rules_python VS dbx_build_tools

What's SV?

I honestly don't know why anyone would use that... as in what does Bazel do better than virtually anything else that can provide this functionality. But, I used to be an ops engineer in a big company which wanted everything to be Maven, regardless of whether it does it well or not. So we built and deployed with Maven a lot of weird and unrelated stuff.

Not impossible, but not anything I'd advise anyone to do on their free time.

Specifically wrt' the link you posted, if you look here: https://github.com/bazelbuild/rules_python/blob/main/python/... it says that only pure Python wheels are supported, but that's also a lie, they don't support half of the functionality of pure Python wheels.

So, definitely not worth using, since lots of functionality is simply not there.

Regarding bazel, the rules_python has a py_wheel rule that helps you creating wheels that you can upload to pypi (https://github.com/bazelbuild/rules_python/blob/52e14b78307a...).

If you want to see an approach of bazel to pypi taken a bit to the extreme you can have a look at tensorflow on GitHub to see how they do it. They don't use the above-mentioned building rule because I think their build step is quite complicated (C/C++ stuff, Vida/ROCm support, python bindings, and multiOS support all in one before you can publish to pypi).

Python support in Bazel now looks more promising with `rules_python`: https://github.com/bazelbuild/rules_python

`rules_go` to my understanding is great too.

Over years, Bazel is not as opinionated as before, mostly because adoptions in different orgs force it to be so.

I have personally run converted build systems to Bazel, and use it for personal projects as well.

Bazel 1.0 was released in October 2019. If you were using it "a few years ago", I'm guessing you were using a pre-1.0 version. There's not some cutoff where Bazel magically got easy to use, and I still wouldn't describe it as "easy", but the problem it solves is hard to solve well, and the community support for Bazel has gotten a lot better over the past years.

https://github.com/bazelbuild/rules_python

The difficulty and complexity of using Bazel is highly variable. I've seen some projects where using Bazel is just super simple and easy, and some projects where using Bazel required a massive effort (custom toolchains and the like).

load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive") #------------------------------------------------------------------------------ # Python #------------------------------------------------------------------------------ # enable python rules http_archive( name = "rules_python", url = "https://github.com/bazelbuild/rules_python/releases/download/0.2.0/rules_python-0.2.0.tar.gz", sha256 = "778197e26c5fbeb07ac2a2c5ae405b30f6cb7ad1f5510ea6fdac03bded96cc6f", )

> hermetic == sandbox which only contains declared deps

Yes, exactly. It also can include other deps like networking. A perfect hermetic build will (for most of the build process) not have access to the network.

> I assume bazel implements this via containers?

Sort of. Bazel's code predates "containers" by ~6 years. Bazel is the build system used at Google (sprinkle on a large amount of asterisks here).

> Also, IMO, this only provides a false sense of safety unless you archive absolutely everything that's inside the sandbox.

Yes, that is what a Bazel monorepo does. You can also include your language's runtime and standard library as part of the build process. See Dropbox's repo for more details on how they do this with Python: https://github.com/dropbox/dbx_build_tools

Some more details about what Google does are available here: https://opensource.google/docs/thirdparty/#

Essentially, everything is built from source (more asterisks here). Since everything is built from source using Bazel you can guarantee that at a certain view of the source repository you can build a reproducible copy of all software in the repo (more asterisks).

One other nice property is you can guarantee that all software in the repo is using a compatible version of each dep: https://opensource.google/docs/thirdparty/oneversion/

The downside is that if you want to upgrade a library that has a breaking change this requires you update a lot of code. The upside is that doing that isn't actually that hard with the tooling available and you can guarantee that you can link together any two libraries in the repo (more asterisks) and have the build succeed without any dep version conflicts. Since you can do this some langauges with runtimes (like JVM language) can see massive speed boosts because they only need to compile changed libraries and can run a `java -cp ... your.jar` to load up deps. You can check out rules_docker for some examples of how this can be used.