Please don't break HN's guidelines by being snarky or putting down others' work in a shallow way. If you know more or have a different perspective to offer, try sharing some of what you know so we can all learn something!

1. The author of this post helped to create the Django framework and runs a successful Python consultancy.

2. Conda is not used as much as you might think... it's really only used within the data science community.

Being in the data science community myself, I prefer straight venv + pip to conda. It’s simpler for me to manage errors. I only use conda when I have to.

Yeah I don't know a single person who chooses to use conda.

Hi whalesalad, good to meet you! Now that you know me, you can never say that again anymore :-) Although, tbf, I only use conda for my machine learning related projects. I've tried using pip for that but was at risk of massive hair loss.

In the scientific community, there is a widespread "just use anaconda" message. Many people spend their entire lives inside anaconda, and equate it with Python.

Conda is ideal if you need to support Windows environments with large, compiled scientific programming libraries.

So it’s niche.

No. I used it for managing my Mac and Linux environments.

On the other hand, almost everyone I know who does scientific python in grad school loves miniconda...

And anyway, you should tell people to install mini conda, not conda, unless they know they'll need everything it includes by default.

Yet another reason to avoid it entirely.

1. Argument from authority doesn’t mean anything to me. I also don’t believe creating Django or running a Python consultancy endow someone with especially useful opinions of Python packaging tooling. (Not that the author isn’t knowledgeable, just you seem to think there’s an A implies B relationship between those two items and having good opinions about Python packaging, and there’s not).

2. Conda is quite widely used outside of data science. It’s for example part of Anaconda enterprise offerings used by huge banks, government agencies, universities, etc., on large projects often with no use cases related to data science. Conda itself has no logical connection with data science, it’s just a package & environment manager.

In each of my last 4 jobs, 2 at large Fortune 500 ecommerce companies, conda has been the environment manager used for all internal Python development. Still use pip a lot within conda envs, but conda is the one broader constant.

> huge banks, government agencies, universities

> large Fortune 500 e-commerce companies

Sorry, but argument from authority doesn't mean anything to me.

In all seriousness though, you literally did not provide any logical reasons to think conda is better.

Giving a counterexample is not argument from authority. I did not respond to the parent comment to discuss any feature of conda, only to dispel the wrong claim that only mostly data science projects rely on it.

I love when people claim to not care about arguments from authority. "Sorry doc, I don't care about your medical degree or years of experience!"

Yes you do.

<20 minutes later> "Sorry, your dependencies could not be resolved"

Unless there's a bug in the dependency solver, isn't this a good thing? conda is preventing you from installing incompatible packages.

It was bad when it took minutes to figure that out. It has gotten better though.