From c4cfceefa91efb9dd3279794d965c41f389fb83c Mon Sep 17 00:00:00 2001 From: jochen Date: Sun, 22 Nov 2026 02:17:20 +0100 Subject: [PATCH] Design 47 H2: H-A-451, the tool runner's review round 1 answers --- 03-DESIGN/01-to-be/47-milestones/H2-lead-answers.md | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/03-DESIGN/01-to-be/47-milestones/H2-lead-answers.md b/03-DESIGN/01-to-be/47-milestones/H2-lead-answers.md index 13844e2b..44606e77 100644 --- a/03-DESIGN/01-to-be/47-milestones/H2-lead-answers.md +++ b/03-DESIGN/01-to-be/47-milestones/H2-lead-answers.md @@ -660,3 +660,10 @@ The lead's one-line answers to the questions H2's develop and review raise (impl registers the module instead). (2) The judgement reads the repository's tree at the registered branch's head through the seat git (the trees read wholeAt uses): a module.json at the tree's root or under modules/ holds a manifest; the indexer's own trunk entries are not read for it (a repository not yet taken in has none). P3, P-failure-loud. +- **H-A-451 (the tool runner's review round 1: a kept delete in a page; T3; the wait's seat list, 02:36)** (1) §779: a + page hands a kept delete as an entry with `deleted: true` and no value (the key's delete marker, at its sequence), so a + reader following pages removes the key and a deleted key never comes back alive from a bucket with history; the + seen-again rule applies to it like any message. (2) §799's again-delays (T3): the tool runner builds nothing; the delay + is the bundle's hand-back (nak) delay, which the runtime already applies as given; the row is closed so. (3) §821: the + tool runner's wait (its `console` package) knows every seat of the seat list by the events it declares (07ebeae7's change to wait.go), the row's + "nothing built" was the map's error. P3, P-failure-loud.