I love Bazel. It's one of those satisfying tools where computer science algorithms and data structures fuse with good software practice to create something that's satisfying on many levels. It knows how to minimally rebuild things. It knows how to hermetically build things in a repeatable way. It's existence made Merkle trees make more sense to me.
I hate Bazel. While trying to figure out building a mix of Python and C++, I've wandered endlessly trying to sort out how to get it working. I dislike the odd mix of a large amount of documentation and a large amount of documentation that doesn't explain what I'm hoping to understand. Python in particular is somewhat inscrutable in that when trying to build a pytest test, it seemed to repeatedly break the hermeticity I was hoping I had. Its complexity led to me repeatedly break things that I thought that I'd fixed. The diff to fix what I thought was the right system is largely one made of witch incantations over a cauldron made by dumping an existing project in and skimming stuff off the top that seemed passable.
So, on that note, thanks to my savior: the Selenium project. I appreciate it as a testing tool (one which I find enjoyable), but more now as a multi-language Bazel project (including Python) whose example I can follow. FormaK's build structure now heavily borrows from Selenium and it's held up to some minor expansions. With the Selenium-inspired structure (and essentially standing on the shoulders of Selenium as my build tools "team") I can now proceed forward with the more FormaK-y aspects of the project:
- The scikit-learn-ification of FormaK to adapt existing machine learning tooling - Design