• 51 Posts
  • 1.93K Comments
Joined 1 year ago
Cake day: September 1st, 2023
  • onlinepersona@programming.devtoProgramming@programming.dev•Is Python's tooling incredibly difficult, or am I just stupid?
    46·
    2 days ago

    Difficult? How so? I find compiling C and C++ stuff much more difficult than anything python. It never works on the first try whereas with python the chances are much much higher.

    What’s is so difficult to understand about virtual envs? You have global python packages, you can also have per user python packages, and you can create virtual environments to install packages into. Why do people struggle to understand this?

    The global packages are found thanks to default locations, which can be overridden with environment variables. Virtual environments set those environment variables to be able to point to different locations.

    python -m venv .venv/ means python will execute the module venv and tell it to create a virtual environment in the .venv folder in the current directory. As mentioned above, the environment variables have to be set to actually use it. That’s when source .venv/bin/activate comes into play (there are other scripts for zsh and fish). Now you can run pip install $package and then run the package’s command if it has one.

    It’s that simple. If you want to, you can make it difficult by doing sudo pip install $package and fucking up your global packages by possibly updating a dependency of another package - just like the equivalent of updating glibc from 1.2 to 1.3 and breaking every application depending on 1.2 because glibc doesn’t fucking follow goddamn semver.

    As for old versions of python, bro give me a break. There’s pyenv for that if whatever old ass package you’re installing depends on an ancient 10 year old python version. You really think building a C++ package from 10 years ago will work more smoothly than python? Have fun tracking down all the unlocked dependency versions that “Worked On My Machine 🏧” at the start of the century.

    The only python packages I have installing are those with C/C++ dependencies which have to be compiled at install time.

    Y’all have got to be meme’ing.

    Anti Commercial-AI license

  • onlinepersona@programming.devtoLinux@programming.dev•KDE's New Distro: Btrfs-Based, Immutable Linux OS, with Flatpak and Snap
    14·
    6 days ago

    My immediate thought was: why not NixOS as a base? Building KDE is such a nightmare that if they had to deal with it themselves on NixOS, it would help them clear up their dependencies. Right now it’s such a big mess of unnamed and implicit dependencies that exposing it to the team would also show them how to cut down on them.

    My hope was also that if the KDE team were invest in a NixOS offshoot, that the OS would finally get proper GUIs or integrations into existing GUIs like Discover (why not Diskover?) Or the system settings and other config management.

    But, to be fair, I could understand if they considered it, took one look at the documentation and noped out.

    Anti Commercial-AI license