On Mon, Apr 11, 2022 at 04:39:30PM +1200, David Rowley wrote:
> I'm not entirely certain this is an improvement. Your commit message
> I'd say is not true going by git grep "compression algorithm". There
> are 3 matches in the docs and take [1], for example. I'd say in that
> one it's better to use "algorithm". In that case, "method" could be
> talking about client or server.
I am not wedded to this change; but, for context, I wrote this patch before
basebackup supported multiple compression ... "things". I didn't touch
basebackup here, since Robert defended that choice of words in another thread
(starting at 20220320194050.GX28503@telsasoft.com).
This change is for pg_column_compression(), and the only other use of
"compression algorithm" in the docs is in pgcrypto (which is in contrib). That
the docs consistently use "method" suggests continuing to use that rather than
something else. It could be described in some central place (like if we
support common syntax between interfaces which expose compression).
> 0010:
> I don't understand this change.
The commit message mentions 959f6d6a1, which makes newbindir optional. But the
documentation wasn't updated, and seems to indicate that it's still required.
https://www.postgresql.org/docs/devel/pgupgrade.html
> 0011:
> I can't quite parse the original. I might not have enough context
> here. Robert, Joe? (new to master)
See the link in the commit message where someone else reported the same
problem.
> 0019:
> -1. pgindent will fix these.
But two of those are from 2016.
Thanks for amending and pushing those. There's some more less obvious ones
attached.
Amit or Masahiko may want to comment on 0012 (doc review: Add ALTER
SUBSCRIPTION ... SKIP).