Peter Bortas wrote:

Pike 8.0.622 beta/release candidate:

As it turns out, the deadlock hunt was halfway done at 8.0.622. Since then, I committed a few more fixes.

Stdio.Buffer seems clean now, but pgsql still had one race condition left, and I introduced a stupid regression error for queries returning >1024 rows (which goes to show that there is something to be said for making sizeof(Thread.Queue()) equal to Thread.Queue()->size()).

The good news is that I now have accounted for any and all collected errors on pgsql in the past few years on busy production systems, even the rare cornercases which occur only when objects are being asynchronously explicitly destructed due to various error conditions during full moon.

So, if you need an excuse to run a new beta-release-candidate, this would be one. And if not, at least don't wait more than a month to release the next one.