Emil Kreutzman ekreutz

Using 7zip to create encrypted archives

Let's use 7zip for strong AES256 encrypted archives with a password. The guide assumes you're on Ubuntu 22 or newer. Note: this creates .7z archives, which is not the same as the regular .zip!

# Install 7zip. Note: don't install "p7zip-full" or "p7zip" (it's an older version!)
# Source, version things: https://askubuntu.com/questions/1465853/difference-between-several-command-line-tools-provided-for-7-zip-compression-li
sudo apt update
sudo apt install 7zip

Convert OpenSSH private key to PEM file format

Assuming that your ssh private key is located at ~/.ssh/id_rsa. We will convert this file to the PEM format.

# Change dir to home directory
cd ~

# Make a copy of the private key and store the copy in the home directory
cp ~/.ssh/id_rsa ~/id_isa_copy

Naughty coroutines

Unexpected behavior when working with asyncio in Python. The behavior is likely intended, but intuitive.

import asyncio


async def fine_job(i: int):
    # A fine job that is well behaved without exceptions!

An antipattern in python asyncio error handling

Don't do this:

The try catch will catch any instance of SomeError, even if it's not emitted inside my_async_function. This happens if my_async_function runs for long enough so that SomeError is emitted elsewhere in the program, while this piece of code is still inside the try-except.

try:

How to trim empty lines in VSCode

What this will do:

Trim empty lines (remove all whitespace)
Trim whitespace at the end of code lines
Note: it will not remove any line at all, but simply trim trailing whitespace.

Steps:

Press Ctrl + F (Cmd + F on mac) to open the search and replace view.
Toggle the regex button .*.

Reset SERIAL counter for a PostgreSQL database table

For the table my_table with a SERIAL column my_id.

SELECT SETVAL(
  pg_get_serial_sequence('my_table', 'my_id'),
  (SELECT COALESCE(MAX(my_id), 0) FROM my_table)
);

Update `conda`

Or, how to update miniconda or Anaconda.

Simply run the following command:

conda update -n base conda

Pixel aspect ratios - a potential issue in object detection?

Updated: Jan 25, 2021

Quick write-up about pixel aspect ratios in video files, and how they might affect performance in an object detection pipeline. I found very little good information about this online, so I decided to write this to summarize my findings.

Tl;dr: scroll down to the "final solution" code, for how to deal with this issue, using Python/OpenCV.

	from numba import njit
	from numpy.typing import NDArray as array


	@njit
	def running_max(values: array, w: int) -> array:
	"""Fast O(n) running maximum.

	For large values of `w` this solution is 100x faster or more, than the naive version.
	"""

	from math import log10, floor
	from decimal import Decimal as Dec

	# Note: the method spec uses a union type float \| Dec introduced in Python 3.10
	# Remove the type definitions to make it work with earlier versions.

	def round_sd(x: float \| Dec, sd: int = 3):
	"""Round a value to a specified amount of significant digits `sd`.
	"""
	return round(x, sd - int(floor(log10(abs(x)))) - 1)