Introversion

Ahh, becoming a Multwinia beta tester seems unreachable for me, giving I'm so new.

So I will just wait patiently and return any bugs I might come across.

sfericz wrote:Ahh, becoming a Multwinia beta tester seems unreachable for me, giving I'm so new.

So I will just wait patiently and return any bugs I might come across.

I'm waiting patiently for the beta of either Subversion or Chronometer myself.

New build (svn838) is up for Mac, Windows and Linux. Fixes this bug and allows you to IP-range-ban DEMO users while keeping authenticated players unaffected (the system is a bit bugged and doesn't recognize beta versions as Windows versions currently and you can't IP-range-ban authenticated users like the documentation claims.)

I got a sync error.
I know I was playing with Tricky, and a demo player who never set his name. It was a rocket riot game on the Second Beginning. The sync error occurred at 16:27 (4:27, EDT)

Its right here

EDIT: I forgot the server name, sorry. The game resumed after tricky left, but I left anyhow.

What's your KeyID (to be found in every debug.txt) so we can search for the game?

May I ask what purpose the sync info does?

Or maybe I should ask for an explaination of what desyncing is in Multiwinia and it's causes.......

I was under the impression that it was caused by packet jumble aka out of order packets and such. So kind of ignored the whole issue.

Out of order packets are reordered by the clients, they're handled just fine, and for any given time, all clients are guaranteed to have received the correct game history. No, sync errors happen when clients disagree on the game state at a given time. To conserve bandwidth, this check is done via a one byte checksum over the gamestate the clients send to the server in every ack packet. The server then compares these checksums, and on disagreement, triggers the sync error message. The sync logs that are produced then contain the full game state, and if all are collected, the reason for the sync error may be found.

New build is up for Non-Macs, svn858. It adds a LogPing command logging connection stats to the event log and fixes the event log CLIENT_QUIT message reasons. The included python scripts got some love, too, they now use a general purpose event log parser.

did you take off the link or what here because I cant find it.

Read back a bit. There never was a link, it's only for the few testers. I'm logging the changes for everyone because a) I'm too lazy to mail them or start a topic on the private forums and b) the players on the servers may be interested in the changes, too.

Selected testers: svn863 is up for all platforms. It fixes Blitzkrieg score logging, doesn't give CPUs colour names that are just wrong, and adds the WaitGameSecondsAbsolute command.

What would you say a good Dedwinian server should be spec-wise?

cheesemoo0 wrote:What would you say a good Dedwinian server should be spec-wise?

Anything from 150MHz with about 40MB RAM onwards. What will matter the most is bandwidth, especially upload. If you have enough and you use QoS, you can get a good service.

NeoThermic

svn869 is up. Attempts to fix the odd team colour bug. The bug mostly happens when CoopMode is activated when players are already online, which is precisely what Dedwinia Test is doing. So do give note if you still get odd team assignments (meaning anything but 2 vs 2).

About how much upload bandwith does each player take up?

Worst case (4 players, each controlling a turret or squad or something the server can't tell apart from that): 10 kbyte/s total.