Our python repository interacts with our C++ repository via a wheel that the C++ CI/CD exports. The wheel has enums, protobufs, and a few shared object files that are accessed from Python. This approach has been working pretty well so far, but upon the switch to pants, while the resolves based approach has it's multitude of benefits, one of the drawbacks is the lockfile churn each time someone needs to consume a custom wheel. Is there any guidance on how to avoid this? Any better approaches suggested? Thanks in advance!

Can you explain what you mean by lockfile churn when someone is consuming a custom wheel? I have had a similar situation to this, but didn’t observe this problem

Sure! Each time I use a custom wheel, I have to generate a fresh lockfile in the resolve that needs it. (my Python repo has multiple resolves)

Assuming "lockfile churn" means "pants updates every external dep in the lockfile to the newest version which satisfies your requirements", then I've also struggled with this problem (any time i add a dependency to my monorepo) I found a workaround and documented it here, if you're interested. It might work for you too