On Tue, Mar 14, 2023 at 10:35:43AM +0530, Bharath Rupireddy wrote:
> My thoughts are simple here - how would one (an end user, not me and
> not you) figure out how to get info/stats till the end of WAL? I'm
> sure it would be difficult to find that out without looking at the
> code or commit history. Search for till end of WAL behaviour with new
> version will be more given the 1.0 version has explicit functions to
> do that. IMO, there's no harm in being explicit in how to achieve till
> end of WAL functionality around in the docs.
Okay. I have kept these notes, but tweaked the wording to be a bit
cleaner, replacing the term "till" by "until". To my surprise while
studying this point, "till" is a term older than "until" in English
literacy, but it is rarely used.
> I get it. I divided the patches to 0001 and 0002 with 0001 focussing
> on the change of behaviour around future end LSNs, dropping till end
> of WAL functions and tests tweakings related to it. 0002 has all other
> tests tidy up things.
>
> Please find the attached v8 patch set for further review.
The tests of 0001 were still too complex IMO. The changes can be much
simpler as it requires only to move the till_end_of_wal() calls from
pg_walinspect.sql to oldextversions.sql. Nothing more.
--
Michael