Re: [webrtc-pc] Should garbage collecting RTCPeerConnection be observable? (#3090)

> Not directly relevant, but a counterpoint to [@docfaraday](https://github.com/docfaraday) : Let me just observe that FIN_WAIT_2 is not a good feature of TCP - having a protocol that is not resistant against the other party ghosting you is an easy opening for a denial-of-service attack.
> 
> The reason why we put keepalives into the protocol is exactly so that if the other party goes away without telling you, it's possible to recover the resources eventually.
> 
> (a quick googling came up with https://medium.com/@WaterBucket/ms13-018-the-tcp-bug-that-made-windows-wait-and-wait-and-wait-0fc98816d12a as one description of the FIN_WAIT_2 issue)

One can certainly engineer transport protocols that are more capable of handling things like this. And you always must leave an escape hatch; sometimes an endpoint loses the ability to reach the other. But that's the rainy-day case. The idea is that an explicit shutdown is performed whenever possible. It's like littering; you have a rule against it, but you also have cleanup crews for accidents and the people who cannot be bothered.

Don't Mess With ~~Texas~~ The Internet

-- 
GitHub Notification of comment by docfaraday
Please view or discuss this issue at https://github.com/w3c/webrtc-pc/issues/3090#issuecomment-3966701970 using your GitHub account


-- 
Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config

Received on Thursday, 26 February 2026 13:39:12 UTC