",
    "image": "mcr.microsoft.com/devcontainers/python:1-3.11-bullseye",
    "features": {
        "ghcr.io/akhildevelops/devcontainer-features/pip:0": {}
    },
    "postCreateCommand": "pip3 install --user -r tracker/requirements.txt",
    "workspaceMount": [
        "source=${localWorkspaceFolder}/tracker,target=/workspace/tracker,type=bind",
        "source=${localWorkspaceFolder}/utils,target=/workspace/utils,type=bind"
    ],
    "workspaceFolder": "/workspace"
}

[rapid sparrow]

why not to mount whole project folder once into dev container 😁

best decision for dev containers in my opinion

[gaunt valve]

also for a mono repo?

[rapid sparrow]

potentially yes. If u will manage making everything launchable from same single forwarded project folder and having launchable all your monorepo processes just under different command path, it will make you simpler life yeah

having single docker image is kind of simpler

kind of fits monorepo. monorepo = monoimage 😄

[indigo zenith]

This is more of a #career-advice question but the money is at doing valuable and efficient work. Whether that means you should focus on learning DevOps skills or something else has a lot to do with your interests, talents, social connections,.etc.

[heavy knot]

Can someone code a small thing for me? I will explain it in Dm and I think it will only take around 20-30min to code

[subtle quarry]

Try #❓｜how-to-get-help

[tawdry needle]

Also there's value in not using microservices when they aren't needed

[ancient grove]

Hi guys, I have successfully setup jenkins and github, so that each time i push new changes to the v1 branch it will trigger a jenkins pipeline to build and push the image to dockerhub. now i have another vm where i pull the image and run it, and i have a service which would check periodically to pull and restart my docker container if the dockerhub image is updated by the jenkins pipeline. Now when its restarting the api is down for some minutes, I know a solution would be to use a load balancer , but i just want to be sure if this is an appropriate way or is there a better way. I already tried using azure app services but its impossible to disable the default url azure assigns an app instance. So ive decided to setup jenkins and host the api in a vm, then use cloudflare tunnel to proxy requests from a subdomain to the server

[astral apex]

Why does it take multiple minutes to restart?
Are you not pulling the image before you restart?

What's wrong with having the app.azurewebsites.net domain floating out there?
I use App Services at work, and it gave me a webhook to hit to prompt an update.
So I have a GitHub Actions workflow that builds and publishes the container, and then hits the webhook to prompt the App Service to download the new container.
And it waits to recycle the container until it's ready, it still finishes all in progress requests, so you don't really have any downtime there.

I'd really recommend that automated workflow that's all-in-one place if you're just looking for the simplest workflow.

[ancient grove]

yes i understand but we are also trying to cut down costs because we have several applications to deploy and doing that in one vm at the moment i think is a great choice as supposed to creating 3-4 azure app services for it

also we are paying for github enterprise subscription

I am currently researching for a better option, from what ive seen we might have to throw in ks8 into our workflow

[astral apex]

You know you don't have to pay separately for each App Service, right?
You buy the "plan" and you can run as many services as you have available resources for on that plan

[astral apex]

GHEC is really nice, but it doesn't really give you anything here
Everything I'm talking about I also run on GitHub's free plan

[astral apex]

K8s is really cool
Though it can be a pain if you're only running a couple services
I only have like two
I actually switched away from K8s to App Services

[ancient grove]

no we cant use free plan, its a company and we also have other workflows that use github actions that use up our credits and free allowances

[astral apex]

No I -- I didn't say you have to
Just that it doesn't really add any benefits to building containers other than the additional minutes

[astral apex]

What's wrong with having the app.azurewebsites.net domain floating out there?

Is there a compliance reason?

Is that the main reason you're not using them?

[ancient grove]

hmm... i see, maybe you can give me idea how to setup our workflow, its a really huge application which ive built with fastapi, i was using github actions before and we ran out of storage, then i thought of a better option to use jenkins, which i already setup and its working

[astral apex]

You ran out of storage on GHEC?

[ancient grove]

yes kind of, we configured a subdomain like api.maindomain.com but i want to connect that to cloudflare tunnel because we also have other subdomains connected to that tunnel.

[ancient grove]

yes we did

artifacts storage

[astral apex]

How big is your container?

[ancient grove]

we were using the free plan and our github actions minutes was almost used up

[ancient grove]

no, not the size, as you use github actions it will keep track of your usage and in freeplans there is a limit to how many minutes and storage you can use, it refreshes every month

[astral apex]

Here's my open-source repo that I have an App Service: https://github.com/letsbuilda/api.letsbuilda.dev

Here's the workflow: https://github.com/letsbuilda/api.letsbuilda.dev/blob/main/.github/workflows/docker-build-push.yaml
It runs this reusable one: https://github.com/darbiadev/.github/blob/main/.github/workflows/docker-build-push.yaml
Which caches the build to and publishes to the GitHub Container Registry
And then after that hits the webhook to prompt Azure to get the new container

Nothing required but the one GitHub repo for the entire build process
And then nothing but a webhook required for the entire deployment process

[ancient grove]

okay i see

[astral apex]

It's a public container, so no creds
But if it was private you'd add a PAT here
AFAIK there's still no packages support for fine grained tokens, so you still have to use the classic PATs

[ancient grove]

one question, i already have jenkins setup and it works file, i think i can still use that with dockerhub and azure

[astral apex]

Yeah
You can manually update in Azure
Or have Jenkins hit the webhook

[ancient grove]

i do not understand, if i deploy a new azure app service and set the source to be a container image instead of github project, I think azure will watch the container registry and pull my new image once i push to it

right ?

currently jenkins will push the built image to dockerhub

[astral apex]

It doesn't automatically pull updates AFAIK, you have to prompt it

[ancient grove]

wow, i didnt know that

anyways now that i know that it works, ill try it out

[astral apex]

I "just" have a CNAME
Is there something specific about Cloudflare Tunnels that makes you want to use them?

[ancient grove]

what are your thoughts on using kubernetes ?

[astral apex]

Kubernetes is great

But it requires management

[ancient grove]

you know its just easy to map a subdomain to some ip address and port

[astral apex]

If you only have one or two containers, just make the App Service for the simplicity
But if you start getting into many containers, it's time for Kubernetes

[ancient grove]

when i turn that proxy switch on, it gives me another error like bad gateway or so

[ancient grove]

okay

please i will ping you after some hours, incase i tag you

[astral apex]

That's my thing
When I had Kubernetes, I had to setup NGINX, I had to setup Cert Manager, I had to setup ingresses and services and deployments
When I tried out App Services I said "hey here's my container" and it said "okay I'm done"
I didn't even believe it at first. Coming from Kubernetes, I couldn't believe it was that simple
It was a big sigh of relief to not have to deal with all that crap anymore

[ancient grove]

yes app service really simplifies the process

ive been using app services for some time now and its been great

[astral apex]

But I'm not bashing on Kubernetes.
It definitely has it's place.
I'm the infra lead at https://vipyrsec.com/ and we just switched to K8s (from App Services BTW)

Lots of containers and it was going to be a pain to make that many App Services

[brazen forge]

just

[astral apex]

Yes?

Also BTW, App Services will kill your container if it doesn't respond to HTTP requests

[ancient grove]

yes

[astral apex]

So it's a non-starter for non web apps

[brazen forge]

(just poking fun at the fact we've been on k8s for quite a while :P)

[ancient grove]

i will definitely learn ks8 soon

@astral apex i really appreciate your time

[astral apex]

Of course, happy to help

[ancient grove]

is this a subdomain ?

[astral apex]

Yeah
The api subdomain is a CNAME that points to apiletsbuildadev.azurewebsites.net

[ancient grove]

haha, i dont know why azure didnt provide a way to remove this default domain

it would have been great to use an ip address

[astral apex]

Not sure
But what's wrong with it being there?

[ancient grove]

its making me feel like i have a backdoor to my server , 😁

the main api domain is protected with cloudflare

if anyone finds out that there is an azure variant they can just make requests to that one

right ?

[astral apex]

true

[ancient grove]

yes, because i also do that when im web scraping

so you see its a bit concerning

[astral apex]

This looks neat
https://omribeladev.medium.com/disabling-the-default-domain-for-azure-web-app-a-step-by-step-guide-4e48f8fb5a7f

[ancient grove]

i actually implemented the solution in this blog more than 2 times

for some reasons its not working for me

[astral apex]

huh

[ancient grove]

it didnt workout for me

they solution is basically to enable access to the sire only by defined ip addresses

then you will add all cloudflare's ip addresses

i did all that but its still saying that azure blocked my request

[ancient grove]

how do i tell azure to pull the images from dockerhub when i push a new image to dockerhub with thesame tag ?

I have turned on ci/cd but its not automatically pulling the new updates from dockerhub

[astral apex]

AFAIK you can't

[astral apex]

That enables the webhook

So when you're ready to update, you send a POST to the webhook

[ancient grove]

okay

so i have to do that manually like write code for it ?

maybe i will just include that in my jenkins pipeline as the last stage

[astral apex]

I do https://github.com/letsbuilda/api.letsbuilda.dev/blob/main/.github/workflows/docker-build-push.yaml#L22-L28

[rancid schooner]

.github/workflows/docker-build-push.yaml lines 22 to 28

deploy:
  needs: build-push
  runs-on: ubuntu-latest
  if: ${{ github.ref == 'refs/heads/main' }}
  steps:
    - name: "Update deployment"
      run: curl -X POST '${{ secrets.DEPLOY_WEBHOOK }}'```

[ancient grove]

thank you

[ancient grove]

it didnt work

i also sent the request manually from postman

but i didnt notice anything

@astral apex it gets triggered if i toggle ci/cd on/off

but when i send post request to the webhook it doesnt work

[astral apex]

Huh

How?

[ancient grove]

see here

[astral apex]

Look right

What do the logs say?

Do they have a bunch of crap about invalid characters because you didn't quote the URL?

[astral apex]

Huh
Not sure
Do you get a response?

[ancient grove]

Nope

When I send in post man its just empty response

[ancient grove]

From here, I dont think any packets if data is being sent

[astral apex]

I normally see the responses

[ancient grove]

yes

here its sending

[ancient grove]

in your case its sending

[ancient grove]

I am getting 403 error

[astral apex]

My guess would be you need to quote the URL

[ancient grove]

but i am sending the requests from postman

[astral apex]

Are you sure you have the entire URL?

Not really sure with Postman

Don't use it much

[ancient grove]

yeah

@astral apex I think using docker-swarm or kubernetes would solve this issue

they offer automatic image update

[astral apex]

FWIW CD for Kubernetes does practially the same thing

[ancient grove]

cos the only issue im having right now is updating the image when i push to dockerhub and still keeping the server running

[astral apex]

How?

[ancient grove]

im talking about imagePullPolicy - always

what do you think ?

[astral apex]

it will always pull
when it starts
it doesn't start (or re start) until you tell it to

https://github.com/vipyrsec/infrastructure/blob/main/kubernetes/manifests/dragonfly/mainframe/deployment.yaml#L19

[rancid schooner]

kubernetes/manifests/dragonfly/mainframe/deployment.yaml line 19

imagePullPolicy: Always```

[ancient grove]

oh

is there a way to tell it to start by sending request to some webhook

*to restart

[astral apex]

I have it in all of my deployments
I still have to manually login and kubectl rollout restart deployment/mainframe --namespace=dragonfly to actually trigger that pull

[ancient grove]

okay

maybe i can write a webhook in flask or fastapi

to call that command

do you think that would work ?

[astral apex]

I haven't explored that yet, but here we use some GitHub Actions that Microsoft make to interact with our cluster
https://github.com/python-discord/bot/blob/main/.github/workflows/build-deploy.yml#L56-L82

[ancient grove]

okay

will this load balance all traffic, because remember the aim is to update the containers without causing downtime

is this the case ? @astral apex

[astral apex]

What is the "this"?

[ancient grove]

I was referring to the workflow file you shared

[astral apex]

No
The workflow file doesn't have anything to do with how Kubernetes handles the running containers

[ancient grove]

Okay

Bro, I honestly dont know what to do

[astral apex]

AFAIK Kubernetes will always automatically try to gracefully restart

You can see when I restarted the bot the old instance stayed "running" while the new one was "pending"
And Kubernetes didn't start "terminating" the old instance until the new instance was "runnning"

Stayed? Staid?
Anyways

[ancient grove]

I see

So its load balancing the operations

[astral apex]

I'd recommend App Service(s) if you're trying to keep it simple
Kubernetes is a LOT to learn

[astral apex]

no

There are multiple types of load balancers, but this is not that
There is only one container at a time, so there is no balancing between them
Right therory, wrong term

[ancient grove]

Im looking for something that works , ive been sitting here since the past 17hours

I may use ks8

[astral apex]

Did App Services not work?
Other than the domain
Did you actually have problems with downtime?

[ancient grove]

I cant trigger the app service to pull my images from dockerhub

That was the webhook thing . When I call the web hook it gives me 403 forbidden error

I already checked , there are no ip restrictions

[astral apex]

Huh
Yeah
Not really sure there
I'd work on that one first

Because you already have it running
Before trying to go down the road of setting up a whole 'nother system

[ancient grove]

Yh

[mystic void]

now you have 15 problems

k8s is a complex beast

[ancient grove]

Haha, thats also what people said about jenkins

[astral apex]

Why do you use Jenkins?

I literally threw my server away after I got rid of Jenkins

[ancient grove]

Its easy to use I think

[astral apex]

It was tainted

Couldn't trust it anymore

[ancient grove]

Why?

[astral apex]

(not really I just replaced it with a newer one)
(but Jenkins was really annoying)

[mystic void]

jenkins is nice, but there are more modern solutions with gha or gitlab ci or circle ci?

[astral apex]

Just got everything I needed from GitHub Actions, with a lot less effort

[ancient grove]

Yeah, I faces alot of issues while setting it up but I found a solution for all of them

[astral apex]

Don't really remember the details, it's been... 6 or 7 years since then

[ancient grove]

I chose jenkins because its opensource and we dont have to pay for anything but the vms to host it on

[mystic void]

same thing with gitlab ci 😉

or drone ci

[ancient grove]

We were using app services, but its expensive

Not app services, I meant github actions

[astral apex]

Are your VMs cheaper than GitHub?

How many minutes do you use?

If you already have the resources available, you could also look into self-hosted runners

[ancient grove]

I am not sure but we are running other workflows apart from our api

Okay

I see

But Why not jenkins?

[astral apex]

Just one more thing to keep track of and figure out

[ancient grove]

What?

[astral apex]

Doesn't really matter all that much

[ancient grove]

Idk Why its just almost impossible to trigger a webhook

[ancient grove]

Hey, I managed to get a 202 accepted status when I call the webhook from postman

Ive been setting an Authorization header all this time 😩

[astral apex]

nice

Why?

[ancient grove]

Was a mistake

[astral apex]

[ancient grove]

I just tested it, after pushing to dockerhub, when I send the request from postman, it pulls the updated image and restarts the container without downtime

[split crypt]

Hi! I'm working on developing a simulation tool for underwater gliders in python and have lately been stuck on an issue. I am required to solve ODEs in my code, but I keep getting different plots when I use scipy.integrate solve_ivp(), odeint(), and RK45(). The matlab function ode45 works well but I am unable to get the same results on python. Any help would be greatly appreciated!

[mystic void]

Please, do not cross post

[split crypt]

sorry, I wasn't sure which channel I should post it in

[rapid sparrow]

We use GitHub actions too.. it has self hosted runners which are free

We just have them with already provided autoscaling solution... and enjoy free gh

[ancient grove]

Great

I solved my issue anyways, after 18hrs of debugging

[ancient grove]

Hey, I solved all issues, thank you, I really appreciate your time

The issue Why the curl was not sending was because I didnt wrap the url in quote

[astral apex]

Happy to help

[astral apex]

Ah

[ancient grove]

Haha

Took me tons of hours to realize

[astral apex]

Oof
Good job getting it fixed

[ancient grove]

If I do this , I am getting 404 site not found

[astral apex]

Huh
Did you add the asuid?
Does the domain show as active?

[ancient grove]

I added the asuid

But azure app service is saying its not able to find the cname

[astral apex]

I think you have to leave proxying off for the initial setup

[ancient grove]

Ok

I disabled proxy and it passed now

[astral apex]

Double check for typos and accidental spaces and all that

Oh good

Yeah, I generally wait half an hour or so for the DNS and the SSL to get propagated a bit before I enable proxying

Should be fine whenever though

[ancient grove]

Wait

Did you select ip based ssl

[astral apex]

Uhh

No

[ancient grove]

Or sni ssl

[astral apex]

SNI I’m 99% sure

I didn’t change the defaults

[ancient grove]

Ok

[astral apex]

The one that’s not the default needs extra configuration

[ancient grove]

Ok

[royal silo]

I currently only have a main branch, if I commit all changes of main branch, then create a new branch rewrite where everything is completely different; there are new folders and files, and then I commit the changes to that branch. Now when I come back to main the folders from the rewrite branch (everything else is as expected) are still there. How do I correctly handle this situation?

[brazen forge]

you committed on main and then created a rewrite branch
so main also has the commit
you'll need to hard reset main to the previous commit

switch to main and then do
git reset --hard HEAD~

[royal silo]

I also heard from SO answers that Git does not track folders, how true is that?

[brazen forge]

yeah, it doesn't track empty folders

a common way to track empty folders is to make a .keep or .gitkeep file in them and track that

[royal silo]

I undid two commits, but the impl folder is still there, this folder is from the rewrite branch. What I found out just now is that impl is filled with empty folders with no files that exists in the rewrite branch's impl folder

[brazen forge]

check for hidden files
ls doesn't show them by default

[royal silo]

fixed it

[ancient grove]

@astral apex i selected ip based ssl for the proxy option to work. Ip based ssl means ill use the certificates from cloudflare and to hide the ip address you have to select this option in the SSL/Tls Tab for your domain on cloudflare

[royal silo]

so i worked a huge amount of hours, like 8 hours, then I stashed my changes, but then my laptop had an unfortunate shutdown, now that I come back, all those files are empty and the stash record is missing.

did i lose all my files to corruption?

[rapid sparrow]

git stash list

       list [<log-options>]
           List the stash entries that you currently have. Each stash entry is listed with its name (e.g.  stash@{0} is the latest entry, stash@{1} is the one before,
           etc.), the name of the branch that was current when the entry was made, and a short description of the commit the entry was based on.

               stash@{0}: WIP on submit: 6ebd0e2... Update git-stash documentation
               stash@{1}: On master: 9cc0589... Add git-stash

           The command takes options applicable to the git log command to control what is shown and how. See git-log(1).

from git stash --help

[split crypt]

I've been using the scipy solve_ivp() function to solve a few ODEs. Can anyone tell me why my plots are tracing backwards? The x-axis is time

[indigo zenith]

Without seeing the code your guess is better than ours

[solemn osprey]

For some reason storing 30 gb of metrics on S3 causes Thanos to waste about 500 gb of S3 traffic daily. What settings should I look into to reduce traffic consumption?

[timid schooner]

I want a little help from anyone with a simple problem in a tool I make!

[rancid schooner]

:ok_hand: applied timeout to @dim tusk until <t:1696446189:f> (10 minutes) (reason: duplicates spam - sent 4 duplicate messages).

The <@&831776746206265384> have been alerted for review.

[hidden atlas]

What do devops do in python? kinda a question but eh

[willow pagoda]

the same stuff as devops in any other language i would assume, i.e. whatever gets an app from development to deployment

something more python specific that i've done is use github workflows (aka CI/CD) for running pytest on push and building/publishing library wheels, alongside https://readthedocs.org/ to host and continuously update my sphinx documentation

[hidden atlas]

I wanna purse a carrer in devops but I legit know nothing about programming

Someone told me to do the aws and azure certs

[willow pagoda]

hmm there much more qualified individuals here that can answer that

though it might also be more appropriate to talk about in #career-advice

[hidden atlas]

Okay

[rapid sparrow]

Scripting glue for CI CD means
Cloud provider lambdas that help to augment infra with more custom solutions
Configuring monitoring systems
Making dev tools
Configuring release/deployment system

If they are free of infra stuff and have strong enough backend side then
Using as means to debug databases
Writing libraries, backend
Doing pretty much backend duties if necessary

In my current company DevOps engineer is for good half is pretty much backend developer.

When I get tired of too much infra, I grab more backend, more pure dev duties

When I just wish to relax a bit more, I go and solve some infra thing like fixing autoscaling

[rapid sparrow]

DevOps engineering for different companies have different meanings though, exact percentage of Dev, Infra and Ops duties will be different

I know a guy that wishes to do 80%+ or even purely infra only stuff. Apparently he worked in place that provided it to him. That is weird to me, there is always stuff in between
Or as I said I prefer to take a break from infra only duties from time to time and just to code some application, more dev project

[hidden atlas]

Ill prob do the same if im dev ops

Will get tired of infrastructure

[rapid sparrow]

Development is kind of more creative u know, involves different set of brain working.
Taking care of infra stuff is going over monitoring systems to find out about issue, reading more docs and writing more simplistic infra code. Different workflow a bit

Both duties have their own merits but I would not have wished only one set of them

Or even Ops duties. Helping to people and taking care of Ops can be nice. I am altruistic enough. But gets tiresome if doing too much too

Preferring to do it only as small part of a day and not every day then I can enjoy it, in getting feedbacks of thanks 😁

[hidden atlas]

How many hours you be working in a day?

[hidden atlas]

Yea I get bored doing repetitive stuff too,

[rapid sparrow]

8-/+. Not sure exactly.
I have flexible schedule, sometime I work more, sometimes less.
I just try to have done enough to write down 40 hours made of work during a week.
Still figuring out work balance thing

Sometimes day can last 12 hours. I am bad at my time management, personal problem pretty much

Often enough I start day only during middle of a day though

[hidden atlas]

You get to choose your own working time?

[rapid sparrow]

Yeah. Very flexible schedule.
Highly preferably I would work 8 hours in same day with start and end at any part of a day with any pauses.
In the end it is important only reporting 40 hours during a week, so I can even jiggle hours to other days if necessary

I need to appear at morning standup though

[hidden atlas]

That would be 1 thing that I would hate the meeting

how are they?

[rapid sparrow]

And I am expected to answer... relatively fast enough (within an hour?) During normal workday time

[rapid sparrow]

Managable, just say what u did yesterday and what u a doing today

I got used to that

[hidden atlas]

Oh that is fine then

[rapid sparrow]

Kind of even nice. I am fully remote, during stand-ups i have at least some live communications with humans and see my colleagues

[hidden atlas]

What about the ops part?

dont you have to be in person

[rapid sparrow]

My Ops is just handling requests (asks for help if smith is broken, or helping to investigate smth) coming to slack channel of requests to DevOps team.

Pure Slack

I just pick stuff I can help there with, that got my interest

Other team members pick stuff they are able to solve

That makes eventually good enough coverage of solved problems

Doing it when wishing to slack off from main duties 😉

[hidden atlas]

You will have something you said you did lol

for them meetings atleast

[rapid sparrow]

Sometimes more emergency stuff appears though

Infra meltdown somewhere

Then it becomes not optional but needed being solved as soon as possible

[hidden atlas]

those moments can be stressful right?

[rapid sparrow]

It can be energy exhausting and going overtime often enough

Important to eat and drink in time 🙂

[fresh mesa]

how to write socket pytho that can ddos a server

[hidden atlas]

don't you get paid extra for overtime

[rapid sparrow]

Really bad to be too low on energy, starting to feel too bad then

[hidden atlas]

but that may come with some downsides too

company finna hate you for taking it

[rapid sparrow]

U know, i don't. But may be I am an idiot that missed how to report it 🤔

I ll check our docs

Usually we report 1x rate any time

And as I said important only having 40 hours per week

So, If I did more today I can be more lazy tomorrow

[hidden atlas]

Why you staying at 40?

8 hours or so a week finna only be 5 days

[rapid sparrow]

Yeah, I am expected to work 5 days in a week

That is why 40 hours

[hidden atlas]

oh that is weird

you cant go above it 40?

[rapid sparrow]

Any overtime for that must be agreed upon with manager for us.

We can, some people are crazy to work much more. I don't wish so

I need time for personal life beraucracy and for self studies and some just for personal mental health

[hidden atlas]

damn but fr if you need time you should get everything done early since you said you was starting work late

[rapid sparrow]

That is questionable approach. If I work too much in one day, I can get exhausted too much and not having mental capacity for tomorrow until very late

I try to be careful regrading my mental and physical health, doing stuff when I am feeling alright

[hidden atlas]

Some days are good days waking up early

You wake up with so mch energy out of no where

[rapid sparrow]

I am owl. My energy is somewhere at middle of day, evening and sometimes during nights

Bless flexible schedule to be friendly for that

[hidden atlas]

im the opposite my energy is gone after 5pm or so

[rapid sparrow]

I care to spend my most productive hours of a day for work. When I am concentrated, full of energy, and in good mental and body health in general

Often enough that can make me many times more productive during certain times

Flexible schedule fully remote work with mind work, important to put effort, not showing imitation of effort

[hidden atlas]

yea

Ima go ahead to sleep now

maybe we can talk tmr

It was nice talking with you

[rapid sparrow]

Likewise

[mighty anvil]

pytube: I have a frustrating couple of issues with pytube: when I do from pytube import YouTube, Channel Channel is white and unseen (see img) ; I can use YouTube, but not Channel...anyone seen this and resolved? https://i.imgur.com/KSeFbmn.jpeg
I have tried installing pytube AND pytube3. I have tried removing the cache and installing/deinstalling both
from pytube3 doesnt work
yet https://pytube.io/en/latest/user/channel.html seems to show I can expect to be able to use Channel functionality

[royal silo]

I am trying to backup some of my dotfiles to github. I initialized a local git repository in my home directory. How do I ignore all other files except those in a few selected folders in .config/ folder?

[rapid sparrow]

you want to ignore all files and folders in .config/
except some selected?

[brazen forge]

!ytdl

[rancid schooner]

Our youtube-dl, or equivalents, policy

Per Python Discord's Rule 5, we are unable to assist with questions related to youtube-dl, pytube, or other YouTube video downloaders, as their usage violates YouTube's Terms of Service.

For reference, this usage is covered by the following clauses in YouTube's TOS, as of 2021-03-17:

The following restrictions apply to your use of the Service. You are not allowed to:

1. access, reproduce, download, distribute, transmit, broadcast, display, sell, license, alter, modify or otherwise use any part of the Service or any Content except: (a) as specifically permitted by the Service;  (b) with prior written permission from YouTube and, if applicable, the respective rights holders; or (c) as permitted by applicable law;

3. access the Service using any automated means (such as robots, botnets or scrapers) except: (a) in the case of public search engines, in accordance with YouTube’s robots.txt file; (b) with YouTube’s prior written permission; or (c) as permitted by applicable law;

9. use the Service to view or listen to Content other than for personal, non-commercial use (for example, you may not publicly screen videos or stream music from the Service)

[brazen forge]

tl;dr: we're not gonna help you with that

[royal silo]

nevermind, I instead settled for read-only bind mounts (i bind mounted each needed config folder in another directory then did git init there) to get away from this

[mighty anvil]

1. Is there a better place to ask my question? I've tried here and Unofficial Django. 2. How can I see my recent posts (on any server) I quite often lose track of what I have posted where!

[mighty anvil]

update: Looking all over the web I'm beginning to think pytube is mostly unmaintained after a YouTube structure change with shorts etc. Also now the basic functionality gives me HTTP Error 410: Gone , eg video_url = "https://www.youtube.com/watch?v=BDgsF2i6nT0" yt = pytube.YouTube(video_url)
this is helpful https://stackoverflow.com/questions/68680322/pytube-urllib-error-httperror-http-error-410-gone but seems like its not stable these days.

What is anyone else using to interact with Youtube Playlists and Channels (I don't need to download vids, I just want the easiest way to get playlists and thumnails and channel latest vids including shorts)

[real hedge]

Does Python have a way to auto reload the browser after changes on file system?

[rapid sparrow]

#tools-and-devops message repeated offence. Ignoring warnings. <@&831776746206265384>

[sacred nacelle]

I'm working on a package called ess which has a number of subpackages. We are currently in the process of splitting those up into separate repositories to help with maintenance. And we are considering making ess a namespace package to keep the existing structure from the user's perspective. However, we noticed that PyPI does not allow the name 'ess'. (But we have so far deployed to conda and it is fine with the name.) So my question is, can it cause problems if we have a namespace package with a name that is forbidden by PyPI? We would call the various subpackages, e.g., esssans for ess.sans. And we might have a meta package essdr (or similar name) that can be used to install the various subpackages with, e.g., pip install essdr[sans].

[indigo zenith]

Python wouldn't know anything about "the browser" unless it's a Selenium session. You probably need at least a little bit of JavaScript to do whatever you think you're trying to do

[real hedge]

Thanks, after giving it some thought I think running a node server is the ideal choice for now

However, I'm curious about other solutions not requiring node.
SSE or websockets could be used in python to detect file changes and send a SSE/WS event to the browser to perform a reload

What I'm creating is a JS framework like react.

My JS framework doesn't need node to compile and has zero dependencies! 👀

[rotund wyvern]

Hey folks, is anyone messing monorepos with multiple pypi packages being pushed? I'm wanting to pursue this via rye actually but i might be a bit in the bleeding edge.

[strong bison]

In WSL, whenever I type in this command:

I always get this:
[+] Building 0.2s (2/2) FINISHED                                                                         docker:default
 => [internal] load build definition from Dockerfile                                                               0.1s
 => => transferring dockerfile: 2B                                                                                 0.0s
 => [internal] load .dockerignore                                                                                  0.1s
 => => transferring context: 2B                                                                                    0.0s
ERROR: failed to solve: failed to read dockerfile: open /var/lib/docker/tmp/buildkit-mount3246307496/Dockerfile: no such file or directory```

how can i fix this issue

[rare anvil]

does anyone know what I need to do to configure the launch.json used for the vscode test debugger? (the one used when I click the second button on the right there)

I need to make justmycode into false

[brazen forge]

in the Debugger pane in the left sidebar, click on the gear icon beside the launch configuration dropdown at the top

that'll open the .vscode/launch.json file in your project where you can configure the Just my code setting

[brazen forge]

do you have a file named Dockerfile/Containerfile in that directory? if there is, do you have a .dockerignore as well?

[wide vessel]

hi guys

I have problem with my VSC someone help me ?? Ithx

[rare anvil]

Yeah but those are just the launch.jsons for regular debuggers u use on ur own right? its not the debugger used when u debug a test with vscode

Like I've tried changing the justmycode in my launch.json, but it isn't taking for the vscode test debugger

In fact even with an empty launch.json, debugging still works. how can I change launch.json config that vscode uses?

[indigo zenith]

#❓｜how-to-get-help

[hollow bay]

So, i was searching about making a python library to a specific chemical engineering work. is this the right chat to talk abou it?

[thorny shell]

maybe, if your questions are about which tools to use to help you

[rapid sparrow]

good enough. #software-architecture is slightly better though depending on which questions u will be asking

[thorny shell]

but if you're, e.g., wondering how to structure it, maybe

[hollow bay]

nice, i'll ask it there. Thank you guys!

[willow pagoda]

continuing discussion from #software-architecture message since here is more appropriate for package management, i recommend reading:

• the Python Packaging Tutorial for the typical packaging workflow
• Setuptools User Guide
• Declaring project metadata for a list of all keys you can add to your configuration

in short, pyproject.toml is the newest standard for declaring project metadata, and setup.py and setup.cfg are the older methods specific to setuptools

Setuptools automatic discovery mentions two common project layouts:

• src-layout

  project_root/
      src/
          my_library/
              __init__.py
              ...
      pyproject.toml
  

• flat-layout

  project_root/
      my_library/
          __init__.py
          ...
      pyproject.toml
  

there are arguments for one vs the other, but personally i stick with src-layout and use editable installs during development

this is what a setuptools-based pyproject.toml might look like: ```toml

pyproject.toml

[build-system]
requires = ["setuptools"]
build-backend = "setuptools.build_meta"

[project]
name = "my-library"
version = "0.1.0"
dependencies = ["numpy>=1.26.0", "sympy>=1.12", "matplotlib>=3.8.0"]

description, readme, license, classifiers, etc.after this, organize your scripts into a [package](<https://docs.python.org/3/tutorial/modules.html#packages>) following one of the above layouts, then adjust your imports as necessary (e.g. `from my_library.submodule import function`), and do an editable install so you can test the package:sh

/project_root $ pip install --editable .```
as per the tutorial you can use build and twine to build and publish your library, and for build systems other than setuptools they may have a command-line tool that can do this for you like poetry

some additional links beyond those are:

• MANIFEST.in, if you have any non-python files that your package should include
• Console Scripts, if you want to register some function as a command-line program for convenience
• PEP 440 and SemVer for version naming conventions (SemVer's major.minor.patch scheme is the most important to know, everything else you should refer to PEP 440 since both documents are not entirely compatible)
• berconpy if you want an example of my project structure

[hollow bay]

oh my, i thought it was a little bit simpler

but let's go, i'm gonna try it

thanks guys

[rapid sparrow]

it is more simple than it looks 😅

[tranquil phoenix]

I'm in the process of researching code-signing certificates. My goal is to sign an application that has versions for Windows, Mac OS, and Linux. What are some good cross-platform code-signing certificate providers that work for signing Windows, Mac OS, AND Linux applications?

[mighty anvil]

Oh, I see the warning now, apologies. Finding Discord quite confusing- where is there a list of all my messages, regardless of the server and messages that mention me?

[rapid sparrow]

there is Inbox that shows everywhere where u a replied across all servers

[fathom veldt]

Hello can i import libs from another file but when i use class or def?

[indigo zenith]

Maybe not a good practice but technically possible, sure

[latent bison]

Recently my system's antivirus/security software apparently started to block > pytest activities and cause some warning message to be posted in my Powershell window. From the warning messages it seems to do something with accessing cache file/folder. Is this normal? Is there something "new" occurring with pipenv and or pytest ?? Thanks!

[rapid sparrow]

😅 use Linux. Best OS to develop web applications that run on Linux servers anyway.
Otherwise, specify which cache folders.
as far as i know, folders of cache like __pycache__ should be freely deleteable at any time. (or for that matter u can even utilize work around of running python without cache)

[rapid sparrow]

Is it possible to turn on logging level and having it seen for specific smth at DEBUG level
without turning DEBUG for every single third party library?

[astral apex]

https://github.com/vipyrsec/dragonfly-mainframe/blob/main/src/mainframe/server.py#L78-L84
Like this?

You can pass a package name to getLogger and then you can set whatever settings you want

[rancid schooner]

src/mainframe/server.py lines 78 to 84

# Disable uvicorn's logging
for _log in ["uvicorn", "uvicorn.error"]:
    logging.getLogger(_log).handlers.clear()
    logging.getLogger(_log).propagate = True

logging.getLogger("uvicorn.access").handlers.clear()
logging.getLogger("uvicorn.access").propagate = False```

[rapid sparrow]

it did not work

def configure_logger(log_level: str) -> None:
    root_logger = logging.getLogger("")
    root_logger.setLevel(logus.log_levels_str_to_int[log_level])
    ch = logging.StreamHandler()
    ch.setLevel(logus.log_levels_str_to_int[log_level])
    if settings.LOG_JSON:
        formatter = logging.Formatter('{"time":"%(asctime)s","filepath":"%(name)s","level":"%(levelname)s","content":%(message)s}')
    else:
        formatter = logging.Formatter('%(name)s:%(levelname)s: %(message)s')
    ch.setFormatter(formatter)
    root_logger.addHandler(ch)

    logus.my_lib_logger.setLevel(logus.log_levels_str_to_int[log_level])
    logus.my_lib_logger.addHandler(ch)

@pytest.fixture(scope="session", autouse=True)
def turn_debug_logging() -> None:
    loggers = [
        logging.getLogger(name) for name in logging.root.manager.loggerDict
        if name.startswith("lib_name") or name.startswith("lib_name") or name.startswith("lib_name")]
    for logger in loggers:
        logger.handlers.clear()
        logger.propagate = True
        logger.setLevel(logging.CRITICAL)
        configure_logger("DEBUG")

setting everything to CRITICAL worked though

actually no

it made weird logging duplication for some reason

...python logging is difficult. sigh

[rapid sparrow]

i am heavily confused

    loggers = [
        logging.getLogger(name) for name in logging.root.manager.loggerDict
        if name in ["lib1","lib2","lib3"]]
    for logger in loggers:
        logger.handlers.clear()
        logger.propagate = True
        logger.setLevel(logging.WARN)

I expected this code to disable loggers...
...for some unknown reason it makes for remaining library amount of logging records duplicated proportional to amount of disabled libs

it is not really remotely expected behaviour, is it? 😄

I know!

i need to print threading, process name just in case. lets check if those are forks or not

[astral apex]

Huh
That’s weird
Yeah it’s a giant pain
I don’t really understand it either
Sorry I can’t be of more help

[rapid sparrow]

printed thread and process names. ID seem to match. yay to no fork bombing.
%(thread)d-%(process)d:%(asctime)s:%(name)s:%(levelname)s: %(message)s

139791544246272-834601:2023-10-11 23:10:34,930:lib_name.status:DEBUG: setting success in django storage >>> task_id=9901825a-06f2-43c5-b8c2-de7a895f9a99
139791544246272-834601:2023-10-11 23:10:34,930:lib_name.status:DEBUG: setting success in django storage >>> task_id=9901825a-06f2-43c5-b8c2-de7a895f9a99
139791544246272-834601:2023-10-11 23:10:34,930:lib_name.status:DEBUG: setting success in django storage >>> task_id=9901825a-06f2-43c5-b8c2-de7a895f9a99

that means it is pure logging smth 🤔

def configure_logger(log_level: str) -> None:
    root_logger = logging.getLogger("")
    root_logger.setLevel(logus.log_levels_str_to_int[log_level])
    ch = logging.StreamHandler()
    ch.setLevel(logus.log_levels_str_to_int[log_level])
    if settings.LOG_JSON:
        formatter = logging.Formatter('{"time":"%(asctime)s","filepath":"%(name)s","level":"%(levelname)s","content":%(message)s}')
    else:
        formatter = logging.Formatter('%(thread)d-%(process)d:%(asctime)s:%(name)s:%(levelname)s: %(message)s')
    ch.setFormatter(formatter)
    root_logger.addHandler(ch)

    logus.mylib_logger.setLevel(logus.log_levels_str_to_int[log_level])
    logus.mylib_logger.addHandler(ch)

for logger in [logging.getLogger(_log) for  _log in ["factory", "faker", "lib3"]]:
    logger.handlers.clear()
    logger.propagate = True
    logger.setLevel(logging.WARN)
configure_logger("DEBUG")

same bug with this

[rapid sparrow]

First time for everything

for logger in [logging.getLogger(_log) for  _log in ["factory", "faker", "lib3"]]:
    logger.handlers.clear()
    logger.propagate = True
    logger.setLevel(logging.WARN)

    configure_logger("DEBUG")

accidentally had configure_logger at the level of the loop 😅

that is the reason for logging duplication :/

[astral apex]

Oop

[rugged bobcat]

Hi guys,
I guess most of you know by now about the latest curl vulnerability. Which tools are you guys using to find the Python dependencies that have this vulnerability? I want to do a quick scan on multiple services. I was thinking about using Black Duck but if there's a better tool I am open to suggestions. I would prefer tools that integrate nicely with poetry and pipenv. Thank you!

[arctic python]

hi guys, I am currently starting to learn more about repo structure by reading some opensource repos.
there's one folder named /contracts I can't get a hang of it.
What does this folder usually storing? is it abnout types, protocols, etc??

[brazen forge]

could you share any repos you've seen this directory in?
(#software-architecture might be a better channel to ask this)

[arctic python]

yes sure. sorry if I get to the wrong chat 😄
this is one repo I saw: https://github.com/microsoft/promptflow/tree/main/src/promptflow/promptflow
and this as well: https://github.com/dbt-labs/dbt-core/tree/main/core/dbt

tried to read from their "how to" doc but seems to be leaving this folder out. So I can't really quite get what is their organisational purpose.

[brazen forge]

those looks like dataclasses (which might also be DB models)

[arctic python]

so it's more like a datastructure that the repo want contributors to follow?

thanks 🙂

[brazen forge]

kinda, those are structs that are used across the project

[arctic python]

thank you

[tawdry needle]

these are "data contracts" used in various parts of the application. it's a little idiosyncratic to use it as the folder name, but it's not bad. you might also see it called "models", as in the standard data models used throughout the application.

[arctic python]

ah that makes sense 🙂 thanks yeah. I saw /models more than contracts probably because it closely related to Pydantic/ ORM.

[tawdry needle]

right, i don't think that distinction is always clear, but maybe it's established practice to distinguish "external data models" and "internal data contracts" in some areas. i don't see those two things as particularly distinct concepts.

imo a contract is something that applies to the interface between two functions or components, consisting of the model(s) that are accepted and returned from that function, as well as pre- and post- conditions. so "contracts" isn't my favorite name for what you see in the dbt repo, and i prefer "models".

[arctic python]

another related question, if I have some abstract protocols. Do I still put it here? or is there some folder name I should be using to indicate that?

[tawdry needle]

types.py, protocols, or types/protocols.py all could work imo. with types and typing also interchangeable

in the python stdlib, types is used for concrete types (like ModuleType) and typing is used for abstract types (like Generic), but i think that's partly because types long predated typing. i don't know what the names would be if we were to reinvent both of those things at the same time.

[quiet harbor]

Im writing this script to move my mouse (its not for antik afk if you want we can talk in a VC about what im exactly doing) inside games, my cursor moves when im tapped out but when i tap in games it doesnt move

sorry if this is the wrong channel i didtn found a better one for this

[sage escarp]

guys how do you ship desktop applications with python?

is installing python on the host computer mandatory?

[thorny shell]

In practice, yes.

There are tools that purport to turn your python code into a standalone executable, but I've never gotten one to work.

[flat lotus]

I have, it's quite easy. The majority of these tools just take the python executable, your app, the standard lib, and put them into a zip

[knotty osprey]

What's the best way to handle creating alembic migrations in a dockerized app? Having to spin up a container with an attached volume and a configurable command for the message seems like overkill, is it the only way?

it honestly seems easier to just manually make the version file and run it, is there any reason I shouldnt do that and manually fill out the previous/next variables?

[rapid sparrow]

if it was my small project but i worked with postgres?

i would have ran command in another terminal, docker-compose up postgres that raises it with forwarded ports for my host machine
and just run migrations commands locally (that i recorded into taskfile.dev). So final command would have been like task makemigrations

if it was my work project, i would have put commands to make migrations into docker-compose service of application. service would have been named makemigrations
colleagues only would have needed to run docker-compose run makemigrations for them to be generated (it would have raised db for them, ran application correctly and did everything for them)

[astral apex]

My teammate makes the migrations run when the Docker image starts -- https://github.com/vipyrsec/dragonfly-mainframe/blob/main/entrypoint.sh#L5
Which I think is dumb, I don't think the app should have an account that allows edit the schema

But I wouldn't generate migrations in Docker
I would always generate new migrations on my host

[rancid schooner]

entrypoint.sh line 5

python -m alembic upgrade head && python -m uvicorn src.mainframe.server:app --host 0.0.0.0 && exec "$0"```

[rapid sparrow]

i think making migrations should be one off command in https://taskfile.dev/ or as docker-compose service.
I would not be doing that in staging or prod

because generating migrations is like writing code... it should not happen in prod, it is part of dev env writing code for me.

i would have ensured with CI that they are correct for usage in staging/prod, that i made all migrations and they will work for our staging/prod db

i will be always git committing all my files of migrations from my dev machine

generating migrations in docker as long as docker is on my host is fine to me.

Docker is just another... more feature rich venv alternative to me 😁

docker running in staging/prod will not have command present to deal with migrations in any way automatically... at serious projects at least. Only at dev env.

[knotty osprey]

Thank you @rapid sparrow and @astral apex , this is a personal project so I'm not terribly worried about running the migrations (though I am trying to use this project as 12-factor practice). It's writing the migrations that has become cumbersome since I've just had to start up a container and run the alembic revision command. Taskfile.dev is new to me, I'm gonna take a look at that

[rapid sparrow]

if u need to raise container to run alembic revision.
U could define docker-compose yml

x-definitions:
  app: &app
    build:
      context: .
    volumes:
      - ".:/code"
    depends_on:
      postgres:
        condition: service_healthy
    environment:
      APP_DATABASE_HOST: postgres

version: '3.8'
services:
  postgres:
    image: postgres:13
    environment:
      POSTGRES_USER: postgres
      POSTGRES_PASSWORD: postgres
      POSTGRES_DB: postgres
    ports:
      - 5432:5432
    healthcheck:
      test: [ "CMD", "pg_isready", "-U", "${PG_USER:-postgres}" ]
      interval: 3s
      timeout: 3s
      retries: 5

  app:
    <<: *app
    command: bash

  makemigrations:
    <<: *app
    command: alembic revision --args

then creating migrations will be just docker-compose run makemigrations, it will create them for you in your project folder then. Without any extra commands. It will await postgres raising completion and then will run them

i prefer to utilize taskfile though (and using to localhost port forwarded localhost), not liking to work from within containers unless necessary (or making it as more easy to replicate dev env for other collegues)

[heavy knot]

I use PyInstaller and add a —onefile argument.

[fresh patio]

#bot-commands

[fierce vale]

How is this password genorator?

import random
op = ("y")
while op == ("y"):
cho = input(" 8 or 11 char:\n---------------\n ")
alphabet = ["a","b","c","d","e","f","g","h","i","j","k","l","m","n","o","p","q","r","s","t","u","v","w","x","y","z","[","]",">","<","/","!","@","#","$","%","^","&","*","-","_","1","2","3","4","5","6","7","8","9","0"]
y = ["a","b","c","d","e","f","g","h","i","j","k","l","m","n","o","p","q","r","s","t","u","v","w","x","y","z"]
print("---------------\n your pass is:\n---------------")

if cho == ("8"):
    a = random.choice(y).upper()
    b = random.choice(y)
    c = random.choice(y)
    d = random.choice(alphabet)
    e = random.choice(alphabet)
    f = random.choice(alphabet)
    g = random.choice(alphabet)
    h = random.choice(alphabet)
    fill = (a+b+c+d+e+f+g+h)
    print("   ", fill)


if cho == ("11"): 
    a = random.choice(y).upper()
    b = random.choice(y)
    c = random.choice(y)
    d = random.choice(y)
    e = random.choice(y)
    f = random.choice(alphabet)
    g = random.choice(alphabet)
    h = random.choice(alphabet)
    i = random.choice(alphabet)
    j = random.choice(alphabet)
    k = random.choice(alphabet)
    fill = (a+b+c+d+e+f+g+h+i+j+k)
    print(" ", fill)

print("---------------")

op = input("  again? y/n\n---------------\n       ")
print("===============")

[heavy knot]

why don’t you try using the string module

so instead of have to make a big ass list you can just do

alphabet = string.letters + string.digits 

• it’ll randomize and u can choose how long u want the password

[fierce vale]

I'll try. I didn't know how to

[fierce vale]

I can't get it to pull items out of the string. lists

[fierce vale]

I figured it out

import random
import string
e = int(input("digits in pass:\n---------------\n "))
print("---------------")
a = string.digits + string.ascii_letters + string.punctuation
b = random.sample(a,e)
c = ""
d = c.join(b)
print(d)
print("===============")

I shortend it a little...

[fathom ledge]

Hi everyone!

Can I ask questions here?

[thorny shell]

yes, ideally if they're about tools and Dev Ops.

If it's a general "plz help me with my python homework" question, then #1035199133436354600 is the place

@fathom ledge ^^

[fathom ledge]

It’s about building a docker image but I already have a lead. Thank you!

[alpine horizon]

can a running process see if an ENV variable was added (globally, by a different process)?

in Ubuntu

[agile stag]

Hey guys

Traceback (most recent call last):                                              |
  File "/app/main.py", line 7, in <module>
    import requests
ModuleNotFoundError: No module named 'requests'

I get this message on my uBuntu console after running my Docker image

I’m not sure how. Maybe it’s because of my Dockerfile

My docker file

# Use the official Python base image
FROM python:3.9-slim

# Set the working directory inside the container
WORKDIR /app

# Copy your application files into the container
COPY . /app

# Copy the requirements.txt into the container
COPY requirements.txt /app

# Install project dependencies
RUN pip install --no-cache-dir -r requirements.txt

# Copy the rest of your project files into the container
COPY . .

# Set the command to run your Python application
CMD ["python", "main.py"]

[brazen forge]

what's in the requirements.txt?

also, you're copying your application/project files twice

the first COPY instruction is unnecessary and actually harming the layer caching

[brazen forge]

in Linux, under normal circumstances, processes cannot (and do not need to) change the environment of other processes

you need the SYS_PTRACE capability to be able to do this, IIRC

if you need to change some application config, it's better to communicate the change to the process via IPC or have the process read some config file in an interval or on some signal like USR1

[agile stag]

I didn’t know I actually had to write something in there. I thought all of my dependencies would automatically be containerized.

[brazen forge]

the layers of an image are stacked, so any change in a layer will mean rebuilding all layers after it
copying in the application files (which includes everything not ignored in a .dockerignore) will mean all layers after it are rebuilt because that generally includes code files that are changed often

anyway, that's not the issue at hand

[brazen forge]

^

[agile stag]

There’s nothing written in it

[brazen forge]

oh

well, that's where the requirements go

like requests

[agile stag]

I see

[brazen forge]

requests==x.y.z

best practice is pinning the packages to an exact version as well

…and the versions of the transitive dependencies as well

[alpine horizon]

Thanks for explaining this, I think I now know why Im having a certain bug

[agile stag]

Do user input programs not work in a Docker image?

[astral apex]

Generally, no

You can make Docker execute a CLI, and pass in the args you give it to the CLI

But most of the time containers will be running web servers or batch jobs that don't need input

[agile stag]

All right. thanks for the info

[rapid sparrow]

They can work if necessary. It is unorthodox additional feature used sparingly depending on needs.
Potential options for usage...

you can hook yourself into container as

• docker exec -it container_id_or_name bash(or sh) if it is present
  • to get into shell and accessing some scripts from inside
  • useful for average python Django application to access manage.py
• docker attach --sig-proxy=false container_id_or_name
  • run commands directly in current stdin of a container and exit with Ctrl+P,Ctrl+Q combination (container needs to be launched opened for stdin + tty manipulations!)
  • useful for game servers like Avorion game server or Minecraft server
    docker-compose with opened stdin for attachment is launched with flags

    stdin_open: true 
    tty: true

• and the most perverted option, you can even run GUI applications in container. And access them with
  • forwarding your GUI into container
  • or just using remote desktop options

while first two are more or less used (especially the first one is very common in python and other dynamic languages world), the third one is extremely perverted option 😅

The easiest option to hook yourself into stdin of container directly, just to run it with tty/stdin enabled as docker run -it stuff
although docker-compose run smth, with written options for that is very straightforward too

[fathom ledge]

Is it normal for a docker image to take hours to build? Literally hours.

Everything else is building fast but it’s the venv with the problem.

I get the message “connection reset by peer” when pip downloads a dependency that needs find compatible with the and the rest and it can take about 30 mins to resume download and move on to the next requirements in the requirements.txt file

[rapid sparrow]

nope.

[rapid sparrow]

the most often reason for me to have that, i have turned on some kind of VPN. I prefer to disable them when using docker

any VPN works really badly with docker often enough

alternative reason could be that you are trying to install stuff not meant to be installed. And u have too weak PC to compile it
i remember some sentry versions require a lot of compiling building and should not be installed
Also installing psycopg2 is not prefered. better installing psycopg2-binary. Monitoring CPU consumption should be enough to confirm this theory

[astral apex]

https://github.com/letsbuilda/anubis/actions/workflows/docker-build-push.yaml

My container can build in like 30 seconds

Though normally it takes a bit over a minute

It doesn't heavily depend on what you actually do inside your container though

[astral apex]

Why would you have a virtual environment inside a container?

a dependency that needs find compatible with the and the rest

So you have incompatible dependencies that pip is trying to reconcile?

Fixing your dependencies would make it to where pip doesn't have to try to fix them for you, so you can skip that step entirely

[fathom ledge]

Oh, because I have airflow and my ML-pipeline in a venv. Some stuff that I use with tensorflow extended is not compartible with airflow

[fathom ledge]

Yes, that's why I have it separate. But I don't understand why the project compiled correctly in Pycharm but not when I try to build the image

FROM apache/airflow:2.7.2-python3.8

USER root

WORKDIR /ml-system

ENV PYTHONBUFFERED=1

RUN apt-get update && apt-get -y upgrade && \
    python3 -m venv /ml-system/tfx-pipelines/venv

COPY requirements.txt /ml-system/tfx-pipelines/venv/requirements.txt
    
USER airflow

RUN /bin/bash -c "source /ml-system/tfx-pipelines/venv/bin/activate"  \
    && pip install --no-cache-dir --verbose -r /ml-system/tfx-pipelines/venv/requirements.txt

RUN pip install --no-cache-dir "apache-airflow==${AIRFLOW_VERSION}"

I haven't even copied the files to the container :/

[fathom ledge]

I will try this, thanks a lot!

[fathom ledge]

Is this something that can be done in Pycharm? tensorflow extended installed all the dependencies for me and I didn't get to choose.

[dusky fjord]

Do you run a local instance of a large language model (llm) to help you write python code? Which flavor do you use? I've been running oobabooga (https://github.com/oobabooga/text-generation-webui) on a debian box and I'd like to run a 34B gguf model (preferably from https://huggingface.co/TheBloke) to help me code. I guess I'd use the llm mostly as a search engine replacement(? other suggestions welcome). If I'm not using python, then I'm probably using java.

What type of huggingface model do you like?

CodeLlama-34B-GGUF (Code completion. Infilling),
CodeLlama-34B-Python-GGUF (Python specialist. e.g. https://huggingface.co/TheBloke/CodeLlama-34B-Python-GGUF),
CodeLlama-34B-Instruct-GGUF (Code completion. Infilling. Instructions / chat.)

or some type of other flavor, e.g.

CodeFuse-CodeLlama-34B-GGUF or
speechless-codellama-34b-v2.0-GGUF?

I mainly code file/data manipulation, mathematical and machine learning type stuff. What model type would be good to choose?

[agile stag]

Apparently, there’s some networking involved when it comes to building a Docker image. I have a program that connects to my local mySQL server on my local machine.

I built a Docker image file for it, pushed it to my Docker Hub, went on my Linode VM, pulled the Docker image from my Docker Hub, ran the image and now the program can’t connect to the mySQL server on my Linode 🤔

[small jay]

Perhaps your MySQL server is blocking certain IPs?

[agile stag]

On my Linode? That’s a good observation. I changed the bind address to 0.0.0.0

[small jay]

No, the MySQL sever I mean

[agile stag]

That’s installed inside my Linode virtual machine right?

I have mySQL installed both on my local machine and virtual machine

[small jay]

Oh you're running both your application and linode on the same machine?

[agile stag]

Well I ssh into my Linode VM on my local machine.

[small jay]

Right, but is the MySQL server running on the Linode VM?

[agile stag]

That’s correct

[small jay]

What did you configure the target address of the DB to be in your application?

Also perhaps you could use docker compose to help make the networking easier

[agile stag]

The DB hostname is called ‘localhost’ in my program

It’s connecting to the hostname ‘localhost’

In my program which is in reference to the localhost on my local machine.

[small jay]

Can you show us the Python code that connects to your database?

[agile stag]

I probably can tomorrow

But every login credential is default on both mySQL servers

Password to connect to the database is exactly the same

Database names are exactly the same on both mySQL servers

[eager rose]

funnily, one of the projects i work on had CI run for about an hour each time 😩. but we killed the part that took that long, thankfully

[fathom ledge]

I still can’t figure out which ones are causing the problems, I even left it building overnight 😂.

Pip is doing something called “backtracking” to resolve dependencies but it can take a very long time and it’s a hit or miss.

Someone on Reddit recommended using conda for machine learning deployments to manage dependencies and venvs instead of what I’m doing.

I’ve been learning tons, that’s for sure!

[agile stag]

Now I’m wondering if a containerized application trying to connect to a mySQL server on a machine is even practical.

Maybe I should containerize the mySQL server as well as the application? 🤔

Maybe my program is within the realms of the container and there’s no way for the application to connect to anything outside of the container

[astral apex]

It is

[astral apex]

You can

[rapid sparrow]

You lack knowledge of docker networking yeah

[agile stag]

I’m trying

[rapid sparrow]

Any ran docker creates by default bridge sub network within which is runs. It is assumed other containers will join same network name for sharing network access between each other by DNS names

If u want to access localhost present MySQL, simple solutions could be

1. binding container to network host. It will disable bridge functionality

[astral apex]

I hate Docker DBs
Mostly because I didn’t understand what I was doing and lost some data

[rapid sparrow]

2. or running MySQL binded to 0.0.0.0

It will listen to all networks then

[agile stag]

This is what I did.

[rapid sparrow]

3. or u can just run MySQL in container that joined same network. P.s. all containers defined in compose, share one network by default (equal to protect current folder name + default suffix) and can address each other by service name as DNS hostnames

[agile stag]

I still can’t connect to the db. I do have the hostname named as ‘localhost’ and the program is written for it to connect to a server named ‘localhost’.

Since it’s inside of a container, I’m thinking the program is trying to connect to the localhost of the Docker container instead of the mySQL server on my Linode VM.

[rapid sparrow]

May be issue in that u create docker subnetwork after MySQL was already binded prerhaps. If u would have created docker network in advance, MySQL would have auto discovered it during binding operation

[agile stag]

Thanks for the information. I can’t wait to take a another jab at it and resolve this.

Do you know if I can configure the network in my container on Docker HUB?

[rapid sparrow]

Start with simple solution of just launching container in network mode host. It will put it into host default network and u will find your localhost binded db easily by localhost hostname

[rapid sparrow]

No idea what is Docker HUB. I am a simple man that uses Docker CLI at Linux to access Docker Engine Server demon at same Linux host. No VMs involved

[agile stag]

Docker HUB is a repository for Docker images

[rapid sparrow]

Ah, that one. Capital letters confused me. Thought it is program name

[agile stag]

i build my Docker image from my local machine, push it in there and then pull it out from my Linode VM

[rapid sparrow]

There are no containers in docker hub. U store images at docker hub
Containers get created on image start.
U can launch multiple containers from same image
Network params are defined on container start

[agile stag]

Oh ok. Good to know.

[rapid sparrow]

Launching containers in host network has disadvantages btw. Containers launched in this way are no longer network isolated (from each other and from host). If they bind any port, it is directly same port gets binded at host. Port forwarding is no longer needed

Just as warning that simplest solution has its cons

So, u can get easily container start prevented if it tries to bind port that was already taken

[cerulean hollow]

Does anyone have experience with using the Terraform CDK for Python in Azure? If so, is it easier to use and read than HCL?

[rapid sparrow]

haven't used it, but i can assure you that HCL is highly readable

it is very consistent and static typed

you aren't promised to have not a shitty code written by infra devs in Terraform CDK

more comfortable, yeah, may be. But highly likely it will not benefit your long term maintanability

[agile stag]

How long does it take for a docker image to run that’s trying to connect to the host’s MYSQL server?

[tawdry needle]

This question is lacking context. But in general creating a connection to a database should be relatively fast, much less than 1 second unless there are some connection limits in place or the database server is under heavy load.

If you are just running mysql on your own machine and trying to connect to it from inside a container image, it should be just as fast as any other connection, you wouldn't see noticeable overhead. So if you are developing an application and it's taking several seconds trying to connect to the database, something has not been set up correctly

[agile stag]

My program creates a log file. It’s built into a Docker image. When I run the image and it creates the the log file, where does it put it and how can I access it?

[rapid sparrow]

Ideally u should switch logging to stdout only if u a able

Docker is not meant to be used with logging files. It is dirty possibility(u can), but not the right way compatible properly with docker ecosystem

[silent sedge]

Is there a metrics collection/storage solution that allows for syncing/backfilling devices that are only online some of the time?

Like a way to have a device record metrics, and sync it into a central database when it comes online, to be queried from Grafana?

Prometheus doesn't support backfilling at all, other TSDB I found don't seem to have a sync feature either

[agile stag]

Would it not be ideal to create a log file for a software that’s meant to be run automatically on a schedule?

[rapid sparrow]

docker already has its own logging management... so your own solution is potentially going to be awkward crutch reinventing, that is not going to be properly compatible with docker logging ecosystem around

[agile stag]

How do I access the Docker log file?

It appears the command for that is docker logs

[eager rose]

the best way to log using containers is just printing to stdout

[rapid sparrow]

1. yes, `docker logs containerIDorName is a regular way.
2. default logging can be further augmented with usage of logging options
   https://docs.docker.com/compose/compose-file/compose-file-v3/#logging
   https://docs.docker.com/config/containers/logging/configure/
3. docker ecosystem solutions can collect log for you into their databases and show with navigation in GUIs

• Loki / Grafana solution (raisable as single docker-compose)
• or just running container directly in AWS with some means, lightsail, elastic beanstalk, ecs, or eks
• kubernetes solution for logging aggregation (Loki, PLG solution is again deployable as Helm) and etc
• we use at work datadog for logging collection from all docker containers running at aws ecs

Docker ecosystem logging solution support autodiscovery. They could be collecting just straight from all discovered containers automatically

[eager rose]

docker logs doesn't print out logs from a file, though. it's just the stdout isn't it?

[rapid sparrow]

kind of... docker still has logs written into files somewhere in his folder

so... so it is reading from file

[eager rose]

err, yeah, but not a file the user is creating

[rapid sparrow]

you can read limited history

[rapid sparrow]

yup

[eager rose]

to do that you would probably need to exec into the container and read it that way

[rapid sparrow]

{
  "log-driver": "json-file",
  "log-opts": {
    "max-size": "10m",
    "max-file": "3",
    "labels": "production_status",
    "env": "os,customer"
  }
}

https://docs.docker.com/config/containers/logging/configure/
You can make configurations to docker logging. to all default, or to specific one

[eager rose]

yeah

[rapid sparrow]

https://docs.docker.com/compose/compose-file/compose-file-v3/
at the level of docker-compose overrides are avaiable

[little yacht]

I need guidance to learn kubernetes, infra Management and efficient microservices communication. I have good experience with python.
can someone guide me where i can practice, i am confused.
if someone can list some resource / technologies where to focus that will be pretty helpful.
plz tag me when reply to this message.

[slim maple]

For kubernetes I think just k8s and helm should be enough, also tiltfile for local development 🤔

There should be some k8s books that explain the main concepts, and helm is just a templating/release management tool really

[rapid sparrow]

helm is nasty to use for personal needs. too much untyped fragile. I think it is great only if u use its results / distributing something as a package
i would have wished something better tbh.

[slim maple]

I mean, it depends on how you use it

[rapid sparrow]

https://registry.terraform.io/providers/hashicorp/kubernetes/latest/docs
There is an interesting alternative in just... doing kubernetes via terraform

or through pulumi https://www.pulumi.com/registry/packages/kubernetes/

or https://cdk8s.io/

[slim maple]

Can I run this locally?

[rapid sparrow]

provider "kubernetes" {
  config_path = "~/.kube/config"
}

Sure. u can use it locally

just feed it kube config, and it will connect to whatever u wish

check first page for all authorization choices

[slim maple]

Why would I use terraform for k8s if I just need to do some templating?

E.g. replace a couple of values, run a for loop, etc

[rapid sparrow]

1. statefulness. you can delete terraform code, and it will erase no longer needed resources absolutely without any left trace
2. terraform is consistent and type safe. You can declare in a good way that your code Requires integer values, or list of complex objects, or string or whatever, it will compile it correctly

[slim maple]

Aren't k8s manifests type safe?

[rapid sparrow]

terraform is great tool in team among many devs because it is ultra consistent.

[rapid sparrow]

when we speak about helm, we depend on helm to make it type safe, and how to treat required and complex values.

[rapid sparrow]

3. you can review changes by terraform plan, what is changed and will be changed! More quality code review that you are doing correct changes

[slim maple]

As of statefulness that's really not that good of an upside, since you mostly use k8s to deploy your applications

[slim maple]

Same with helm or k8s manifests?

Just use git

[rapid sparrow]

no no. When we use Terraform... we can compare how our next change is compared to previously applied state, comparing to what is already running in prod

git is not answer to everything, it only shows diff to previous its commits/tags (we can't be always responsible and rely on that nobody made manual changes)

[slim maple]

You can just check what commit you applied before that, in gitlab you could look up all the deployments for example

[rapid sparrow]

(we can't be always responsible and rely on that nobody made manual changes)

there are always manual changes drifting around.

[slim maple]

It seems that it can be useful for cluster components but for apps I'd prefer helm

[rapid sparrow]

seeing diff in to what is Really running in prod is very valuable

[slim maple]

Plus that's what probably 90% of people use

[slim maple]

What do you mean?

[rapid sparrow]

someone could have toggle something via GUI

updated already something to other thing

changed smth. and etc

[slim maple]

Doesn't sound like a tool problem to me

[rapid sparrow]

terraform will build plan in comparison to real state currently running

[slim maple]

Solution is as simple as "don't change anything manually"

[rapid sparrow]

there will be always manual changes. Just from person experimenting in staging to introduce new infrastructure (even as a code in some PR)
Or someone did not deploy smth

there is always... state.... problems to what is really now

terraform allows to verify, your plan was deployed and clear now

and see that thing you deploy is next, going to change things compared to prod things now

[slim maple]

^

[rapid sparrow]

this is idealistic view that is not scaling onto big organization with a lot of stuff

[slim maple]

Ok

[rapid sparrow]

we try to achieve at this point... in 99% we do things as a code

but there is always 1% of stuff that can be done due to emergency or experiments or whatever

[slim maple]

Experiments should probably not be done on a live cluster

[rapid sparrow]

mostly i have problem with someone applying stuff from pull request not manually

but PR was not merged yet

but i discovered it only during terraform plan

[rapid sparrow]

Also, not all stuff is locked in versions always... that makes terraform plan building changes just from having downloaded newer stuff
What will be changed just from you applying same code :/ (helps to lock stuff to versions if possible)

[slim maple]

That's just not useful for me

[rapid sparrow]

well... those are problems for a lot of infra

for small stuff it can be not a problem