From 4bdc4983b830fcb7e98470a3b786bfe31a80e991 Mon Sep 17 00:00:00 2001 From: jochen Date: Fri, 20 Nov 2026 22:00:48 +0100 Subject: [PATCH] Design 47 H: H-A-196, a process pulls a durable once --- 03-DESIGN/01-to-be/47-milestones/H-lead-answers.md | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/03-DESIGN/01-to-be/47-milestones/H-lead-answers.md b/03-DESIGN/01-to-be/47-milestones/H-lead-answers.md index e101bdb2..8f8e2ee9 100644 --- a/03-DESIGN/01-to-be/47-milestones/H-lead-answers.md +++ b/03-DESIGN/01-to-be/47-milestones/H-lead-answers.md @@ -1044,3 +1044,9 @@ Where two preparers disagreed, the chosen one is named. of them per branch (a setting of the module, default 4, oldest dropped; a run's candidate counts within its branch's bound), and only for the branches it follows (the watched branches; a branch dropped from them frees its revisions). A kept revision is a cache of the catalog seat's own truth, never a state of the mesh. P1, P4, P11. +- **H-A-196 (a process pulls a durable once, 22:00)** Met twice today: a tool runner built before H-A-132/133 ran two + pull loops on the tools bundle's one durable consumer, so each event reached one loop only and the catalog seat's + revision verb answered a stale revision while the same process had heard the newer one; nothing said so. The SDK + refuses, loud at bind time, a second pull loop on a durable the same process already pulls (the error names the + durable and both binders), so the kind cannot run in silence again; the cause itself is delivered by the tool + runner's own release (H-A-132/133). P1 (failure must be loud), P7.