sorokin.oleg.edu/learnharness
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 6 Wiki Activity

the last resolution decides the wake (owner ruling 2026-10-03, round-1 incident)

Browse Source 
Live test finding: answer A then dismiss B woke the agent — the continuation
fold treated a mixed [answered, dismissed] trailing batch as 'reaction owed'
(all-dismissed rule only). Dismiss A then answer B and dismiss-both were
already correct.

The law now: the trailing batch's LAST resolution decides. Ending on a
dismissal = the user signaled stop — no wake, the answers' reactions ride the
next plain message. Ending on a real answer = the reaction is owed. One law
with the pump's pause (endsInDismissal reads the same last result), so
continuation, delivery-holding, and the pause can never disagree. The wrong
pin ('a real answer in the batch still demands the reaction') flipped.
This commit is contained in:
amosblomqvist
2026-10-03 02:27:44 +03:00
parent 7a5a55e94a
commit a3a048f572
3 changed files with 103 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/session-core.test.ts
+17 -4
View File
@@ -854,7 +854,20 @@ describe("the dismissal pause (needsContinuation + endsInDismissal)", () => {
assert.equal(needsContinuation(messages), false, "dismissed by id, not by position — the batch rule holds");
});
it("mixed tail (answered + dismissed) keeps the normal rule — the answer is owed its reaction", () => {
it("mixed tail answered-then-dismissed: the LAST resolution decides — no wake (round-1 incident)", () => {
const assistant = {
...assistantWithCall("a1", "call-a"),
content: [
{ type: "toolCall", id: "call-a", name: "quiz", arguments: {} },
{ type: "toolCall", id: "call-b", name: "quiz", arguments: {} },
],
} as unknown as LhMessage;
const messages = [userText("u1", "topic"), assistant, resultMsg("call-a", "quiz"), dismissedMsg("call-b")];
assert.equal(endsInDismissal(messages), true, "tail ends in the dismissal");
assert.equal(needsContinuation(messages), false, "ending on a dismissal = the user signaled stop — the reaction rides the next message");
});
it("mixed tail dismissed-then-answered: the answer IS owed its reaction (round-2 wake)", () => {
const assistant = {
...assistantWithCall("a1", "call-a"),
content: [
@@ -862,9 +875,9 @@ describe("the dismissal pause (needsContinuation + endsInDismissal)", () => {
{ type: "toolCall", id: "call-b", name: "quiz", arguments: {} },
],
} as unknown as LhMessage;
const messages = [userText("u1", "topic"), assistant, resultMsg("call-a", "quiz"), dismissedMsg("call-b")];
assert.equal(endsInDismissal(messages), true, "tail ends in the dismissal");
assert.equal(needsContinuation(messages), true, "a real answer in the batch still demands the reaction");
const messages = [userText("u1", "topic"), assistant, dismissedMsg("call-a"), resultMsg("call-b", "quiz")];
assert.equal(endsInDismissal(messages), false);
assert.equal(needsContinuation(messages), true, "ending on a real answer = the reaction is owed");
});
it("any user message un-pauses (the derived tail flips)", () => {