abseil-py VS awesome-python

> Sounds a lot like Nix's philosophy of totally sandboxed/declarative package builds

They are similar but I think of Nix as being closer to Dockerfiles than to Bazel. Nix describes how to compose and build things, of multiple build systems, in a unixy environment. It sandboxes things and runs commands in those things like `make install` and then pulls the files those commands generates out and uses them for other things. This is very close to how people use Dockerfiles but very different to how someone would go about using Bazel. Bazel's main selling points are that it allows you to describe how your software is built by using application level concepts. Bazel has `library`, `binary`, and `test`. Regardless of what the language we're talking about any two programmers will probably know what these high level concepts mean and, due to conventions in rules, any developer - regardless if they have experience in the language in question - can understand what is going on.

Bazel is also much finer grained than something like Nix. For example, the NixPkg for abseil-py is defined here: https://github.com/NixOS/nixpkgs/blob/nixos-21.05/pkgs/devel...

The BUILD files for abseil-py are found in the repo. Something you'll notice is that there isn't a single build target for all of abseil-py. Instead, each python module or "library" has it's own target: https://github.com/abseil/abseil-py/blob/master/absl/BUILD#L...

This allows for some cool stuff when deploying your software. Suppose you import just the flags module from this repo. That means that your built `py_binary` artifact will only contain just the modules that you're actually importing. You can also see the dependancy graph for all of your source code by talking to Bazel's query API. Find a bug in a library? You can now instantly query every single `binary` that those source files went into without false positives. That's just an example of something you may want to do but really you now have a way to query the dep graph for all languages and tools so the sky is the limit.

> instead of needing a monorepo with every piece of software you can just grab a reproducible build from nixpkgs

You also don't need a monorepo for Bazel to work. You can use `http_archive` or `git_repository` to load specific WORKSPACEs. For example: https://abseil.io/docs/cpp/quickstart#set-up-a-bazel-workspa...

Also bzlmod is being worked on (a package manager that automates this stuff). This also has provisions for running all of the `_test` targets on deps after they're compiled by your toolchain and on your hardware. This validates that all importing and compiling worked as expected (which most build systems can't accommodate).

Python: https://github.com/vinta/awesome-python

6) Awesome Python: Embrace the power of Python with this extensive collection of awesome libraries, frameworks, resources, and software. Whether you're a seasoned Pythonista or just starting your journey, this repository is your ultimate guide to maximizing the potential of this versatile language. (https://github.com/vinta/awesome-python)

- https://github.com/vinta/awesome-python

🎇 Repository Link: Awesome Python

Explore here.

I know of curated lists like https://github.com/vinta/awesome-python but they are nowhere close to alternativeto.net in terms of information (relations) and community involvement.

https://github.com/vinta/awesome-python

There are a lot of python packages for engineering and scientific applications (as well as other applications in general, and, thanks to the inherently collaborative nature of free software, they are only growing in quantity and quality. Many MATLAB toolboxes already have Python equivalents.