From ec27abd1da51c0681b172009fdf9cd4d47f784dd Mon Sep 17 00:00:00 2001 From: sms <114885497+seanmartinsmith@users.noreply.github.com> Date: Tue, 6 Oct 2026 11:43:29 -0400 Subject: [PATCH] docs(cli): say why config writes skip the kv.* screen (GH#4078) commitConfigWrite stages every dirty config row, while commitMemoryWrite refuses internal keys. Record the reason in the code, not only in the commit history: bd config set/unset/set-many are deliberate config writes, the same trust level as bd dolt commit. --- cmd/bd/dolt_autocommit.go | 2 ++ 1 file changed, 2 insertions(+) diff --git a/cmd/bd/dolt_autocommit.go b/cmd/bd/dolt_autocommit.go index 638b372aa..35430188e 100644 --- a/cmd/bd/dolt_autocommit.go +++ b/cmd/bd/dolt_autocommit.go @@ -170,6 +170,8 @@ func commitPendingIfEmbedded(ctx context.Context, st storage.DoltStorage, actor // dolt.auto-commit batch/off, which defer it to `bd dolt commit` (CommitAll, // config included) exactly as they defer every other server-mode write. // +// Unlike commitMemoryWrite it does not screen other dirty config rows: the +// operator is writing config on purpose, the trust level of `bd dolt commit`. // rename-prefix and migrate also write config (issue_prefix, sync.branch) and // are deliberately not wired here: they are multi-step operations, and when // their config row should become a commit is the GH#2455 question itself.