On Thu Aug 29 19:45:37 2024 +0000, Tim Clem wrote:
Strange, me too. I only tested the final result on macOS, assuming (apparently incorrectly) that MR !6021 had fixed such issues there. It sure seems that proc_pidinfo isn't enumerating those sockets for some reason. Worth digging into since this series replaces that anyway? I could just move the tests later in the series, or mark them flaky until the end.
Moving the tests later in the series is fine.