Philip Semanchuk <philip@americanefficient.com> writes:
>> On Feb 27, 2020, at 4:31 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> Looking at the original discussion, it seems clear that the choice of
>> pg_monitor was intentional; see in particular
>> https://www.postgresql.org/message-id/CA%2BOCxowV7eL-DS1Hr-h5N7Tr8Gvn5VGW%2B%2BYJ2yo6wMN9H3n9Gg%40mail.gmail.com
>> So I think the code is correct and the documentation is a typo.
>> That's a much easier answer to back-patch, as well.
> Sounds good to me. Thanks for the context!
Re-reading the pg_buffercache documentation, I was reminded that that
view can have a pretty significant performance hit if you've got lots
of shared buffers. So I think being restrictive about it is good
policy, reinforcing the view that the code made the right choice.
I pushed a patch fixing the docs, in v10 and up.
regards, tom lane
