Buck VS pants

Hyper-large tech companies managing hyper-large monorepos using Bazel (google), buck (Facebook), please (thought machine), pants (Twitter, Foursquare & Square) enjoy them but also have a lot of resources devoted to running and maintaining it.

We use Buck company wide. Our packaging / deployment system, for example, expects to be given a Buck target to build, not a pre-built binary - I can’t just build my app with dotnet and upload it. While it is possible for a Buck target to be a simple bash command (i.e dotnet publish), doing so makes the target “opaque” - Buck wouldn’t have any knowledge of my app’s build graph so I’d lose many of the benefits it gives us (incremental cached builds etc.)

Oh excellent, then better (and more portable!) tools are available:

http://pants.build

https://ninja-build.org

https://buck.build

and, if you hate yourself: https://bazel.build

I’m feeling sorry for everyone affected.

Let’s hope that this isn’t going to impact Buck [0] too much. It’s one of the best things Facebook has ever made.

[0]: https://github.com/facebook/buck/tree/dev

Pioneered by tech giants like Google and Meta with tools like Bazel and Buck, monorepos are seeing widespread adoption across companies of all sizes and industries.

Buck has a http_file() that you can use this way, and it has first-class support for Java.

That's a good bridge into saying that we don't use pretty much any standard tooling. Our build system is Buck, we use Mercurial instead of Git, and the IDE of choice seems to be Visual Studio (although Android Studio is supported, with some custom plugins required).

Didn't know it was python based. Their github repo shows 90% of the code is java, only 2% is python. Regardless, C++ build systems should be written by people who are familiar with C++ and the specific problems they need to solve. That means the build system should be written in C++ or C and not in java, lua, or python. I certainly do not want to install python just to build my C++ programs in a CI environment, it just increases my build complexity and attack surface

The landing page of https://buck.build/ explains some of the most basic stuff. This Meta Engineering blog post summarizes some of the deeper optimizations possible through buck: https://engineering.fb.com/2017/11/09/android/rethinking-android-app-compilation-with-buck/

> C/C++'s header system with conditional inclusion

Wouldn't it be more accurate to say something like "older build systems"? I don't think any of the things you listed are "modern". Which isn't a criticism of their legacy! They have been very useful for a long time, and that's to be applauded. But they have huge problems, which is a big part of why newer systems have been created.

FWIW, I have been using pants[0] (v2) for a little under a year. We chose it after also evaluating it and bazel (but not nix, for better or worse). I think it's really really great! Also painful in some ways (as is inevitably the case with any software). And of course it's nearly impossible to entirely stomp out "genrules" use cases. But it's much easier to get much closer to true hermeticity, and I'm a big fan of that.

0: https://www.pantsbuild.org/

Does pants/bazel can help me?

Hyper-large tech companies managing hyper-large monorepos using Bazel (google), buck (Facebook), please (thought machine), pants (Twitter, Foursquare & Square) enjoy them but also have a lot of resources devoted to running and maintaining it.

Yeah there's definitely some alternatives out there. Pants is another one that has a lot of traction.

You should look into pex, or it’s parent build system pants. A PEX (Python EXecutable) file can package up all your code including dependencies and run on another machine of similar OS with just an available compatible interpreter.

We don't have the CD yet, but I think what I put in place counts as simple CI (even if incomplete)? Every push & PR trigger an azure pipeline, which runs pants. This install the dependencies from the lockfile, run some linters, uses DVC to pull the data necessary for tests, and run unit tests (mypy check is deactivated until I solve a weird error). Basically the same script runs on laptops cross-platform (one of us uses Max, one Ubuntu with GPU, one Ubuntu with CPU, the scripts runs on every platform). The only difference with CI is the installation of Pants and the gestion of Cache (needs to be downloaded in CI so it takes ~3min in CI versus 20 seconds on my laptop).

This has also been recently integrated in pants.

https://github.com/pantsbuild/pants/issues/13393