Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

As a non python native, I sigh whenever I have to install anything in python. There's always a dependency I can't find and never know why. I never understand why the dependencies aren't just resolved by an application and local to that application. I mostly end up with something working. Sometimes I just create a VM for that single application.


I once really desperately needed a Python tool. Wasted an hour and getting more and more enraged trying to install it, overwriting hell-knows-what and ruining god-knows-how-many other programs in the process. Eventually it did occur to me to just use Docker and that did it. <facepalm>

Every time I am forced to do something with Python I end up hating it. It seems those who write it are just not interested in anyone who doesn't have a complete dev environment setup for it. :(


> It seems those who write it are just not interested in anyone who doesn't have a complete dev env...

And that's probably what makes Python so popular, and so great. I get it. It's attractive to people who aren't developing software for other developers.

But what you described is surely the result of that atmosphere.


> And that's probably what makes Python so popular, and so great. I get it.

I don't follow because that's actually a problem. Imagine having to install Haskell compiler -- exactly one particular version of it with the exact right combination of flags (and there are hundreds of them) -- for each project... while also having the whole thing kind of compile from source.

And that for people who just want to do something like `brew install a_valuable_tool` and have no patience (and certainly not the time) to tinker with Python's dev environment just so they can install a tool they need.

So maybe I am misunderstanding but no, that's not what makes Python popular or great, at all. It makes it a legendary pain for anyone who doesn't code it every day.

If I can't do `brew install your_tool` but have to fiddle with the intricacies of your language's ecosystem then you have failed as a tool author.

Contrast this with Golang tools (disclaimer: I dislike Golang) which in the worst case scenario involve 1-2 commands to download and compile+install the tool but usually don't involve even that because Golang's process is self-contained and that makes it super easy for all package managers to distribute it pre-compiled for their platform.


The distro should be packaging software for end users, relying on the python version they maintain. The problem is that Apple isn’t doing that at all, and brew seems to have issues getting all formulas to converge on known-good dependencies.


Try Anaconda (or miniconda/mamba): like virtualenvs but reproducible, dependency pinning, and a great solver. All in one handy tool


Thank you.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: