Suggestion for Ladle
Moderator: Light
Suggestion for Ladle
I don't know who the leader/creator of the ladle is, so I posted here.
I was looking through the guidelines the other day, as Im playing on Sunday, and there was one thing that I thought was surprising. If I read correctly, it is 8v8 with best of 3. The problem I see with that, is with everyone (except subzero and his amazingly low ping) will probably lag a lot more than normal with 17 people in a server (one admin). I asked around and a good amount of people seem to agree with me. I was wondering if it could be changed to 6v6, or 7v7.
Thanks.
I was looking through the guidelines the other day, as Im playing on Sunday, and there was one thing that I thought was surprising. If I read correctly, it is 8v8 with best of 3. The problem I see with that, is with everyone (except subzero and his amazingly low ping) will probably lag a lot more than normal with 17 people in a server (one admin). I asked around and a good amount of people seem to agree with me. I was wondering if it could be changed to 6v6, or 7v7.
Thanks.
- 2020
- Outside Corner Grinder
- Posts: 1322
- Joined: Thu Dec 29, 2005 9:21 pm
- Location: the present, finally
there should be enough flexibility in the system to deal with this...
captains could arrange beforehand in irc
to play 7v7 or less
ultimately
however
a team can field 8 whatever the circumstances
however many opponents there or ping...
if we empower captains to make these kinds of decisions
to actually agree on the circumstances for a game
within a short period of time
(ie changeover between rounds)
the system becomes scalable...
captains could arrange beforehand in irc
to play 7v7 or less
ultimately
however
a team can field 8 whatever the circumstances
however many opponents there or ping...
if we empower captains to make these kinds of decisions
to actually agree on the circumstances for a game
within a short period of time
(ie changeover between rounds)
the system becomes scalable...
hold the line
It's not my decision, I'm just one of several server admins. If both teams rather want to play with a lower (or higher) number of players, I'd suggest the server admin follows that wish. 8 should be the default for this Ladle, simply because it has always been that way and teams have prepared for that number.
Experiences with smaller teams on the AFL were good, IMHO; not every team can muster full eight players. The Ladle is player organized; so I would suggest you simply proclaim to be the initiator of the next ladle and declare it to be one for teams of five players, people would be too lazy to object
Experiences with smaller teams on the AFL were good, IMHO; not every team can muster full eight players. The Ladle is player organized; so I would suggest you simply proclaim to be the initiator of the next ladle and declare it to be one for teams of five players, people would be too lazy to object

just another suggestion, we could give the top 4 teams a rank (the teams that make it to semis) and make it so that each team plays one of the bottom 4 teams. For example, if the top 4 teams in the last ladle were |x|,Ct,|DS|,TKR, then none of those teams would play each other, each would play one of the other 4 teams in the next ladle.
-
- Round Winner
- Posts: 246
- Joined: Thu Jun 08, 2006 2:39 am
I've happily lived my life up till now without knowing that existed. Now I can't get those images out of my headLegit wrote:2girls1cup

Thanks Legit.
(anyone who doesn't know what it is, you're better off that way. don't, like me, let curiousity get the better of you)
Oh, and I think 6 per team make for better matches also. Where skill plays a greater part than luck/lag.
I'm still busy not understanding what was so funny about Star Wars Kid and Numa Numa Guy. Besides, the server runs on a different connection than my home network, so I could torrent all Linux distributions in existence (and, piggybacking on it, some pr0n, of course) simultaneously without harming the server.
About the "lag" problems (really, as far as I observed them, they were performance problems that had nothing to do with lag), they're clientside bugs. Analysis showed that they are triggered by a seemingly innocent server change; something was sending duplicate cycle sync messages after death. They triggered an unfortunate chain of events: when a client receives a cycle sync message, it MOVES the cycle. When any game object is moved, it references a triangle of the grid datastructure, and keeps hold of it. While a game object keeps hold of a triangle, it and all triangles it gets transformed to by grid operations (happening every time a wall gets inserted) will not be simplified. That proliferation of non-simplifyable triangles usually is not a performance problem, because on every timestep, game objects update the referenced current triangle to be one that really is in the grid, and thus clears the non-simplification proliferation list.
Not so when it is dead, however. The end result is this: when the server sends a duplicate death message, the clients' grid datastructures get infested by a non-simplification virus. That virus first only has the triangle the cycle died in infested, but every time someone drives through that triangle, the virus spreads. Really quickly, it can infect the whole grid. Then the grid does not get simplified, and the inefficiency of the 0.2.8.2.1 rendering code hits: it traverses the whole grid datastructure for its work, which of course then takes longer and longer.
Apart from fixing the client errors where I could find them, I added guards to the server code that no duplicate death messages are sent any more. Except when the original packet is lost, of course.
So, sorry about that. But really, it was not the server's fault.
About the "lag" problems (really, as far as I observed them, they were performance problems that had nothing to do with lag), they're clientside bugs. Analysis showed that they are triggered by a seemingly innocent server change; something was sending duplicate cycle sync messages after death. They triggered an unfortunate chain of events: when a client receives a cycle sync message, it MOVES the cycle. When any game object is moved, it references a triangle of the grid datastructure, and keeps hold of it. While a game object keeps hold of a triangle, it and all triangles it gets transformed to by grid operations (happening every time a wall gets inserted) will not be simplified. That proliferation of non-simplifyable triangles usually is not a performance problem, because on every timestep, game objects update the referenced current triangle to be one that really is in the grid, and thus clears the non-simplification proliferation list.
Not so when it is dead, however. The end result is this: when the server sends a duplicate death message, the clients' grid datastructures get infested by a non-simplification virus. That virus first only has the triangle the cycle died in infested, but every time someone drives through that triangle, the virus spreads. Really quickly, it can infect the whole grid. Then the grid does not get simplified, and the inefficiency of the 0.2.8.2.1 rendering code hits: it traverses the whole grid datastructure for its work, which of course then takes longer and longer.
Apart from fixing the client errors where I could find them, I added guards to the server code that no duplicate death messages are sent any more. Except when the original packet is lost, of course.
So, sorry about that. But really, it was not the server's fault.
- wrtlprnft
- Reverse Outside Corner Grinder
- Posts: 1679
- Joined: Wed Jan 04, 2006 4:42 am
- Location: 0x08048000
- Contact:
Is that by any change the same bug as the one shown in these old screenshots with DEBUGLEVEL=3?
There's no place like ::1