On Sat Jan 21 00:43:00 2023 +0000, Jinoh Kang wrote:
No, the actual culprit turned out to be binding to `INADDR_ANY` after more testing. It is not clear that `SO_REUSEADDR` is triggering the firewall. This MR does not prevent the firewall alert dialog. I'm submitting a separate PR for that. (I actually realized this much earlier, but I have only so much time...)
My original theory was that the firewall dialog was caused by _both_ `SO_REUSEADDR` _and_ binding to a fixed port (perhaps only in well-known or registered range?) I thought not hard-coding ports in tests was a good idea in general, but now that the actual cause of the alert turned out to be different, it's true that the extra complexity introduced by dynamic ports is much harder to justify.