Thomas Eizinger
Thomas Eizinger
@conectado This is ready for review. I tried to make meaningful commits.
> I don't understand how this fixes #4290, I remember that we have priorities set where we always pick a direct connection over a relayed one, and str0m should test...
> This PR may be fixing an issue that no longer exists. I've removed the 2nd "resolves". With that taken into account, this PR is primarily an optimisation to avoid...
> > I can remove the "resolves" again and close the other one as not reproducible if you think that is more correct? > > yeah :) Done!
> could 2 seconds be too high of a timeout? Yeah, doing this time-based isn't ideal. I'd like to work on https://github.com/firezone/firezone/issues/4060 which allows us to not pay this time-penality...
I think the tests are currently failing for the same reason as https://github.com/algesten/str0m/pull/496. The non-roaming party doesn't know that the candidate was invalidated and thus keeps sending messages to it.
> I obviously don't know the full story here, but from a birds eye perspective it appears like working against the spirit of the ICE agent. > > The whole...
That seems to have changed. The latest docs say: > This trait only accepts UTF-8–encoded data and is not [flushable](https://doc.rust-lang.org/std/io/trait.Write.html#tymethod.flush). If you only want to accept Unicode and you don’t...
> Would this use the `releases/generate-notes` [endpoint](https://docs.github.com/en/rest/releases/releases?apiVersion=2022-11-28#generate-release-notes-content-for-a-release) from the github api? Unlikely because that doesn't follow the "keep-a-changelog" format as far as I am aware.