Eugene

I'm reworked retransmission and bring back serverId update, please check https://github.com/ec-/Quake3e/commit/f8e7b429b9098e3fc009967cb5f6e5dc4dec9260

> May I ask what is the compatibility reason for bringing back the serverid change during map restart? It is needed to discard outdated commands for CS_ACTIVE clients > One...

> I thought we had already addressed this with client->lastUsercmd.serverTime = sv.time - 1 to prevent pre-restart **move commands** being executed. Can you be more specific about what was wrong...

* At `User_Move()` you might have `serverId = sv.serverId` but incorrect `messageAcknowledge` which will place client into game so exact gamestate acknowledge is important * Issue with multiple `CL_WritePacket()` has...

It is not correct to allow client to join by simple `serverId == sv.serverId` check as it may be a reply to _any_ previous gamestate client received - so exact...

- client could receive incorrect configstrings because `csUpdated[]` now is reset after sending (to avoid potential overflows) - dedicated gamestateAcked variable is needed to skip sending snapshots to clients

> Seems questionable this is very useful. For an overflow to happen because of extra csUpdated values accumulated prior to a gamestate resend, but not due to the (likely much...

> Could you explain what you mean by the potential overflow? every `csUpdated` is +1 to `cl->reliableSequence` -> more chance to overflow > With the original gamestate check it shouldn't...

* I see no real improvements over existing code, moreover this https://github.com/ec-/Quake3e/pull/286/commits/8135f46b0861ad67f63f89afc8a9231eeb4b6308#diff-6424df15aa5cce99d5418d3af2588838eec3027f261fae9ff9d648154a8deac6R2271-R2275 is not correct * I like more universal apparoach with `gamestateAck` state tracking instead of tiding to post-download...

it says ```c // this client has acknowledged the new gamestate ``` which is NOT correct