Marten Seemann

A similar argument applies to architecture information. It would be useful to have a field to log OS / architecture / kernel information / compiler version.

> Note that it's in my opinion perfectly acceptable to have qvis display a field that isn't defined in the qlog docs as such Are you suggesting that qvis defines...

This only captures a subset of the possible changes. For example, you wouldn't be able to tell if a qlog implementation already supported the recently added `stateless_reset` packet type or...

Thinking about it, I'm not so sure about it any more. For stateless reset, we already use the `trigger` field. Maybe we should do the same for VN? Then we...

Do you have any idea where the `Public` measurement is coming from? Is there any address that the node is dialable at?

Just pointing out that our WebTransport will be marked as "experimental" at first. This means that we won't make any guarantees regarding backwards-compatibility as the IETF draft evolves towards RFC...

How would we set the congestion window then? Like this? ```go type Stat struct { CongestionWindow *uint64 } ``` Alternatively, we could say that for values where the zero value...

Just noticed, of course this doesn't live at the transport level, but at the connection level. This means we can either extend the `ConnStats` struct: https://github.com/libp2p/go-libp2p/blob/fcf408c65d99f9a2f45c761a1003d66d3a664f30/core/network/network.go#L100-L104 or introduce a new...

Agreed. From the user's perspective it's better to not define that constant, than to define it and then not to emit it. Let's make it a TODO.

Going one step further, one could argue that this is purely a yamux (and mplex?) problem, and that reasonable stream multiplexers will have a connection-level flow control window. For QUIC,...