I've got a set of changes I'd like to contribute as a 5-part stack, each based on main:
- lint — adopt ruff + pre-commit, enforce in CI
- cloud engines — Livy / Spark Connect + AST-based multi-part name qualification (fixes a silent catalog-dropping bug; makes fuzzy column remap opt-in)
- cli —
lakebench CLI + profiles (fixes a token_env credential-dropping bug)
- tpcdi — TPC-DI benchmark port across six engines
- databricks — Databricks Connect engine
Each is independently green (ruff + tests). Do you prefer a base-chained stack of PRs, or sequential PRs against main (open #1, merge, rebase next)? Happy to do whichever fits your review workflow. I'll also call out two design questions inline (the Databricks constructor re-exec, and the W291/W293 lint ignore).
I've got a set of changes I'd like to contribute as a 5-part stack, each based on
main:lakebenchCLI + profiles (fixes atoken_envcredential-dropping bug)Each is independently green (
ruff+ tests). Do you prefer a base-chained stack of PRs, or sequential PRs againstmain(open #1, merge, rebase next)? Happy to do whichever fits your review workflow. I'll also call out two design questions inline (the Databricks constructor re-exec, and the W291/W293 lint ignore).