don't compare peer IDs when hole punching

[marten-seemann]

@mxinden pointed out that comparing peer IDs when hole punching is not necessary. We'll have logic in our DCUtR protocol anyway to assign the role of client and server (we'll need the same logic to coordinate the TCP sim open).

[vyzo]

Sounds like a good idea, but we'll have to change the code in the hole punching pr in go-libp2p so that only one of the peers uses the option, predictably.

[marten-seemann]

I added a comment to that PR, so we don't forget. I suggest moving forward with this PR, and not blocking on that change.

[vyzo]

SGTM

…

On Tue, Jul 6, 2021, 02:07 Marten Seemann ***@***.***> wrote: I added a comment to that PR, so we don't forget. I suggest moving forward with this PR, and not blocking on that change. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#210 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAI4SXTAVEBE7AXMNQ453TTWI3JJANCNFSM473MIRLA> .

[mxinden]

@mxinden pointed out that comparing peer IDs when hole punching is not necessary. We'll have logic in our DCUtR protocol anyway to assign the role of client and server (we'll need the same logic to coordinate the TCP sim open).

Approving the idea of assigning roles through the DCUtR protocol instead of by comparing peer IDs.

[Stebalien]

Someday, maybe, we can expose a separate HolePunch (or ActiveListen) function instead....

[marten-seemann]

Someday, maybe, we can expose a separate HolePunch (or ActiveListen) function instead....

I wouldn't be opposed to that.