Hi Wang-san,
Here is some more information about my v21-0001 review [2] posted yesterday.
~~
If the streaming=parallel will be disallowed for publishers not using
protocol 4 (see Amit's post [1]), then please ignore all my previous
review comments about the protocol descriptions (see [2] comments #4b,
#7b, #47a, #47b).
~~
Also, I was having second thoughts about the name replacement for the
'main_worker_pid' member (see [2] comments #1b, #49). Previously I
suggested 'apply_leader_pid', but now I think something like
'apply_bgworker_leader_pid' would be better. (It's a bit verbose, but
now it gives the proper understanding that only an apply bgworker can
have a valid value for this member).
------
[1] https://www.postgresql.org/message-id/CAA4eK1JR2GR9jjaz9T1ZxzgLVS0h089EE8ZB%3DF2EsVHbM_5sfA%40mail.gmail.com
[2] https://www.postgresql.org/message-id/CAHut%2BPuxEQ88PDhFcBftnNY1BAjdj_9G8FYhTvPHKjP8yfacaQ%40mail.gmail.com
Kind Regards,
Peter Smith.
Fujitsu Australia