On Tue, 29 Nov 2022 at 20:46, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> I wrote:
> > That seems like a fairly bad idea: it will add extra contention
> > on ProcArrayLock, and I see no real strong argument that the path
> > can't get traversed often enough for that to matter. It would
> > likely be better for KnownAssignedXidsCompress to obtain the lock
> > for itself, only after it knows there is something worth doing.
>
> Since we're running out of time in the current commitfest,
> I went ahead and changed that, and made the cosmetic fixes
> I wanted, and pushed.
That is a complete patch from multiple angles; very happy here.
Thanks for a great job.
--
Simon Riggs http://www.EnterpriseDB.com/