Hi all,

Reviving #868: https://github.com/apache/cloudberry/discussions/868 — renaming 
gp*/greenplum* identifiers to cb*/cloudberry* with aliases. The surface: CLI 
tools (gpstart, gpstop, gpconfig… → cbstart, cbstop…), catalog 
(gp_segment_configuration, gp_distribution_policy…), GUCs and system functions 
(gp_role, gp_session_id…), SQL keywords, error messages, docs. We should move 
on this soon.

Benefits are well-rehearsed: ASF trademark hygiene, Cloudberry's own identity, 
clean namespace for cb_ai_* and friends. Two things have stalled it before, 
both addressable now.

First, bulk renames are exactly where AI does better than humans. Missed 
instances in comments and strings, inconsistencies across files, fat-fingered 
names in one file out of fifty — the typical human failure modes are precisely 
what coding agents don't do. They don't get bored, don't skim, don't miss dusty 
corners. With our test suite and PR review backstopping, what used to be a 
quarter of error-prone toil is a few weeks of agent-driven work.

Second, past rename passes polluted history — `git blame` archaeology often 
lands on "rename gpdb → cbdb" with the real change 2–3 hops back. This time we 
should land it as a single atomic commit and add the SHA to 
`.git-blame-ignore-revs`. Git skips listed revs in blame and the rename becomes 
invisible to archaeology. PostgreSQL, Rust, and others use this pattern.

Let's move on this. Anyone willing to drive it — or own a slice (CLI, catalog, 
docs) — would be doing the project a real favor.

Best,
Zhang Mingli

Reply via email to