From ba08827ff4078448bffed4125f436e0bfdf56c54 Mon Sep 17 00:00:00 2001 From: "Edward A. Lee" Date: Mon, 25 Sep 2023 07:20:10 +0200 Subject: [PATCH] Elaborated explanation --- examples/C/src/distributed/CALDecentralized.lf | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) diff --git a/examples/C/src/distributed/CALDecentralized.lf b/examples/C/src/distributed/CALDecentralized.lf index e967452b..3b82103e 100644 --- a/examples/C/src/distributed/CALDecentralized.lf +++ b/examples/C/src/distributed/CALDecentralized.lf @@ -22,10 +22,13 @@ * * This program uses decentralized coordination, so if the processing latency plus communication * latency exceeds the 200ms tolerance for inconsistency plus the 30ms tolerance for unavailability - * (the deadline), then the coordinator will preserve availability at the expense of consistency. If - * you remove the 200ms `after` delay, however, then you will need to increase the STP offset in the - * `Actuate` reactor to get proper decentralized control. Any number significantly larger than the - * 35ms processing latency should be sufficient to prevent STP violations. + * (the deadline), then the coordinator can preserve availability at the expense of consistency. + * However, in order to do that, you need to adjust the STP offset, which corresponds to the + * processing offset in the CAL theorem. If you remove the 200ms `after` delay, you will need to + * increase the STP offset in the `Actuate` reactor to get proper decentralized control. Any number + * significantly larger than the 35ms processing latency should be sufficient to prevent STP + * violations, but such a number will result in a deadline violation. Setting a lower number + * prevents the deadline violation, but at the cost of getting consistency violations. * * @author Edward A. Lee */