Especially if you’re around bleeding edge technologies and always after the latest fuzz you’ll have a hard time to get around Python and Docker. Don’t get me wrong – both are brilliant and do have a right to exist and many beautiful use cases. But lately I have seen so many “releases” that felt so “half baked” due to overuse of Python and Docker.
Really – nothing against Python. Some old school devs would still refer to it as a “scripting” language – which doesn’t do it justice due to the high level approach, countless modules and widespread use as applications. However if you come from a classic dev background and the word “compiling” still means a lot to you, you might have a different perspective. There are workarounds and tricks to compile Python – but honestly – it was never meant to be compiled and if you ever tried to compile Python scripts with loads of dependencies or value scmall binaries you get what I mean. If you’re a geek using Debian and wanna share your creations with other Linux users then no problem – you’ll barely find anyone who doesn’t have python installed. But for other OS like MacOS or Windows this often isn’t the typical end user scenario.
So what I often see when people have like a hundred .py scripts with like 150 dependencies and people are like “WTF, dude no idea how to run it!” is people starting to use Docker (Which I personally prefer over some runbooks).
While Docker has a beauty of it’s own, practically being a complete system that “just works”, I feel it’s not comparable to a classic binary. It sure beats weird instructions on how to install a hundred dependencies or runbooks that use package managers you don’t even have installed – but back to MacOS and Windows users – it’s by far no standard that these have Docker installed.
And even if they do – what you get is a giant monolithic blackbox that “somehow works” if everything went right. This is a bit like if you ordered a toaster and someone delivers a complete mobile home that has a toaster in the kitchen. Yes it does work and makes toast, but I sometimes feel that the beauty of efficient coding gets lost here. There are beautiful use cases when I feel Docker is the best solution but I think it often gets absused to put a complete mess of code (“Dude why did you install this module? Hmm yeah not sure it’s even necessary…” “Oh that .py? yeah nothing to do with it, I just copied some stuff from an old project…”) into a huge file that is anything but performant.
And yeah joke is on me – if it’s open source just rewrite it and release something better. I just feel the quality of stuff released lately has decreased. Is there still anyone left believing in “beautiful code”? /rant