Well, I had a quick talk with KRABBELBABY. Of course, he vowed never to tk again and that he wasn't KRABBELBABY at all. Yeah, right. Half an hour later KRABBELBABY with coincidentally the same IP joins, leaves, and rejoins as [fe]kinked, volunteering to test the "stickiness" of kick votes
And my little ban-by-regex script.
Code: Select all
[2] [fe]meriton: BUG killing was experiencing strange network problems
From the sound of it ("Netobject not spawned"), that's the same bug you and me had again. Network objects that get deleted accidentally. Since it was recurring, it may have been a player or team. Now if I had killer's clientside recording or at least the number of the object from the message...
Code: Select all
[10] Poll "Kick sean" submitted by sean.
[10] New poll: "Kick sean". Enter the main menu vote on it.
[15] Lucifer: BUG
[15] Lucifer: sean started a poll to kick himself
[15] Lucifer: ok, it may not be a bug, but it used to be that way
[15] Lucifer: that he couldn't
Just accept it, or he'll get kick vote protection for some minutes and can do what he wants. Oh, that's what you did
6 to 1.
And there was a genuine crash! We have code that kills a cycle once it gets too many reference counts (from walls, mostly) to protect against performance DOS attacks Jonathan discovered during the 0.2.7.1 release process. Well, I added a reference to the cycle itself in gCycle::Kill, causing an endless recursion. Fixed for 0.2.8.0.