Bruce Momjian wrote:
>
> This seems quite involved. Can we get the basic functionality I
> described first?
On the way.
> Also I am not sure how all this information is going
> to be passed from the logging process to the backend requesting the
> information, and it seems overly complicated.
There's *no* information passing from the logging process, with the
single exception of the latest logfile timestamp (if allowed). I'd
rather like to have that information from the logger, to be safe in case
the system time was manipulated and the last logfile is not the current one.
The rest is just a reworked version of pg_dir_ls, with internal
knowledge of how the timestamp is formatted.
Regards,
Andreas