Monorepo Build Systems. DAG vs. depth-first recursive, featuring sparse-checkout [video]

An edu piece on Google style monorepos - https://github.com/paul-hammant/google-monorepo-sim and video https://www.youtube.com/watch?v=L67ri_xe2oQ - showing how their Piper would work for slimming checkouts (git sparse checkout for the sim).

I go into the DAG-centric nature of Google's in my video, and that was one of the reasons for making it and the sim repo.

I did at the outset think that the Google way lead to faster builds. It could could be with parallel step execution and a build cache is setup, but I don't think it is significantly different in a vanilla serial mode.

Perhaps there's another benefit - slimming a monorepo to feed to a coding AI, in order to bring down tokens needed for a task.

I've WIP for a C# module, but I'm stuck on some fine-grained setup for 'vstest'. I also asked Google's Jules agent to migrate one of the depended-on modules to Kotlin, and it has been 24 hours now, with it assuring me that it hasn't stalled.