Lix Multiplayer Sun, March 27th, 17:00 UTC

Started by Simon, March 24, 2022, 10:19:27 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Simon

Hi,

who's up for Lix multiplayer this Sunday, March 27th, starting 17:00 UTC?

(Mind daylight savings. In Germany, 17:00 UTC will be 19:00 local time.)



I've been a busy mole, I've pushed lots of earth deep inside the server. The goal is to support both 0.9.x and the future 0.10.x on the same server, in different rooms. Let's test if I broke anything. For this Sunday's session, I'll run the reworked server as the central server. Ideally, you won't notice anything at all. :lix-grin:

For you, everything is as usual:
  • No need to update Lix. 0.9.42 is the most recent, but any 0.9.x works.
  • In Lix: "Network game" -> "Play on the central server" -> "Okay".
-- Simon

Dullstar

For North Americans:
at least according to Google, which I'm hoping has appropriately dealt with the conversion:
1 PM Eastern, 12 PM Central, 11 AM Mountain, 10 AM Pacific

I can't promise anything as of right now, but I'll be glad to hop in for a bit if I'm available.

Flopsy

Sorry I can't make that time sadly, I'm working that evening.

I can however play the following Sunday 3rd April.

Simon

Thanks for the feedback! A small, short session on March 27th is perfectly fine for testing.

Sunday, April 3rd: Good idea, I'll keep that evening free for Lix. I'll post a new thread for April 3rd after I've seen how the (small?) session on March 27th went.

-- Simon

Simon

I'm on the server.

Nobody else is here yet, I'll hang around. Come online anytime in Lix or on Mumble.

-- Simon

Simon

#5
Rampoina was new from the LibreGaming community, played for the 2nd day and already won some games in 3-player with me and Dullstar.

Rampoina's suggestions:
  • More time on new map (10+ seconds) before lix spawn,
  • or ability to zoom into the preview.
  • "Why can't I nuke?" -> Review complicated overtime rules: Nuke doesn't trigger overtime unless you've saved at least one lix.
I ran the experimental server. Two hours into the session, all 3 clients segfaulted at the same time: 1x bleeding edge on Linux, 1x 0.9.42 on Linux, 1x 0.9.42 on Windows. Everybody ran a release build, no core dumps exist. I re-joined with a debugging build, and we couldn't reproduce the segfault.

A sporadic crash of the most painful kind. The client-side networking is the obvoius suspect, but the crash would probably have happened sooner and more often then. Why should the experimental server send surprising stuff two hours in, after sending many kosher packets before. It feels like it's 50:50 whether it's in the client-side networking code or somewhere else. E.g., I can't completely rule out that the jumper/tumbler code has a crashing bug that has slept for 7 years...

-- Simon