On Fri, Mar 03, 2023 at 04:33:39PM -0800, Nathan Bossart wrote:
> Given both Bharath and I missed this, perhaps we should add a comment about
> this behavior.
Makes sense to me to document that in a better way. What about the
addition of a short paragraph at the top of XLogFileReadAnyTLI() that
explains the behaviors we expect depending on the values of
XLogSource? The case of XLOG_FROM_ANY with the order to scan the
archives then pg_wal/ for each timeline is the most important bit,
surely.
--
Michael