简体中文界面。帖子、指南及政策正文保留原文；未对原文作自动翻译。
ClaraBennett0698 · 2026年6月25日 15:47 UTC
Ask what stays false after the request arrives. Make them answer that separately.
16分We had a review where everyone agreed with the same word because they were imagining different moments. Drawing the normal sequence didn't expose it. Asking the machine person to leave the machine busy after receiving the request did. Then the robot person had to explain what they would wait for, instead of nodding at the word ready.
12分Priya, that is a better meeting question than my usual plea for definitions. People can agree on a noun for hours.
15分I've put the busy-machine case first. Receipt can occur in that case; availability cannot. We need names that preserve the distinction in the interface table.
9分Who owns withdrawing availability when the machine state changes before the transfer?
14分And does the integrator actually receive that withdrawal, or just remember the earlier ready?
4分Louis, yes, the receiver's interpretation belongs beside the sender's definition; otherwise Clara gets two improved documents that still disagree with each other.
18分Naomi and Louis, both teams now put availability under the machine side and require the receiver to respond to its loss. They're still arguing about the interrupted-transfer state.
-1分I wouldn't rush that last argument. Before transfer and during transfer may need different responses, and neither should be improvised from a signal name. Have them identify the physical state they can establish in each case and the authorised recovery owner; that is where our tidy table grew useful.
20分We have separate receipt and availability entries signed by both teams. Interrupted transfer stays an explicit design issue, with the machine and integration leads jointly responsible for resolving it.
15分