Cycling, Liverpool FC, Python, AI/ML, Emacs, GNU/Linux, DevOps, Pink Floyd -- not necessarily in that order.
elpy is a great minor-mode for python-mode (major-mode) in Emacs, but managing multiple venvs (virtual environments) has always been tricky as pyvenv (that elpy uses) doesn't have a way to have buffer-local venv (virtual environment) -- any venv activated using e.g. pyvenv-activate/pyvenv-workon is (process) global.
I use blacken-mode (uses black) to auto-format Python code on save -- it works great but not much helpful while dealing with strings. For example, if I want to set a max limit on columns (characters) per line when writing a string, black is of no help. So I needed to come up with something in emacs that does the auto-format of strings -- set a max limit on-the-fly on number of columns per line for writing strings.
Emacs has a global fringe-mode for fringes, which I always keep disabled as it takes a bit of space in the frame and I like to start typing the first character on a line to the very left. But there are some cases where fringes are quite useful e.g. when flycheck shows different checker warnings/errors in code, it uses fringes (if enabled) to have a nice arrow to indicate the line(s) with the warnings/errors -- great to have a quick glance at everything.
The eww browser of emacs (GNU Emacs 29.3) uses gnutls-cli to setup TLS connections to web servers. And on my Linux Mint 21.3 (based on Ubuntu 22.04), it wasn't installed; so whenever I was trying to visit any HTTPS site, the following error was shown (truncated):
Opening TLS connection with ‘gnutls-cli --x509cafile /etc/ssl/certs/ca-certificates.crt -p 443 ’...failed
Recently, I had a sort of unusual requirement to get the last modified file (tracked by git) in a certain directory for a specific branch. I tried to look for any existing porcelain or plumbing command in git that can be used to advantage, but couldn't find anything, so had to stitch up something combining several utilities.
Here's how it looks eventually:
Oftentimes, I need to record my webcam and microphone, so I created a tiny snippet to automate this on my GNU/Linux system (ffmpeg devices are different in different operating systems as they depend on the kernel modules, one can do ffmpeg -devices to find the available devices on their system). One straight-up ffmpeg command to record webcam and microphone is:
In Linux, V4L (Video4Linux) dictates how a video-capture device (as character node) is created in the device tmpfs filesystem (devtmpfs specifically, mounted on /dev). The webcam devices are usually created as /dev/video<index> e.g. /dev/video0. Different udev rules are then used to give these devices locations in /dev based on their ID (in /dev/v4l/by-id/), bus path (in /dev/v4l/by-path/) and so on.
When working with torchaudio for audio-oriented neural network implementations, one of the very first things we do is to get the metadata of the audio input using torchaudio.info.
Panel is the perfect place in Linux Mint to put the launchers for programs/commands that are used most often. To create a launcher we need a .desktop file that describes some attributes of the program. Most installed GUI programs come with their launchers (.desktop files), but when we have a custom command/program of our own that we want to create the launcher for, the only thing we need to do is create a .desktop file.
tensorboard is a quite popular among deep learning practitioners for experiment tracking. I also use it often, but what was bothering me for a while was to start the server (using tensorboard command) on one emacs buffer every time I want to track something. It gets tedious very quickly considering restarts as well. So, I've written a tiny elisp recipe to start/restart tensorboard directly from the (Python) file buffer (python-mode) in emacs.