From f906b7c8b32a09a46e81727b9ed93ee2b1018415 Mon Sep 17 00:00:00 2001 From: jochen Date: Tue, 17 Nov 2026 16:00:12 +0100 Subject: [PATCH] Answer A-475: a validation ask whose request moved is asked again, not left could-not-validate --- 03-DESIGN/01-to-be/47-milestones/G-22-lead-answers-2.md | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/03-DESIGN/01-to-be/47-milestones/G-22-lead-answers-2.md b/03-DESIGN/01-to-be/47-milestones/G-22-lead-answers-2.md index 094109e9..084a4781 100644 --- a/03-DESIGN/01-to-be/47-milestones/G-22-lead-answers-2.md +++ b/03-DESIGN/01-to-be/47-milestones/G-22-lead-answers-2.md @@ -1837,3 +1837,11 @@ the principle that settles it, cited as "design 47 G A-n". pipeline's test step places no other repository beside the one it tests, and a module's tests read nothing outside their module: a test that needs a program of another module builds a stand-in from its own testdata (a fixture composer that answers as the composer's contract says), never that module's source tree. +- **A-475** (§713, met twice: 2026-11-17 14:02 and 15:37; principles 3 and 2) A validation ask whose revision's request + moved while it was out (a take-in of the branch's head, the catch-up's included, wrote a newer revision) is a stale + ask, as §713 drops one, never a could-not-validate: the indexer asks the revision again under its next try at once, + and a pending revision whose merges are all covered by what was taken in ends with them. The catch-up takes in no + repository while a validation ask of it is out; it waits for that ask's end. A stale ask asked again past + `validate-tries` tries in a row ends could-not-validate with its words, so a request that keeps moving is loud. + Could-not-validate keeps §713's other causes (the step could not run, the program exited other than 0, the answer + not both cases), asked again by a later merge or `indexer_validate_again`.