Re: Add pg_accept_connections_start_time() for better uptime calculation

On Sun, Feb 16, 2025 at 07:53:06PM -0500, Tom Lane wrote:
> Yeah.  Making that happen would require extending the contents of
> postmaster.pid, which is likely to break assorted peoples' tooling.
> I doubt that this feature clears the bar for justifying that.

Sure, agreed to not touch postmaster.pid.

Now my point is also that I would not object to a patch that wants to
show the information of postmaster.pid in a nicer way than it is now
through SQL, as one tuple with one attribute per field written, or
something like a JSON object.  With the format of postmaster.pid being
very stable across releases, perhaps one attribute per field is
better.
--
Michael