DELTA after SECOND reviews
Follow-ups
What happens after a SECOND review is followed-up with new commits addressing the concerns?
AI disclaimer: OK, basically everything below other than my pick of DELTA initialisms.
| Letter | Check | Why a delta needs it | Prompt it like this |
|---|---|---|---|
| Diff | Old head against new with range-diff, not the whole MR diff again. |
A rebase moves the base under unchanged lines; only the range-diff separates our delta from it. | "Range-diff the old head against the new and list only what changed." |
| Echoes | The ripple into the files and consumers the MR touches, including what a rebase pulled in. | A small hunk can break a caller the first pass had cleared. | "For each changed symbol, name what reads it and whether that still holds." |
| Loops | Each open thread of ours against the delta: answered, or still open. | The threads are the debts the delta may have paid or left open. | "Per thread: answered by this delta, or still open?" |
| Truth | Docs and tests still match the code; Summation and Neglect re-run on the new head. | A fix can leave the description or a test describing the old behaviour. | "Verify the description and tests against the new head." |
| Additions | Anything new becomes a draft note; nothing new means zero drafts. | Re-raising the first pass buries the one real new finding. | "List only findings absent from the first pass; none is a complete answer." |
Re-read the issue and MR text before drafting from memory — a DELTA once drafted a note from a recollection of the issue that a re-read refuted. FORCED is not a step after a DELTA; it lists how any pass goes wrong.