RE: Failed transaction statistics to measure the logical replication progress

On Thursday, December 16, 2021 8:37 PM I wrote:
> Hi, created a new version v17 according to the recent discussion with changes
> to address other review comments.
FYI, in v17 I've removed one part of commit message
about spool file statistics on the subscriber.
My intention is just to make the patches more committable shape.

Although I deleted it, I'd say still there would be some room
for the discussion of the necessity. It's because to begin with,
we are interested in the disk writes (for the logical replication,
pg_stat_replication_slots is an example), and secondly there can
be a scenario that if the user of logical replication dislikes and wants
to suppress unnecessary writes of file on the subscriber
(STREAM ABORT causes truncate of file with changes, IIUC)
they can increase the logical_decoding_work_mem on the publisher.
I'll postpone this discussion, till it becomes necessary
or will abandon this idea, if it's rejected. Anyway,
I detached the discussion by removing it from the commit message.

Best Regards,
Takamichi Osumi