Bother you, but this picture from other articles (" PBFT practical Byzantine fault-tolerant algorithm in-depth explanation "https://blog.csdn.net/TurkeyCock/article/details/81672759? Utm_source=copied in the app), I study for a long time, because ability is limited, really don't understand, I contact the author of the text, the author should not live for a long time, so can't contact),

I have three main problem
One is in the picture: (green) why is continuous, and the interval of CHECKPOINT are hundreds of (for example, the 100 message interval),
The second is the picture inside why still have to be PRE - PREPARE COMMITED news? I remember COMMITED news has been confirmed and no need to send the PRE - PREPARE, collect in P collection is not COMMITTED, is between min_s and max_s has COMMITTED message still have to restart the PRE - PREPARE?
3 it is up to VIEW the VIEW to five, at least two, wonder why VIEW will be out of sync? NEW - the VIEW should be from the same VIEW of the master node to each send a copy, so I think the VIEW should be synchronized, so I don't understand the picture of the situation,

Thank you,