Floris Bruynooghe
Floris Bruynooghe
Hi, I've been looking around in `poll_receive` recently and started wondering if it could be simplified. This is my first attempt at a step in this direction, it pulls together...
This is a continuation of #2168 and concentrates on further simplifying the logic needed in the poll_transmit function. * The buffer to write into is now owned by the `PacketBuilder`...
When a tail-loss probe needs to be sent the current `queue_probe` algorithm adds a PING frame even if a DATAGRAM frame could also be sent. This wastes a single byte,...
Sometimes the application knows about network path changes that Quinn may not detect immediately to. If the network changed significantly it can often be faster to reach optimal throughput and...
It is scary when folks have to worry about secrets being printed. #84 will solve the key persistence in a much better way anyway.
If I understood this correctly this change is editorial. "The peers that ..." reads rather strange combined with "... from the peer ..." later in that sentence. I think using...
The relay client in the MagicSock sometimes wants to reconnect to be sure the connection is good. When doing so however it does not use the RLS session resumption, here...
We used to have logic that would re-holepunch at regular intervals if the latency was "not good enough". We have some commented out constants related to this still lying around...
It is probably possible for a client to add these metrics: - holepunch_attempt: counter - holepunch_success: counter But it might takes some careful figuring out, and should be done on...
This is unlikely to be needed for 1.0. But there's a TODO in the code so creating this issue already for visibility. And someone might disagree with how important this...